WO2025237147A1 - Ai单元监控方法、装置及设备 - Google Patents

Ai单元监控方法、装置及设备

Info

Publication number
WO2025237147A1
WO2025237147A1 PCT/CN2025/093340 CN2025093340W WO2025237147A1 WO 2025237147 A1 WO2025237147 A1 WO 2025237147A1 CN 2025093340 W CN2025093340 W CN 2025093340W WO 2025237147 A1 WO2025237147 A1 WO 2025237147A1
Authority
WO
WIPO (PCT)
Prior art keywords
information
measurement
measurement data
data
unit
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Pending
Application number
PCT/CN2025/093340
Other languages
English (en)
French (fr)
Inventor
姚健
贾承璐
孙布勒
姜大洁
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Vivo Mobile Communication Co Ltd
Original Assignee
Vivo Mobile Communication Co Ltd
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Vivo Mobile Communication Co Ltd filed Critical Vivo Mobile Communication Co Ltd
Publication of WO2025237147A1 publication Critical patent/WO2025237147A1/zh
Pending legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Classifications

    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/04—Architecture, e.g. interconnection topology
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/08—Learning methods
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04W—WIRELESS COMMUNICATION NETWORKS
    • H04W4/00—Services specially adapted for wireless communication networks; Facilities therefor
    • H04W4/30—Services specially adapted for particular environments, situations or purposes
    • H04W4/38—Services specially adapted for particular environments, situations or purposes for collecting sensor information

Definitions

  • This application belongs to the field of communication technology, specifically relating to a method, apparatus and equipment for monitoring artificial intelligence (AI) units.
  • AI artificial intelligence
  • Some communication systems support perception and allow inference of perceived data using AI units.
  • the AI unit is often used continuously for inference of perceived data after training.
  • AI units generally suffer from generalization problems; for example, in some perception scenarios or over time, their processing power may not meet the perception requirements. Therefore, continuously using the same AI unit for inference of perceived data may easily lead to a decrease in the accuracy of the perceived data, resulting in a deterioration in perception performance.
  • This application provides an AI unit monitoring method, apparatus, and device, which can solve the problem that continuous use of AI unit inference perception data may easily lead to a decrease in the accuracy of perception data, resulting in a deterioration in perception performance.
  • an AI unit monitoring method including:
  • the first device acquires the perception data output by the first AI unit
  • the first device acquires reference data for monitoring the first AI unit
  • the first device determines the monitoring result of the first AI unit based on the reference data and the perception data output by the first AI unit.
  • an AI unit monitoring method including:
  • the second device transmits data, and the transmitted data includes at least one of the following:
  • the first measurement data is used to be input into the first AI unit for reasoning to obtain the perception data output by the first AI unit, and the first measurement data includes the first measurement result; or, the first measurement data includes the perception data output by the first AI unit.
  • the second measurement data is used to determine a reference signal, and the second measurement data includes a second measurement result; or, the second measurement data includes reference data used to monitor the first AI unit.
  • an AI unit monitoring device including:
  • the processing module is used to acquire the perception data output by the first AI unit
  • the processing module is also used to acquire reference data for monitoring the first AI unit
  • the processing module is further configured to determine the monitoring result of the first AI unit based on the reference data and the perception data output by the first AI unit.
  • an AI unit monitoring device comprising:
  • a sending module is configured to send data, wherein the sent data includes at least one of the following:
  • the first measurement data is used to be input into the first AI unit for reasoning to obtain the perception data output by the first AI unit, and the first measurement data includes the first measurement result; or, the first measurement data includes the perception data output by the first AI unit.
  • the second measurement data is used to determine a reference signal, and the second measurement data includes a second measurement result; or, the second measurement data includes reference data used to monitor the first AI unit.
  • an AI unit monitoring device is provided, the device being configured to perform the steps of the first device-side AI unit monitoring method provided in the embodiments of this application, or to implement the steps of the second device-side AI unit monitoring method provided in the embodiments of this application.
  • a device including a processor and a memory, the memory storing a program or instructions executable on the processor, the program or instructions, when executed by the processor, implementing the steps of the first device-side AI unit monitoring method provided in the embodiments of this application.
  • a device including a processor and a communication interface, wherein the processor or the communication interface is used to acquire perception data output by a first AI unit; acquire reference data for monitoring the first AI unit; and the processor is used to determine the monitoring result of the first AI unit based on the reference data and the perception data output by the first AI unit.
  • an apparatus comprising a processor and a memory, the memory storing a program or instructions executable on the processor, the program or instructions, when executed by the processor, implementing the steps of the second device-side AI unit monitoring method as provided in the embodiments of this application.
  • a ninth aspect provides a device including a processor and a communication interface, wherein the communication interface is used to transmit data, the transmitted data including at least one of the following: transmitting first measurement data and transmitting second measurement data; wherein the first measurement data is used to be input to a first AI unit for inference to obtain perception data output by the first AI unit, and the first measurement data includes a first measurement result; or, the first measurement data includes perception data output by the first AI unit; the second measurement data is used to determine a reference signal, and the second measurement data includes a second measurement result; or, the second measurement data includes reference data, the reference data being used to monitor the first AI unit.
  • a readable storage medium on which a program or instructions are stored, which, when executed by a processor, implement the steps of the AI unit monitoring method on the first device side as provided in the embodiments of this application, or implement the steps of the AI unit monitoring method on the second device side as provided in the embodiments of this application.
  • a wireless communication system comprising: a first device and a second device, wherein the first device is configured to perform the steps of the AI unit monitoring method on the first device side as provided in the embodiments of this application, and the second device is configured to perform the steps of the AI unit monitoring method on the second device side as provided in the embodiments of this application.
  • a chip including a processor and a communication interface, the communication interface being coupled to the processor, the processor being used to run programs or instructions to implement the AI unit monitoring method on the first device side as provided in the embodiments of this application, or to implement the AI unit monitoring method on the second device side as provided in the embodiments of this application.
  • the first device acquires the perception data output by the first AI unit; the first device acquires reference data for monitoring the first AI unit; and the first device determines the monitoring result of the first AI unit based on the reference data and the perception data output by the first AI unit.
  • the first device determines the monitoring result of the first AI unit based on the reference data and the perception data output by the first AI unit.
  • Figure 1 is a block diagram of a wireless communication system applicable to an embodiment of this application
  • Figure 2 is a schematic diagram of a measurement scenario provided in an embodiment of this application.
  • Figure 3 is a schematic diagram of another measurement scenario provided by an embodiment of this application.
  • Figure 4 is a schematic diagram of a neural network provided in an embodiment of this application.
  • Figure 5 is a schematic diagram of a neuron provided in an embodiment of this application.
  • FIG. 6 is a flowchart of an AI unit monitoring method provided in an embodiment of this application.
  • Figure 7 is a schematic diagram of a signal resource provided in an embodiment of this application.
  • FIG. 8 is a flowchart of another AI unit monitoring method provided in an embodiment of this application.
  • FIG. 9 is a schematic diagram of an AI unit monitoring method provided in an embodiment of this application.
  • Figure 10 is a schematic diagram of a time delay spectrum information subset extraction provided in an embodiment of this application.
  • FIG 11 is a schematic diagram of another time delay spectrum information subset extraction provided in an embodiment of this application.
  • Figure 12 is a schematic diagram of an AI unit monitoring method provided in an embodiment of this application.
  • Figure 13 is a schematic diagram of a signal path detection provided in an embodiment of this application.
  • FIG. 14 is a schematic diagram of an AI unit monitoring device provided in an embodiment of this application.
  • FIG. 15 is a schematic diagram of another AI unit monitoring device provided in an embodiment of this application.
  • Figure 16 is a structural diagram of a device provided in an embodiment of this application.
  • Figure 17 is a structural diagram of another device provided in an embodiment of this application.
  • Figure 18 is a structural diagram of another device provided in an embodiment of this application.
  • Figure 19 is a structural diagram of another device provided in an embodiment of this application.
  • first and second are used to distinguish similar objects and not to describe a specific order or sequence. It should be understood that such terms can be used interchangeably where appropriate so that embodiments of this application can be implemented in orders other than those illustrated or described herein, and the objects distinguished by “first” and “second” are generally of the same class, not limited in number; for example, the first object can be one or more.
  • “or” in this application indicates at least one of the connected objects.
  • the scope of protection for "A or B” covers at least three scenarios: Scenario 1: including A but not B; Scenario 2: including B but not A; Scenario 3: including both A and B.
  • the terms “A and/or B,” “at least one of A and B,” and “at least one of A or B” also cover at least the above three scenarios.
  • the character “/” generally indicates that the preceding and following objects are in an "or” relationship.
  • instruction in this application can be either a direct instruction (or explicit instruction) or an indirect instruction (or implicit instruction).
  • a direct instruction can be understood as one in which the sender explicitly informs the receiver of specific information, the operation to be performed, or the requested result, etc., in the instruction sent.
  • An indirect instruction can be understood as one in which the receiver determines the corresponding information based on the instruction sent by the sender, or makes a judgment and determines the operation to be performed or the requested result, etc., based on the judgment result.
  • LTE Long Term Evolution
  • LTE-A Long Term Evolution-Advanced
  • CDMA Code Division Multiple Access
  • TDMA Time Division Multiple Access
  • FDMA Frequency Division Multiple Access
  • OFDMA Orthogonal Frequency Division Multiple Access
  • SC-FDMA Single-carrier Frequency-Division Multiple Access
  • NR New Radio
  • FIG. 1 shows a block diagram of a wireless communication system applicable to an embodiment of this application.
  • the wireless communication system includes a terminal 11 and a network-side device 12.
  • Terminal 11 can be a mobile phone, tablet computer, laptop computer, notebook computer, personal digital assistant (PDA), handheld computer, netbook, ultra-mobile personal computer (UMPC), mobile internet device (MID), augmented reality (AR), virtual reality (VR) device, robot, wearable device, flight vehicle, vehicle user equipment (VUE), shipboard equipment, pedestrian user equipment (PUE), smart home (home devices with wireless communication capabilities, such as refrigerators, televisions, washing machines, or furniture), game console, personal computer (PC), ATM, or self-service machine, etc.
  • PDA personal digital assistant
  • UMPC ultra-mobile personal computer
  • MID mobile internet device
  • AR augmented reality
  • VR virtual reality
  • robot wearable device
  • flight vehicle vehicle user equipment
  • VUE shipboard equipment
  • pedestrian user equipment PUE
  • smart home home devices with wireless communication capabilities, such as refrigerators, televisions, washing machines
  • Wearable devices include: smartwatches, smart bracelets, smart headphones, smart glasses, smart jewelry (smart bracelets, smart chains, smart rings, smart necklaces, smart anklets, smart anklets, etc.), smart wristbands, smart clothing, etc.
  • in-vehicle devices can also be referred to as in-vehicle terminals, in-vehicle controllers, in-vehicle modules, in-vehicle components, in-vehicle chips, or in-vehicle units, etc. It should be noted that the specific type of terminal 11 is not limited in this application embodiment.
  • Network-side equipment 12 may include access network equipment or core network equipment, wherein access network equipment may also be referred to as Radio Access Network (RAN) equipment, radio access network function, or radio access network unit.
  • Access network equipment may include base stations, Wireless Local Area Network (WLAN) access points (APs), or Wireless Fidelity (WiFi) nodes, etc.
  • WLAN Wireless Local Area Network
  • WiFi Wireless Fidelity
  • a base station may be referred to as a Node B (NB), Evolved Node B (eNB), Next Generation Node B (gNB), New Radio Node B (NR Node B), Access Point, Relay Base Station (RBS), Serving Base Station (SBS), Base Transceiver Station (BTS), Radio Base Station, Radio Transceiver, Basic Service Set (BSS), Extended Service Set (ESS), Home Node B (HNB), Home Evolved Node B, Transmit/Receive Point (TRP), or any other suitable term in the relevant field, as long as the same technical effect is achieved.
  • the base station is not limited to specific technical terms. It should be noted that in this application embodiment, only a base station in an NR system is used as an example for introduction, and the specific type of base station is not limited.
  • Core network equipment also known as core network nodes, core network functions, or core network elements, includes, but is not limited to, at least one of the following: Mobility Management Entity (MME), Access and Mobility Management Function (AMF), Session Management Function (SMF), User Plane Function (UPF), Policy Control Function (PCF), Policy and Charging Rules Function (PCRF), Edge Application Server Discovery Function (EASDF), Unified Data Management (UDM), and Unified Data Warehouse (UDM).
  • MME Mobility Management Entity
  • AMF Access and Mobility Management Function
  • SMF Session Management Function
  • UPF User Plane Function
  • PCF Policy Control Function
  • PCF Policy and Charging Rules Function
  • EASDF Edge Application Server Discovery Function
  • UDM Unified Data Management
  • UDM Unified Data Management
  • UDM Unified Data Warehouse
  • the core network equipment includes: Data Repository (UDR), Home Subscriber Server (HSS), Centralized Network Configuration (CNC), Network Repository Function (NRF), Network Exposure Function (NEF), Local NEF (or L-NEF), Binding Support Function (BSF), Application Function (AF), Location Management Function (LMF), Gateway Mobile Location Centre (GMLC), and Network Data Analytics Function (NWDAF).
  • UDR Data Repository
  • HSS Home Subscriber Server
  • CNC Centralized Network Configuration
  • NEF Network Exposure Function
  • L-NEF Local NEF
  • BSF Binding Support Function
  • AF Application Function
  • LMF Location Management Function
  • GMLC Gateway Mobile Location Centre
  • NWDAF Network Data Analytics Function
  • the core network equipment can be implemented by one or more functional modules in a single device, or by multiple devices working together; this application does not specifically limit this. It is understood that the aforementioned functional modules can be network elements in hardware devices, software functional modules running on dedicated hardware, or virtualized functional modules instantiated on a platform (e.g., a cloud platform).
  • a platform e.g., a cloud platform
  • network-side devices and terminals in addition to communication capabilities, may possess sensing capabilities.
  • Sensing capabilities refer to the ability of one or more devices to sense information such as the location, distance, and speed of a target object through the transmission and reception of wireless signals, or to detect, track, identify, and image target objects, events, or environments.
  • Integrated communication and sensing refers to the integrated design of communication and sensing functions in the same system through spectrum sharing and hardware sharing. While transmitting information, the system can sense information such as location, distance, and speed, and detect, track, and identify target devices or events. The communication system and the sensing system complement each other, thereby improving overall performance and bringing a better service experience.
  • the integration of communication and radar is a typical application of communication and sensing integration (communication and sensing fusion).
  • the integration of communication and radar systems can bring many advantages, such as cost savings, size reduction, power consumption reduction, improved spectrum efficiency, and reduced mutual interference, thereby improving the overall system performance.
  • each sensing link in Figure 2 is illustrated with one transmitting node and one receiving node. In actual systems, different sensing links can be selected according to different sensing requirements. Each sensing link may have one or more transmitting and receiving nodes, and the actual sensing system may include multiple different sensing links.
  • the sensing targets in Figure 2 are people and vehicles as examples, and it is assumed that neither people nor vehicles carry or have installed signal transceiver equipment. The sensing targets in actual scenarios will be much more diverse.
  • Sensing Link 1 Base station self-transmitting and self-receiving sensing. In this method, the base station sends sensing signals and obtains the sensing results by receiving the echo of these signals;
  • Sensing Link 2 Inter-base station air interface sensing. In this mode, base station 2 receives sensing signals sent by base station 1 and obtains the sensing results.
  • Sensing Link 3 Uplink air interface sensing. In this mode, the base station receives sensing signals sent by the terminal and obtains the sensing results.
  • Sensing Link 4 Downlink Air Interface Sensing. In this mode, the terminal receives sensing signals sent by the base station and obtains the sensing results.
  • Sensing Link 5 Terminal Self-Sending and Receiving Sensing.
  • the terminal sends a sensing signal and obtains the sensing result by receiving the echo of the sensing signal.
  • Sensing Link 6 Sidelink sensing between terminals. For example, terminal 2 receives a sensing signal sent by terminal 1 and obtains a sensing result, or terminal 1 receives a sensing signal sent by terminal 2 and obtains a sensing result.
  • signaling transmission between radio access network devices and terminals, and between different terminals may be via Radio Resource Control (RRC) signaling, Medium Access Control Control Element (MAC CE), Layer 1 signaling, or other newly defined sensing signaling; signaling transmission between sensing network functions and terminals may be via Non-Access-Stratum (NAS) signaling (forwarded via AMF), or via RRC signaling, MAC CE, Layer 1 signaling, or other newly defined sensing signaling; interaction between sensing network functions and base stations may be via AMF forwarding to the radio access network through the N2 interface; or the core network sensing network function may send the signal to the UPF, which in turn sends it to the radio access network through the N3 interface; or the signal may be sent to the radio access network (e.g., a base station) through a newly defined interface; signaling transmission between radio access network devices may be via the Xn interface.
  • RRC Radio Resource Control
  • MAC CE Medium Access Control Element
  • Layer 1 signaling or other newly defined sensing signaling
  • the sensing network function can also be called a sensing network element or sensing management function (Sensing MF). It can be located on the radio access network side or the core network side, that is, it can be a radio access network device or a core network device. Specifically, it can refer to a network node in the core network or RAN that is responsible for at least one of the following functions: sensing request processing, sensing resource scheduling, sensing information interaction, and sensing data processing. It can be an upgrade based on the AMF or LMF in the mobile communication network, or it can be other network nodes or newly defined network nodes. Specifically, the functional characteristics of the sensing network function/sensing network element can include at least one of the following:
  • the system interacts with wireless signal transmitting devices or wireless signal measuring devices (including target terminals or base stations serving the target terminals or associated with the target area) to exchange target information.
  • the target information includes sensing processing requests, sensing capabilities, sensing auxiliary data, sensing measurement types, sensing resource configuration information, etc., in order to obtain the value of the target sensing result or sensing measurement (uplink measurement or downlink measurement) sent by the wireless signal measuring device.
  • the wireless signal can also be referred to as the sensing signal.
  • the sensing method used is determined based on factors such as the type of sensing service, the information of sensing service consumers, the required Quality of Service (QoS) requirements, the sensing capabilities of the wireless signal transmitting equipment, and the sensing capabilities of the wireless signal measuring equipment.
  • the sensing method may include: wireless access network device A transmitting and wireless access network device B receiving, or wireless access network device transmitting and terminal receiving, or wireless access network device A transmitting and receiving, or terminal transmitting and receiving, or terminal A transmitting and terminal B receiving, etc.
  • the sensing equipment serving the sensing service is determined based on factors such as the type of sensing service, information about the sensing service consumers, the required sensing QoS requirements, the sensing capabilities of the wireless signal transmitting equipment, and the sensing capabilities of the wireless signal measuring equipment.
  • the sensing equipment includes either wireless signal transmitting equipment or wireless signal measuring equipment.
  • the overall coordination and scheduling of resources required for managing sensing services such as configuring sensing resources for wireless access network devices or terminals accordingly;
  • the system processes or calculates the values of sensed measurements to obtain sensing results. It can also verify sensing results and estimate sensing accuracy.
  • the LMF is a core network element in the 5G core network that provides control plane positioning, completes the calculation and feedback of location information in the 5G network, and provides functions such as positioning process management, terminal capability acquisition, auxiliary data provision, and terminal location estimation. Specifically, it provides at least one of the following functions:
  • NG RAN Next Generation Radio Access Network
  • LMF supports medium-to-high precision positioning methods such as cell identifier (CELL ID), uplink time difference of arrival (UL-TDOA), and Assisting Global Navigation Satellite System (A-GNSS).
  • CELL ID cell identifier
  • UL-TDOA uplink time difference of arrival
  • A-GNSS Assisting Global Navigation Satellite System
  • radars can be classified into monostatic radars and bistatic/multistatic radars based on whether the transmitter and receiver are separate.
  • Bistatic radars generally require a long distance between the transmitting and receiving antennas, comparable to the radar's effective range.
  • external radiation source radar is a special case of bistatic radar. It utilizes relevant electromagnetic wave detection theories and signal processing techniques to acquire non-cooperative electromagnetic signals emitted by a third party (such as a communication base station) to achieve target detection, location, tracking, and identification. It is also called passive radar, bistatic/multistatic passive radar, passive radar, non-cooperative illumination source radar, or non-cooperative passive detection system.
  • RT is the distance from the signal transmitter (Tx) to the target
  • RR is the distance from the signal receiver (Tx) to the target
  • L is the baseline distance
  • ⁇ T is the angle of the target relative to the signal transmitter
  • ⁇ R ⁇ R1 , ⁇ R2 ) are the angles of the target relative to the signal receiver
  • ⁇ is the bistatic angle.
  • AI has been widely applied in various fields. Integrating artificial intelligence into wireless communication networks to significantly improve technical indicators such as throughput, latency, and user capacity is an important task for future wireless communication networks.
  • AI modules can be implemented in various ways, such as neural networks, decision trees, support vector machines, and Bayesian classifiers.
  • a neural network is used as an example for illustration, but the specific type of AI module is not limited.
  • a schematic diagram of a neural network is shown in Figure 4.
  • the neural network consists of neurons, and a schematic diagram of a neuron is shown in Figure 5.
  • a1, a2, ..., aK are the inputs
  • w is the weight (multiplicative coefficient)
  • b is the bias (additive coefficient)
  • ⁇ (.) is the activation function.
  • Common activation functions include Sigmoid, tanh, Rectified Linear Unit (ReLU), etc.
  • the parameters of the neural network are optimized using gradient optimization algorithms.
  • Gradient optimization algorithms are a class of algorithms that minimize or maximize an objective function (sometimes called a loss function), which is often a mathematical combination of model parameters and data.
  • an objective function sometimes called a loss function
  • f(.) the probability density function
  • the optimization algorithm can be based on the error back propagation (BP) algorithm.
  • BP error back propagation
  • the basic idea of the BP algorithm is that the learning process consists of two processes: forward propagation of the signal and backward propagation of the error.
  • forward propagation the input sample is introduced from the input layer, processed layer by layer by the hidden layers, and then propagated to the output layer. If the actual output of the output layer does not match the expected output, the process transitions to the error back propagation stage.
  • Error back propagation involves propagating the output error back to the input layer layer by layer through the hidden layers in a certain form, distributing the error to all units in each layer, thereby obtaining the error signal of each unit. This error signal serves as the basis for adjusting the weights of each unit.
  • This process of adjusting the weights of each layer through forward and backward propagation is repeated continuously. This continuous adjustment of weights is the learning and training process of the network. This process continues until the error of the network output is reduced to an acceptable level, or until the predetermined number of learning iterations is reached.
  • optimization algorithms include gradient descent, stochastic gradient descent (SGD), mini-batch gradient descent, momentum descent, Nesterov (named after the inventor, specifically stochastic gradient descent with momentum), adaptive gradient descent (Adagrad), Adadelta, root mean square propagation (RMSprop), and adaptive momentum estimation (Adam).
  • these optimization algorithms calculate the gradient based on the error/loss obtained from the loss function with respect to the current neuron, add the learning rate, previous gradients/derivatives/partial derivatives, etc., and then pass the gradient to the previous layer.
  • the AI unit may also be referred to as an AI model, machine learning (ML) model, ML unit, AI structure, AI function, AI characteristic, machine learning model, neural network, neural network function, neural network functionality, etc.
  • an AI unit may refer to a processing unit capable of implementing specific algorithms, formulas, processing flows, capabilities, etc., related to AI.
  • An AI unit may also be a processing method, algorithm, function, module, or unit for a specific dataset.
  • an AI unit may be a processing method, algorithm, function, module, or unit running on AI/ML related hardware such as a graphics processing unit (GPU), neural processing unit (NPU), tensor processing unit (TPU), or application-specific integrated circuit (ASIC). This application embodiment does not specifically limit this.
  • the aforementioned specific dataset includes the input and/or output of the AI unit.
  • the identifier of the AI unit may be an AI model identifier, an AI structure identifier, an AI algorithm identifier, or an identifier of a specific dataset associated with the AI unit, or an identifier of a specific scenario, environment, channel characteristics, or device related to AI/ML, or an identifier of a function, feature, capability, or module related to AI/ML. This application embodiment does not specifically limit this.
  • Figure 6 is a flowchart of an AI unit monitoring method provided in an embodiment of this application. As shown in Figure 6, it includes the following steps:
  • Step 601 The first device acquires the perception data output by the first AI unit.
  • the aforementioned first device can be a terminal or a network-side device, such as a wireless access network device or a core network device.
  • the first AI unit mentioned above is a perception-related AI unit, such as the perception data output by the first AI unit being the perception result or intermediate data used to determine the perception result.
  • the aforementioned first AI unit may be an AI unit deployed on the aforementioned first device, such as the aforementioned perception data being perception data obtained by the aforementioned first device using the aforementioned first AI unit for reasoning.
  • the aforementioned first AI unit may be an AI unit deployed on other devices, such as the aforementioned perception data being perception data output by the first AI unit sent by other devices to the first device.
  • the device on which the first AI unit is deployed is not limited.
  • the first AI unit may be deployed on the aforementioned first device, or it may be deployed on other devices.
  • the first AI unit may be deployed in the core network sensing network function or positioning management function, or it may be deployed on a base station or terminal.
  • the base station or terminal may include a base station or terminal responsible for transmitting or receiving the aforementioned sensing-related signals and processing sensing data, or a base station or terminal that is not responsible for transmitting or receiving the aforementioned sensing-related signals but only for processing sensing data.
  • the perception data output by the first AI unit can also be called the perception data obtained through reasoning by the first AI unit. Specifically, it can be the perception result or the intermediate data used to determine the perception result.
  • Step 602 The first device acquires reference data for monitoring the first AI unit.
  • the reference data acquired by the first device for monitoring the first AI unit can be reference data obtained by the first device through measurement, calculation, or inference.
  • the reference data acquired by the first device for monitoring the first AI unit can be reference data received by the first device from other devices.
  • the aforementioned reference data can be understood as the actual label or reference label of the first AI unit, that is, as a reference for the perception data output by the first AI unit.
  • steps 601 and 602 are not limited.
  • step 601 may be executed before step 602, or steps 601 and 602 may be executed simultaneously, and step 601 may be executed after step 602.
  • Step 603 The first device determines the monitoring result of the first AI unit based on the reference data and the perception data output by the first AI unit.
  • the first device determines the monitoring result of the first AI unit based on the reference data and the perceived data output by the first AI unit. This can be achieved by comparing the reference data with the perceived data, determining the difference between the two, and then determining the monitoring result of the first AI unit. Alternatively, it can be by calculating the loss value between the reference data and the perceived data output by the first AI unit, and determining the monitoring result of the first AI unit based on this loss value, etc., without any specific limitation.
  • the first device acquires the perception data output by the first AI unit; the first device acquires reference data for monitoring the first AI unit; and the first device determines the monitoring result of the first AI unit based on the reference data and the perception data output by the first AI unit.
  • the first device determines the monitoring result of the first AI unit based on the reference data and the perception data output by the first AI unit.
  • the monitoring results of the first AI unit are determined based on the reference data and the perception data output by the first AI unit, the monitoring of the first AI unit is realized, thereby ensuring the effective operation of the first AI unit and ensuring the reliability of the perception performance based on the first AI.
  • the first device determines the monitoring result of the first AI unit based on the reference data and the perception data output by the first AI unit, performance monitoring can be performed to address the generalization problem of the first AI unit, thereby avoiding a significant decline in the performance of the AI unit.
  • the reference data includes at least one of the following:
  • the perception data output by the second AI unit has a processing capability that is stronger than that of the first AI unit.
  • non-AI methods refer to methods that do not rely on AI reasoning.
  • the perception data acquired through non-AI methods includes at least one of the following:
  • Perception data acquired through target perception devices acquired through target perception devices.
  • the aforementioned estimation algorithm can be a non-AI parameter estimation algorithm. For example, it can acquire measurement data using wireless sensing, but instead of inputting the acquired measurement data into the AI unit to obtain sensing data, it can calculate the sensing data through other methods. For instance, sensing data calculated based on channel information using spectral estimation or parameter estimation algorithms can be used as reference data for sensing data inferred from the first AI unit. Sensing data calculated using algorithms such as 2D-Fast Fourier Transform (2D-FFT), 3D-Fast Fourier Transform (3D-FFT), Multiple Signal Classification (MUSIC), or ESPRIT can be used as reference data for sensing data inferred from the first AI unit.
  • 2D-FFT 2D-Fast Fourier Transform
  • 3D-Fast Fourier Transform 3D-Fast Fourier Transform
  • MUSIC Multiple Signal Classification
  • ESPRIT can be used as reference data for sensing data inferred from the first AI unit.
  • the aforementioned sensor data can be obtained through sensors such as visible light cameras, infrared cameras, Global Navigation Satellite System (GNSS), lidar, millimeter-wave radar, thermometers, hygrometers, barometers, gyroscopes, accelerometers, magnetometers, gravity sensors, sonar, and rain gauges.
  • the sensor data can be used as reference data for the sensor data obtained by reasoning based on the first AI unit in wireless sensing.
  • the aforementioned target perception mode can be a monostatic perception mode or a bistatic perception mode.
  • the perception data obtained by the monostatic perception mode can be used as reference data for the perception data obtained by reasoning through the first AI unit in the bistatic perception mode.
  • the aforementioned target sensing device can be a specific device, such as a base station or a terminal.
  • the sensing data acquired by the base station can be used as reference data for the sensing data obtained by the terminal through reasoning by the first AI unit.
  • the perception data output by the second AI unit can be understood as reference data for the perception data obtained by the first AI unit through reasoning, using the perception data output by the more capable AI unit.
  • reference data mentioned above may be reference data obtained by the first device in the manner described above, or it may be reference data sent by other devices to the first device, that is, other devices obtain reference data in the manner described above and send it to the first device.
  • reference data in multiple ways can be obtained to improve the flexibility of AI unit monitoring and meet the needs of more business or scenarios.
  • the first device acquires the perception data output by the first AI unit, including:
  • the first device receives the first measurement data
  • the first measurement data is used as input to the first AI unit for inference to obtain the perception data output by the first AI unit, and the first measurement data includes the first measurement result;
  • the first measurement data includes the perception data output by the first AI unit.
  • the aforementioned receiving of the first measurement data may be receiving measurement data sent by the second device, that is, the aforementioned measurement data is acquired by the second device, which may be a terminal or a network-side device.
  • the aforementioned first measurement data is used to input the first AI unit for inference to obtain the perception data output by the first AI unit.
  • This can be understood as the perception data being obtained through AI inference based on the aforementioned first measurement data.
  • the aforementioned first device inputs the aforementioned first measurement data into the aforementioned first AI unit for inference to obtain the aforementioned perception data.
  • the first measurement result mentioned above is the measurement result obtained by executing the measurement process. It can be a measurement result such as channel information, spectrum information, time delay, Doppler, angle or intensity, and specifically it can be the value of a specific measurement quantity or the value of a measurement quantity that meets specific format requirements.
  • the aforementioned first measurement data can be understood as at least inputting the first measurement result into the aforementioned first AI unit for inference.
  • the aforementioned first measurement result may be the measurement result obtained by the second device executing the measurement process.
  • the aforementioned first measurement result may be measurement results such as channel information, spectral information, or basic measurement quantities.
  • the aforementioned first measurement data can be understood as the perception data being received by the first device from other devices. That is, the first AI unit is deployed on other devices, such as on the second device. If the second device executes the measurement process, obtains the measurement result, and then inputs the measurement result into the first AI unit for reasoning to obtain the perception data.
  • the first measurement data further includes: first explanatory information, which includes at least one of the following:
  • the first measurement data includes timestamp information, perception mode information, indication information for indicating that the first measurement result is used for AI unit inference, information of the AI unit corresponding to the first measurement data, performance index information of the first measurement data, indication information for indicating that the first measurement data includes perception data output by the first AI unit, reference data type information associated with the first measurement data, accuracy information corresponding to the first measurement data, performance level information corresponding to the first measurement data, and confidence level information corresponding to the first measurement data.
  • the timestamp information mentioned above can be absolute time, such as International Standard Time (UTC), or time relative to a certain reference time point, such as radio frame number, half-frame number, subframe number, transmission time interval (TTI) sequence number, slot number, sub-slot number, symbol number, etc.; wherein, the slot number or symbol number can also refer to the sequence number of the starting slot or symbol relative to the sensing coherent processing time.
  • UTC International Standard Time
  • TTI transmission time interval
  • the sensing mode information is used to indicate the sensing mode corresponding to the first measurement data, such as at least one of the sensing modes shown in Figure 2.
  • the aforementioned indication information for indicating that the first measurement result is used for AI unit inference can be understood as indicating that the first measurement result is a measurement result carried by the AI unit, or indicating that the first measurement result is a common measurement result used for AI inference and AI unit monitoring, that is, the first measurement result can be used for AI inference and AI unit monitoring.
  • the information of the AI unit corresponding to the first measurement data is used to indicate the identifier of the AI unit to which the first measurement data is input, such as the model ID, specifically indicating which AI unit the first measurement data is input to.
  • the aforementioned performance metrics may include, but are not limited to, Reference Signal Received Power (RSRP), Reference Signal Received Quality (RSRQ), Signal to Interference plus Noise Ratio (SINR), and sensing performance metrics.
  • RSRP Reference Signal Received Power
  • RSS Reference Signal Received Quality
  • SINR Signal to Interference plus Noise Ratio
  • Performance index information can better indicate the performance of measurement data, so the first device can selectively perform AI unit based on the first measurement data to better ensure perception performance. For example, if the performance index information indicates that the performance index of the first measurement data is higher than a preset threshold, the aforementioned perception data can be obtained by performing AI inference based on the first measurement data.
  • the aforementioned reference data type information can be perceptual data obtained through non-AI means or perceptual data output by the second AI unit.
  • the aforementioned accuracy information can represent the accuracy of the first measurement result, specifically the specific accuracy or accuracy level.
  • the aforementioned performance level information can represent the performance level of the first measurement result, such as the level of the performance index indicating the first measurement result.
  • the confidence level information described above can represent the confidence level of the first measurement result.
  • the reliability of the perception data output by the first AI unit can be improved when performing AI inference based on the first measurement data.
  • the input data of the first AI unit includes the aforementioned first measurement result and all or part of the aforementioned first explanatory information; or in some embodiments, the input data of the first AI unit includes the aforementioned first measurement result but does not include the aforementioned first explanatory information.
  • the information related to the first measurement data in the above first explanatory information can be understood as information related to the first measurement result.
  • the timestamp information of the first measurement data can be understood as the timestamp information of the first measurement result
  • the perception mode information of the first measurement data can be understood as the perception mode information of the first measurement result
  • the information of the AI unit corresponding to the first measurement data can be understood as the information of the AI unit corresponding to the first measurement result
  • the performance index information of the first measurement data can be understood as the performance index information of the first measurement result
  • the reference data type information associated with the first measurement data can be understood as the reference data type information associated with the first measurement result
  • the accuracy information corresponding to the first measurement data can be understood as the accuracy information of the first measurement result
  • the performance level information corresponding to the first measurement data can be understood as the performance level information corresponding to the first measurement result
  • the confidence level information corresponding to the first measurement data can be understood as the confidence level information corresponding to the first measurement result.
  • the first device acquires reference data for monitoring the first AI unit, including:
  • the first device receives the second measurement data
  • the second measurement data is used to determine the reference data, and the second measurement data includes the second measurement result; or,
  • the second measurement data includes the reference data.
  • the aforementioned second measurement data used to determine the reference data can be understood as the first device determining the reference data based on the measurement data after receiving the second measurement data, such as inputting the second measurement result into the second AI unit for reasoning to obtain the reference data, or performing non-AI calculations based on the second measurement result to obtain the reference data.
  • the second measurement result mentioned above is the measurement result obtained by executing the measurement process. It can be a measurement result such as channel information, spectrum information, time delay, Doppler, angle or intensity, and specifically it can be the value of a specific measurement quantity or the value of a measurement quantity that meets specific format requirements.
  • the aforementioned second measurement data can be understood as the reference data being generated by other devices, and the first device receiving the reference data sent by other devices.
  • multiple methods can be used to obtain reference data, thereby improving the flexibility of AI unit monitoring.
  • the second measurement data further includes second explanatory information, which includes at least one of the following:
  • the second measurement data includes timestamp information, perception mode information, indication information for monitoring the second measurement result, reference data type information associated with the second measurement data, performance index information of the second measurement data, indication information for indicating that the second measurement data includes the reference data, information of the AI unit corresponding to the second measurement data, accuracy information corresponding to the second measurement data, coordinate system information of the second measurement data, and sensor type information corresponding to the second measurement data.
  • the measurement result of the second measurement data is considered to be a measurement result that can be used for monitoring of the first AI unit.
  • the coordinate system information mentioned above can be the relationship between the sensor's corresponding coordinate system and the global coordinate system, such as the transformation parameters from the local coordinate system to the global coordinate system, that is, the rotation angles of the local coordinate system relative to the global coordinate system: ⁇ (bearing angle), ⁇ (tilt angle) and ⁇ (slant angle).
  • the second measurement data since the second measurement data also includes the aforementioned second explanatory information, the reliability of the reference data can be improved when determining the reference data based on the second measurement data.
  • the input data of the second AI unit includes the aforementioned second measurement result and all or part of the aforementioned second explanatory information; or in some embodiments, the input data of the second AI unit includes the aforementioned second measurement result but does not include the aforementioned second explanatory information.
  • the information related to the second measurement data in the above second explanatory information can be understood as information related to the second measurement result.
  • the timestamp information of the second measurement data can be understood as the timestamp information of the second measurement result
  • the perception mode information of the second measurement data can be understood as the perception mode information of the second measurement result
  • the information of the AI unit corresponding to the second measurement data can be understood as the information of the AI unit corresponding to the second measurement result
  • the performance index information of the second measurement data can be understood as the performance index information of the second measurement result, etc.
  • the monitoring results include at least one of the following:
  • Effectiveness information for performance level information, and perception accuracy information
  • the perception accuracy information includes at least one of the following:
  • Distance accuracy information speed accuracy information, angle accuracy information, positioning accuracy information, detection accuracy information, and recognition accuracy information.
  • the aforementioned validity information may be used to indicate whether the first AI unit is valid or whether the first AI unit is applicable.
  • the aforementioned perception accuracy information can be perception error information or perception error statistics.
  • the first AI unit can be monitored from multiple dimensions to improve the monitoring effect of the AI unit.
  • the aforementioned distance accuracy information can represent how close the estimated distance of the target object is to its true distance (i.e., the distance represented by the reference data).
  • the distance of the target object can refer to the distance between the target object and the signal receiving device;
  • the distance of the target object can refer to the distance between the target object and the signal receiving device, or the sum of the distances between the target object and the signal receiving device and the transmitting device, or the sum of the distances between the target object and the signal receiving device and the transmitting device minus the distance between the transceivers, or the projection of the sum of the distances between the target object and the signal receiving device and the transmitting device onto the bistatic angle bisector.
  • the aforementioned distance accuracy information can be any target distance error information calculated based on the results of a single sensing operation, i.e.
  • the aforementioned distance accuracy information can also be the sum of all target distance errors calculated based on a single sensing result, i.e.
  • the aforementioned distance accuracy information can also be calculated based on the root mean square error of the distance obtained from multiple sensing results, i.e.:
  • R ⁇ sub> ij ⁇ /sub> be the distance value of the j-th target in the perception data output by the first AI unit obtained in the i-th calculation
  • R ⁇ sub>ij ⁇ /sub> be the distance value of the j-th target in the reference data obtained in the i-th calculation
  • N represents the number of perception result calculations (or the number of times the second device reports the first and second measurement data)
  • M represents the number of targets in each perception result calculation.
  • the aforementioned velocity accuracy information indicates how close the measured velocity of the target object is to its true velocity (i.e., the velocity represented by the reference data).
  • the velocity of the target object can be the projection of the target object's velocity onto the line connecting the target and the signal receiving device.
  • the velocity of the target object can be the projection of the target object's velocity onto the bistatic angle bisector, or the sum of the projections of the target object's velocity onto the line connecting the target and the signal receiving device and the projections onto the line connecting the target and the signal transmitting device.
  • it can also refer to the target object's rotational speed, repetition rate of motion (e.g., respiratory/heart rate), etc.
  • the aforementioned speed accuracy information can be any target speed error information calculated based on the results of a single sensing operation, i.e.
  • the aforementioned speed accuracy information can also be the sum of all target speed errors calculated based on a single sensing result, i.e.
  • the aforementioned speed accuracy information can also be calculated based on the root mean square error of the speed obtained from multiple sensing results, i.e.:
  • the aforementioned angular accuracy indicates how close the measured angle of the target object is to its true angle (i.e., the angle represented by the reference data), including azimuth and elevation accuracy.
  • the angle of the target object can refer to the angle of arrival of the target object relative to the signal receiving device; for bistatic sensing, the angle of the target object can refer to the angle of arrival of the target object relative to the signal receiving device, or the angle of departure of the target object relative to the signal transmitting device.
  • the aforementioned angle accuracy information can be any target angle error information calculated based on the result of a single sensing operation, i.e.
  • the aforementioned angle accuracy information can also be the sum of all target angle errors calculated based on a single sensing result, i.e.
  • the aforementioned angle accuracy information can also be calculated from the root mean square error of the angle based on multiple sensing results, i.e.:
  • the aforementioned positioning accuracy information indicates how close the measured position of the target object is to its true position (i.e., the position represented by the reference data).
  • the target object's position coordinates can be further calculated from at least one of its distance and angle.
  • the aforementioned positioning accuracy information can be any target position coordinate error information calculated based on the results of a single sensing operation, i.e.:
  • the aforementioned positioning accuracy information can also be the sum of all target position coordinate errors calculated based on a single sensing result, i.e.:
  • the aforementioned positioning accuracy information can also be calculated based on the root mean square error of the position coordinates obtained from multiple sensing results, i.e.:
  • the aforementioned detection accuracy information or recognition accuracy information may represent the number of times the detection/recognition results of the perception data output by the first AI unit are consistent with or inconsistent with the reference data regarding the presence, number, or category of the target; or the aforementioned detection accuracy information or recognition accuracy information may also be the proportion of the number of times the detection/recognition results of the perception data output by the first AI unit are consistent with or inconsistent with the reference data regarding the presence, number, or category of the target to the total number of perception result calculations.
  • the method further includes:
  • the first device sends target information, which includes at least one of the following: signal configuration information, measurement configuration information, and indication information; wherein the indication information is used to indicate at least one of the following:
  • the first measurement data is used to input the first AI unit for reasoning to obtain the perception data output by the first AI unit, and the first measurement data includes the first measurement result.
  • the first device sending the target information may be sending the target information to the second device.
  • the aforementioned signal configuration information includes configuration information for signals used to acquire input data of the first AI unit, and may also include configuration information for signals used to acquire the aforementioned reference data.
  • the signal used to acquire the input data of the first AI unit refers to a signal upon which the perceived data is acquired; for example, measuring the signal yields a measurement result, which serves as the input data for the first AI unit.
  • the signal used to acquire the aforementioned reference data refers to a signal upon which the reference data is acquired; for example, measuring the signal yields a measurement result, and then determining the reference data based on that measurement result.
  • the signal used to acquire the input data of the first AI unit and the signal used to acquire the reference data are the same signal or different signals, or the signal used to acquire the input data of the first AI unit is a subset of the signal used to acquire the reference data.
  • the above measurement configuration information can sense data-related measurements or reference signal-related measurements.
  • the aforementioned instruction information to acquire first measurement data, acquire second measurement data, or report first measurement data and report second measurement data refers to instructing the device receiving the aforementioned target information to acquire first measurement data, acquire second measurement data, or report first measurement data and report second measurement data, such as instructing a second device to acquire first measurement data, acquire second measurement data, or report first measurement data and report second measurement data.
  • an instruction if an instruction is given to obtain first measurement data, it can implicitly indicate that the first measurement data be reported; if an instruction is given to report first measurement data, it can implicitly indicate that the first measurement data be obtained; if an instruction is given to obtain second measurement data, it can implicitly indicate that the second measurement data be reported; if an instruction is given to report second measurement data, it can implicitly indicate that the second measurement data be obtained.
  • the acquisition of measurement data or the reporting of first measurement data, second measurement data, first measurement data, and second measurement data can be explicitly or implicitly indicated through the aforementioned signal configuration information or measurement configuration indication.
  • the device receiving the target information can acquire or report measurement data based on the information, thereby enabling the first device to obtain the aforementioned sensing data and reference data in a timely manner, so as to improve the monitoring performance of the AI unit.
  • the signal configuration information includes at least one of the following:
  • the configuration information of the first signal and the configuration information of the second signal wherein the first signal is a signal used to acquire input data of the first AI unit, and the second signal is a signal used to acquire the reference data, and the first signal and the second signal satisfy at least one of the following:
  • the frequency domain resource length of the second signal is greater than that of the first signal
  • the frequency domain resource interval of the second signal is smaller than that of the first signal
  • the time-domain resource length of the second signal is greater than that of the first signal
  • the time-domain resource unit of the second signal is smaller than the time-domain resource interval of the first signal
  • the transmission power of the second signal is greater than the transmission power of the first signal
  • the first signal is a subset of the second signal.
  • the configuration information mentioned above can be configuration information such as the purpose, type, and resources of the signal.
  • the configuration information includes at least one of the following:
  • Signal resource identifiers are used to distinguish different signal resource configurations
  • the signal's purpose indicates whether it's used for communication (e.g., channel measurement, channel estimation, synchronization, carrying data information, etc.), sensing, or both. Specifically, it can also specify the type of sensing service or category of sensing service it's used for.
  • Sensing services can include at least one of the following: detecting target presence, location, speed detection, distance detection, angle detection, acceleration detection, material analysis, composition analysis, shape detection, category classification, radar cross section (RCS) detection, polarization scattering characteristic detection, fall detection, intrusion detection, quantity statistics, indoor positioning, gesture recognition, lip reading, gait recognition, facial expression recognition, face recognition, respiration monitoring, heart rate monitoring, pulse monitoring, humidity/brightness/temperature/atmospheric pressure monitoring, air quality monitoring, weather condition monitoring, environmental reconstruction, terrain and building/vegetation distribution detection, pedestrian or vehicle flow detection, crowd density, vehicle density detection, etc.
  • RCS radar cross section
  • the sensing service type can be a classification of multiple different sensing services based on certain characteristics, such as classifying them into detection-type sensing services according to function.
  • detection-type sensing services include intrusion detection, fall detection), parameter estimation-based perception services (distance, angle, speed calculation), and recognition-based perception services (action recognition, identity recognition), etc.; they can also be classified according to the type of perceived target, including unmanned aerial vehicles (UAVs), humans, automotive vehicles, automated guided vehicles (AGVs), and objects on roads/railways; they can also be classified according to the range of perception (short-range perception, medium-range perception, long-range perception), the level of perception (coarse-grained perception, fine-grained perception, etc.), power consumption/energy consumption, and resource consumption, etc.
  • Waveforms such as orthogonal frequency division multiplex (OFDM), single-carrier frequency-division multiple access (SC-FDMA), orthogonal time-frequency space (OTFS), frequency-modulated continuous wave (FMCW), pulse signals, etc.
  • OFDM orthogonal frequency division multiplex
  • SC-FDMA single-carrier frequency-division multiple access
  • OTFS orthogonal time-frequency space
  • FMCW frequency-modulated continuous wave
  • pulse signals etc.
  • Subcarrier spacing for example, 30 kHz in an OFDM system
  • the guard interval is the time interval between the end of signal transmission and the latest echo signal of the signal being received. This parameter is proportional to the maximum sensing distance. For example, it can be calculated by c/(2R max ), where R max is the maximum sensing distance (belonging to the sensing requirement information). For example, for self-transmitting and self-receiving sensing signals, R max represents the maximum distance from the sensing signal transmission point to the signal transmission point.
  • the OFDM signal cyclic prefix (CP) can serve as the minimum guard interval.
  • CP the OFDM signal cyclic prefix
  • the starting frequency domain position i.e. the starting frequency point, can also be the starting resource element (RE) or resource block (RB) index;
  • the starting time domain position i.e. the starting time point, can also be the starting symbol index, time slot index, or frame index;
  • the terminating frequency domain position i.e., the terminating frequency point, can be represented by the terminating RE and RB indices;
  • the termination time domain position i.e. the termination time point, can be represented using the termination RE and RB indices;
  • Frequency domain resource length i.e. frequency domain bandwidth
  • the frequency domain bandwidth B of each first signal is B ⁇ c/(2 ⁇ R), where c is the speed of light and ⁇ R is the distance resolution.
  • the time-domain resource length also known as the burst duration, is inversely proportional to the Doppler resolution.
  • the frequency domain resource spacing is inversely proportional to the maximum unambiguous distance/delay. For OFDM systems, when subcarriers are continuously mapped, the frequency domain spacing is equal to the subcarrier spacing.
  • the frequency domain resource unit spacing can also be represented by the comb mapping parameter K comb .
  • Time-domain resource interval is the time interval between two adjacent signal resource units.
  • the time-domain resource interval is related to the maximum unambiguous Doppler frequency shift or the maximum unambiguous velocity.
  • Temporal resource characteristics such as periodic transmission, semi-persistent transmission, and non-periodic transmission.
  • the time-domain burst resource interval or time-domain burst transmission period is related to the refresh frequency of the sensing results.
  • Signal power for example, taking a value every 2dBm from -20dBm to 23dBm;
  • Sequence information including sequence type information (ZC sequence, PN sequence, etc.), sequence generation method, sequence length, etc.;
  • Signal direction including the angle or beam information of signal transmission
  • Quasi-co-location (QCL) relationships such as a signal comprising multiple resources, each resource being associated with a Synchronization Signal Block (SSB) QCL, where the QCL can be of type A, type B, type C, or type D.
  • SSB Synchronization Signal Block
  • Cyclic prefix (CP) information can include CP type or CP length, such as normal cyclic prefix (NCP), extended cyclic prefix (ECP), or newly designed sensing measurement-specific CP.
  • NCP normal cyclic prefix
  • ECP extended cyclic prefix
  • the first signal is a subset of the second signal, as shown in Figure 7.
  • the coherent processing time of the sensing signal i.e., the length of the signal time-domain resources required to calculate one sensing information
  • Tp2 80 slots (or Tp1)
  • the transmission period of the first signal is 1 slot
  • the transmission period of the second signal is 0.5 slots.
  • the frequency domain density (frequency domain resource unit spacing) of the first signal is such that one out of every 12 subcarriers is used to carry the first signal, while the second signal occupies consecutive subcarriers in the frequency domain. Therefore, measurements based on the second signal, compared to measurements based on the first signal, have a larger unambiguous ranging range and unambiguous velocity measurement range, and also exhibit higher processing gain.
  • the reliability of the reference data can be higher than that of the perception data output by the first AI unit, thereby enabling better monitoring of the first AI unit.
  • the first signal and the second signal are not limited to satisfying the above relationship.
  • the first signal and the second signal are the same signal.
  • the above signal configuration information may include all or part of it, or it may be agreed upon by the protocol or pre-configured.
  • the measurement configuration information includes at least one of the following:
  • the signal resource indication of the above measurement can indicate at least one of the first signal and the second signal, such as indicating an identifier of at least one of the first signal and the second signal, which has a mapping relationship with the signal resource, thereby implicitly indicating the resource of at least one of the first signal and the second signal to save configuration overhead.
  • the measurement purpose information corresponding to the first measurement data can indicate that the first measurement data is used for AI inference.
  • the measurement purpose information corresponding to the second measurement data mentioned above can indicate that the second measurement data is used for AI unit monitoring.
  • the measurement quantities and formats for different measurement purposes are preset, so that the measurement quantity or format can be determined by the purpose information, thereby saving configuration costs.
  • the above processing information can indicate whether to use AI processing or non-AI processing.
  • AI processing is used for the first measurement data
  • non-AI processing is used for the second measurement data.
  • the measurement quantities or measurement formats corresponding to different processing methods are preset.
  • the measurement quantity of the first measurement data includes at least one of the following:
  • Received signal channel information, spectrum information, and basic measurement quantities
  • the measurement quantity of the second measurement data includes at least one of the following:
  • Received signal channel information, spectrum information, and basic measurement quantities.
  • the aforementioned channel information can be raw channel information, such as time-domain channel response and frequency-domain channel response, and specifically can include at least one of the complex result of the channel response, amplitude/phase, and I-channel/Q-channel data.
  • the aforementioned spectral information can be spectral information calculated based on channel information or received signals, and may include at least one of the following: time delay (distance) spectrum, Doppler (velocity) spectrum, angle spectrum, time delay (distance)-Doppler (velocity) spectrum, time delay (distance)-angle spectrum, time delay (distance)-Doppler (velocity)-angle spectrum, time-Doppler spectrum (micro-Doppler spectrum).
  • the spectral information mentioned above can refer to complex results, such as a time-delay-Doppler spectrum, which refers to the time delay, Doppler index, and corresponding complex values (including phase information) in a 2D spectrum; the spectral information can also refer to a power spectrum, such as a time-delay-Doppler spectrum, which refers to the time delay, Doppler index, and corresponding power values (excluding phase information) in a 2D spectrum.
  • the aforementioned basic measurement quantities may include at least one of the following:
  • Time delay Doppler, angle, intensity, power.
  • the aforementioned basic measurement quantity can be a quantification of the actual value or soft information.
  • the measurement quantity for the second measurement data may also be reference data, including at least one of the following: time delay, Doppler, angle, distance, speed, orientation, spatial position, acceleration, presence of target, number of targets, trajectory, gesture, action, expression, vital signs, quantity, imaging results, weather, air quality, shape, material, and composition of the perceived target.
  • the aforementioned format information is used to represent the format of the measurement data, such as the measurement quantity format.
  • the aforementioned format information of the first measurement data represents the format of the measurement data input to the first AI unit (i.e., the first measurement data), and the aforementioned format information of the second measurement data represents the format of the measurement data used to calculate the reference data (i.e., the second measurement data).
  • the format information of the first measurement data is used to indicate at least one of the following:
  • channel information The dimensions of channel information, the range of spectral information, the types of basic measurements, and the format of sensor data.
  • the format information of the second measurement data is used to indicate at least one of the following:
  • the dimensions of the aforementioned channel information can be frequency domain channel response information on a single symbol (one-dimensional channel information), time-frequency domain channel response information on multiple symbols (two-dimensional channel information), time-frequency spatial domain channel response information of multiple symbols and multiple antennas (three-dimensional channel information), as well as different dimensions of scale, i.e. the number of sampling points or sampling interval (e.g., time-frequency domain density).
  • the range of spectral information can be used to limit the range of different spectral information, or it can be called a truncation window for different dimensions of spectral information, such as a subset of spectral information corresponding to a specific time delay or Doppler range in a time-delay-Doppler spectrum; it can also indicate the upper limit of the number of paths or the upper limit of the number of sampling points of a specific spectrum; or it can indicate the minimum granularity of a specific spectrum, that is, the interval between two adjacent sampling points (corresponding to the sensing resolution).
  • the types of the basic measurements mentioned above can be the quantification method of the measurement value (e.g., quantification granularity) or soft information types, such as providing the mean + variance, or soft information such as confidence intervals and confidence levels.
  • quantification method of the measurement value e.g., quantification granularity
  • soft information types such as providing the mean + variance, or soft information such as confidence intervals and confidence levels.
  • the format of the measurement data can be matched, thereby reducing the amount of computation required for AI inference or determination of reference data.
  • the above reporting configuration can indicate the criteria for reporting the first or second measurement data, such as at least one of the following: the time-frequency domain resource configuration for reporting, the reporting period, and the triggering event for reporting.
  • the triggering event includes at least one of the following:
  • the event of entering a specific area e.g., a residential community
  • the device orientation can be the orientation of components such as the device's antenna or screen;
  • the above-mentioned configuration information allows the second device to make more reliable reports.
  • the sensor type information mentioned above can be the sensor's data format, such as binary format, ASCII format, etc., or it can be a data format associated with a specific sensor type, such as YUV, RGB, etc. for an image sensor (camera).
  • the above measurement method can indicate whether AI processing or non-AI processing is used. If AI processing is indicated, it can also indicate information about the AI unit, as well as the accuracy information of the AI unit, or the performance level or confidence level that can be achieved.
  • the aforementioned measurement configuration information enables the second device to perform more accurate measurements to assess the complexity of AI unit monitoring.
  • measurement configuration information may include all or part of it, or it may be agreed upon in the agreement or pre-configured.
  • the first device sends target information including at least one of the following:
  • the first device sends all or part of the target information to the second device;
  • the first device sends all or part of the target information to the third device;
  • the second device is a device for providing at least one of the first measurement data and the second measurement data
  • the third device is a device for providing the second measurement data.
  • the aforementioned third device can be a terminal or a network-side device.
  • the second device satisfies at least one of the following conditions:
  • the supported transmit power exceeds the preset threshold
  • the number of supported antennas exceeds a preset threshold
  • Supported array aperture exceeds preset threshold
  • the amount of resources available for sensing exceeds a preset threshold
  • the clock synchronization error between the device and the third device is less than a preset threshold.
  • the target perception mode mentioned above can be at least one of the perception modes shown in Figure 2.
  • the aforementioned preset thresholds can be agreed upon in the protocol or configured on the network side.
  • the measurement data provided by the second device can be more reliable, thereby making the AI unit monitoring more reliable.
  • the third device satisfies at least one of the following conditions:
  • the supported transmit power exceeds the preset threshold
  • the number of supported antennas exceeds a preset threshold
  • Supported array aperture exceeds preset threshold
  • the amount of resources available for sensing exceeds a preset threshold
  • the clock synchronization error between the device and the second device is less than a preset threshold.
  • the target perception mode mentioned above can be at least one of the perception modes shown in Figure 2.
  • the aforementioned preset thresholds can be agreed upon in the protocol or configured on the network side.
  • the measurement data provided by the third device can be more reliable, thereby making the AI unit monitoring more reliable.
  • the method further includes:
  • the first device determines a target operation for the first AI unit based on the monitoring results, and the target operation includes at least one of the following:
  • the target operation for the first AI unit determined based on the monitoring results can be a target operation for the first AI unit when the monitoring results meet preset conditions, such as the monitoring results indicating that the first AI unit is invalid, not applicable, has a low performance level, or low accuracy.
  • the first device After determining the target operation, if the first AI unit is deployed on the first device, the first device performs the target operation; if the first AI unit is deployed on the second device, the first device sends a notification to the second device, notifying the second device to perform the target operation.
  • the AI unit is stopped based on monitoring results, such as by using other AI units or other methods to perceive, in order to improve or ensure perception performance.
  • the first device acquires the perception data output by the first AI unit; the first device acquires reference data for monitoring the first AI unit; and the first device determines the monitoring result of the first AI unit based on the reference data and the perception data output by the first AI unit.
  • the first device determines the monitoring result of the first AI unit based on the reference data and the perception data output by the first AI unit.
  • Figure 8 is a flowchart of an AI unit monitoring method provided in an embodiment of this application. As shown in Figure 8, it includes the following steps:
  • Step 801 The second device sends data, wherein the data sent includes at least one of the following:
  • the first measurement data is used to be input into the first AI unit for reasoning to obtain the perception data output by the first AI unit, and the first measurement data includes the first measurement result; or, the first measurement data includes the perception data output by the first AI unit.
  • the second measurement data is used to determine a reference signal, and the second measurement data includes a second measurement result; or, the second measurement data includes reference data used to monitor the first AI unit.
  • the reference data includes at least one of the following:
  • the perception data output by the second AI unit has a processing capability that is stronger than that of the first AI unit.
  • the perception data acquired through non-AI methods includes at least one of the following:
  • Perception data acquired through target perception devices acquired through target perception devices.
  • the first measurement data further includes: first explanatory information, which includes at least one of the following:
  • the first measurement data includes timestamp information, perception mode information, indication information for indicating that the first measurement result is used for AI unit inference, information of the AI unit corresponding to the first measurement data, performance index information of the first measurement data, indication information for indicating that the first measurement data includes perception data output by the first AI unit, reference data type information associated with the first measurement data, accuracy information corresponding to the first measurement data, performance level information corresponding to the first measurement data, and confidence level information corresponding to the first measurement data.
  • the second measurement data further includes second explanatory information, which includes at least one of the following:
  • the second measurement data includes timestamp information, perception mode information, indication information for monitoring the second measurement result, reference data type information associated with the second measurement data, performance index information of the second measurement data, indication information for indicating that the second measurement data includes the reference data, information of the AI unit corresponding to the second measurement data, accuracy information corresponding to the second measurement data, coordinate system information of the second measurement data, and sensor type information corresponding to the second measurement data.
  • the method further includes:
  • the second device receives target information, which includes at least one of the following: signal configuration information, measurement configuration information, and indication information; wherein the indication information is used to indicate at least one of the following:
  • the signal configuration information includes at least one of the following:
  • the configuration information of the first signal and the configuration information of the second signal wherein the first signal is a signal used to acquire input data of the first AI unit, and the second signal is a signal used to acquire the reference data, and the first signal and the second signal satisfy at least one of the following:
  • the frequency domain resource length of the second signal is greater than that of the first signal
  • the frequency domain resource interval of the second signal is smaller than that of the first signal
  • the time-domain resource length of the second signal is greater than that of the first signal
  • the time-domain resource unit of the second signal is smaller than the time-domain resource interval of the first signal
  • the transmission power of the second signal is greater than the transmission power of the first signal
  • the first signal is a subset of the second signal.
  • the measurement configuration information includes at least one of the following:
  • the measurement quantity of the first measurement data includes at least one of the following:
  • Received signal channel information, spectrum information, and basic measurement quantities
  • the measurement quantity of the second measurement data includes at least one of the following:
  • Received signal channel information, spectrum information, and basic measurement quantities.
  • the format information of the first measurement data is used to indicate at least one of the following:
  • channel information The dimensions of channel information, the range of spectral information, the types of basic measurements, and the format of sensor data;
  • the format information of the second measurement data is used to indicate at least one of the following:
  • the method further includes:
  • the second device performs a target operation based on the target information, the target operation including at least one of the following:
  • the second device satisfies at least one of the following conditions:
  • the supported transmit power exceeds the preset threshold
  • the number of supported antennas exceeds a preset threshold
  • Supported array apertures exceed a preset threshold
  • the amount of resources available for sensing exceeds a preset threshold
  • the clock synchronization error between the device and the third device is less than a preset threshold, wherein the third device is a device used to provide the second measurement data.
  • this embodiment is an implementation of the second device corresponding to the embodiment shown in FIG6.
  • This embodiment mainly describes the monitoring of obtaining reference data based on non-AI parameter estimation.
  • the first AI unit is deployed on the first device side.
  • the first device is a sensing network function, a base station, or a terminal
  • the second device is a base station or a terminal.
  • the first device obtains first measurement data input to the first AI unit from the second device, and second measurement data used to calculate reference data through parameter estimation.
  • the specific process is shown in Figure 9.
  • Step 1 The first device (taking SensingMF as an example) acquires sensing requirement information.
  • the source of the sensing requirement information can be:
  • the sensing requirement comes from an external application.
  • the application function (AF) sends the sensing requirement information to the NEF, which then sends it to the AMF.
  • the AMF selects SensingMF and sends the sensing requirement information to SensingMF.
  • AF can directly send the sensing demand information to SensingMF;
  • Sensing demand information can also come from base stations and/or terminals.
  • the base station and/or terminal send the information to the AMF, and the AMF selects SensingMF and sends the sensing demand information to SensingMF.
  • the base station and/or terminal may directly send the sensing demand information to SensingMF;
  • Sensing demand information can also come from core network elements, which send sensing demands to the AMF.
  • AMF selects SensingMF and sends the perceived demand information to SensingMF
  • core network elements can directly send sensing requirement information to SensingMF.
  • the method of forwarding sensing requirement information through AMF may occur, but is not limited to, in scenarios where multiple sensing network elements are deployed in the network, and the AMF needs to select a suitable sensing network element from multiple sensing network elements based on information such as the location of the sensing object, the type of sensing service, or the sensing QoS requirement information; the method of forwarding sensing requirements without AMF may occur, but is not limited to, in scenarios where one or fewer SensingMFs are deployed in the network.
  • the first device may obtain sensing requirement information by receiving sensing requirement information sent by the sensing network function; or this step may be optional.
  • the perceived demand information includes at least one of the following:
  • sensing service or sensing service type is described in the corresponding description of the above implementation method, and will not be repeated here.
  • the target area for perception can refer to the area where the object being perceived may exist, or the area where imaging or environmental reconstruction is required.
  • the sensing object type can be a classification of sensing objects based on their possible motion characteristics.
  • Each sensing object type contains information such as the motion velocity, motion acceleration, and typical RCS of a typical sensing object.
  • Sensitive QoS can be a performance metric for sensing target areas or objects, including at least one of the following:
  • Perception resolution can be categorized into: ranging resolution, angle measurement resolution, velocity measurement resolution, and imaging resolution, etc.
  • Sensing accuracy can be categorized into: ranging accuracy, angle measurement accuracy, velocity measurement accuracy, positioning accuracy, etc.
  • the sensing range can be divided into: ranging range, velocity measuring range, angle measuring range, imaging range, etc.
  • Perception latency can be the time interval from the transmission of a sensing signal to the acquisition of a sensing result, or the time interval from the initiation of a sensing demand to the acquisition of a sensing result.
  • Perception update rate such as the time interval between two consecutive perception operations and obtaining perception results
  • Detection probability such as the probability of correctly detecting an object given its presence
  • Recognition probability (used in multivariate detection scenarios, representing the probability of correctly detecting a target state or category when the target is in a specific state or belongs to a specific category);
  • False alarm probability such as the probability of falsely detecting a target when the target does not exist
  • Step 2 The first device sends a first instruction message to the second device (specifically, it may be all or part of the target information in the above embodiments).
  • the first instruction message is used to instruct the second device to acquire and/or report measurement data, that is, to request measurement data for inference and monitoring of the first AI unit from the second device.
  • the first instruction message includes at least one of the following:
  • the signal configuration information includes at least one of the configuration information of a first signal and the configuration information of a second signal; the first signal is a signal used to acquire input data of the first AI unit, and the second signal is a signal used to acquire reference data.
  • first signal and the second signal are the same signal, or the first signal and the second signal are different signals, and the characteristics of the first signal and the second signal satisfy at least one of the following:
  • the frequency domain resource length (bandwidth) of the second signal is greater than that of the first signal
  • the frequency domain resource cell spacing of the second signal is smaller than that of the first signal
  • the time-domain resource length of the second signal is greater than that of the first signal
  • the time-domain resource unit interval (transmission period) of the second signal is smaller than the time-domain resource interval of the first signal
  • the transmission power of the second signal is greater than that of the first signal.
  • the measurement performance based on the second signal is better than that based on the first signal.
  • the first signal is a subset of the second signal, as shown in Figure 7.
  • the transmission period of the first signal is 1 slot, and the transmission period of the second signal is 0.5 slots.
  • the frequency domain density (frequency domain resource cell spacing) of the first signal is such that one out of every 12 subcarriers is used to carry the first signal, and the second signal occupies consecutive subcarriers in the frequency domain.
  • the above measurement configuration information includes at least one of the following:
  • the measurement includes the signal resource indication, the purpose of the measurement or the method of processing the measurement data, the measured quantity, the measurement quantity format, the reference data type associated with monitoring, and the reporting configuration.
  • the measured signal resource indicator can be the identifier of the first signal or the identifier of the second signal;
  • the indication of measurement purpose or measurement data processing method includes at least one of the following: the measurement is an indication for inference or monitoring of the AI unit; the measurement quantities for different purposes are preset, and the second device can determine the corresponding measurement quantity according to the indication; the indication of whether the measurement data is subsequently processed by AI or not; optionally, the measurement quantities or measurement quantity formats corresponding to different processing methods are preset, and the second device can determine the corresponding measurement quantities and measurement quantity formats according to the indication; in this embodiment, AI processing refers to measurement data used for inference of the AI unit, and non-AI processing refers to measurement data used for monitoring of the AI unit; the identification information of the AI unit into which the measurement data is input, i.e., different AI units correspond to different measurement quantities or measurement quantity formats, and the second device can determine the corresponding measurement quantities and measurement quantity formats according to the indication.
  • the measured quantity can be a measured quantity used by the first device as input to the first AI unit, or a measured quantity used to calculate reference data, including at least one of the following: received signal or raw channel information (including time-domain channel response and frequency-domain channel response), including at least one of the following: complex result of received signal or channel response, amplitude/phase, I-channel/Q-channel data; spectral information calculated based on the raw channel information, including at least one of the following: time delay (distance) spectrum, Doppler (velocity) spectrum, angle spectrum, time delay (distance)-Doppler (velocity) spectrum, time delay (distance)-Doppler (velocity)-angle spectrum, time delay (distance)-Doppler (velocity)-angle spectrum, time-Doppler spectrum (micro-Doppler spectrum); wherein, the spectral information can refer to complex result, for example, the time delay-Doppler spectrum refers to the time delay, Doppler index and corresponding complex value (including phase information)
  • the basic measurement quantities include at least one of the following: time delay, Doppler, angle, and intensity (power);
  • the basic measurement quantity can be a quantification result of the actual value or soft information.
  • the aforementioned measurement quantities can be the measurement quantities corresponding to AI inference and monitoring respectively, namely, the measurement quantities used by the first AI unit for inference (e.g., channel information or spectral information) and the measurement quantities used to calculate reference data based on non-AI parameter estimation (e.g., basic measurement quantities, specifically, such as the time delay value, Doppler value, or angle value corresponding to the path or sampling point in the spectral information that exceeds a preset threshold after obtaining spectral information based on channel information).
  • the measurement quantities used by the first AI unit for inference e.g., channel information or spectral information
  • non-AI parameter estimation e.g., basic measurement quantities, specifically, such as the time delay value, Doppler value, or angle value corresponding to the path or sampling point in the spectral information that exceeds a preset threshold after obtaining spectral information based on channel information.
  • the measurement quantities used to calculate the reference data can also be perception results, including at least one of the following: time delay, Doppler effect, angle, distance, velocity, orientation, spatial location, acceleration, target presence, number of targets, trajectory, gesture, action, expression, vital signs, quantity, imaging results, weather, air quality, shape, material, and composition of the perceived target.
  • the perception results are the reference data; that is, the calculation of the reference data based on the non-AI parameter estimation method is completed on the second device side and the reference data is directly sent to the first device.
  • the aforementioned measurement format can be the format of measurement data input to the first AI unit (i.e., the first measurement data), or the format of measurement data used to calculate reference data (i.e., the second measurement data), including at least one of the following:
  • the original channel information can be dimensional, such as the frequency domain channel response information on a single symbol (one-dimensional channel information), the time-frequency domain channel response information on multiple symbols (two-dimensional channel information), the time-frequency spatial domain channel response information of multiple symbols and multiple antennas (three-dimensional channel information), and the scale of different dimensions, namely the number of sampling points or the sampling interval (e.g., time-frequency domain density).
  • spectral information it can be a limitation on the range of different spectral information (or a cutoff window for different dimensions of spectral information), such as a subset of spectral information corresponding to a specific time delay or Doppler range in a time-delay-Doppler spectrum; it can also indicate the upper limit of the number of paths or the upper limit of the number of sampling points of a specific spectrum; or it can indicate the minimum granularity of a specific spectrum, that is, the interval between two adjacent sampling points (corresponding to the sensing resolution).
  • the aforementioned spectral information subset includes partial spectral information used for device positioning and partial spectral information used for sensing, such as information on the N2-N3 sampling points or the N2-N3 paths in the time delay spectrum, as shown in the boxed portion of Figure 10.
  • the aforementioned subset of spectral information is the portion of the time-delay-Doppler spectrum where the absolute Doppler value is less than X1 and the time delay value is less than X2, as shown in the boxed area in Figure 11.
  • the aforementioned measurement format can refer to the measurement format corresponding to the inference and monitoring of the first AI unit, for example, the two have different measurement quantities, and therefore the measurement format is also different; or it can refer to the measurement format shared by the inference and monitoring of the AI unit, for example, the two have the same measurement quantities and the same signal configuration (the first signal and the second signal are the same signal), and therefore the corresponding measurement format is also the same.
  • the reference data types associated with monitoring include at least one of the following:
  • Sensing data acquired through specific sensing modes such as monopolar sensing mode and base station self-transmitting and self-receiving sensing mode.
  • Sensing data acquired through specific sensing devices or specific types of sensing devices such as sensing data acquired by base stations or sensing data acquired by stationary devices;
  • Perceptual data is obtained through reasoning by more capable AI units.
  • the above-mentioned reporting configuration for the second device's measurement data reporting criteria includes at least one of the following:
  • the reporting cycle can be based on the coherent processing time of the sensed data, and the measurement results related to sensed perception and positioning can be reported.
  • the reporting trigger can be a pre-defined event, including but not limited to:
  • the second measurement data is reported for monitoring of the AI unit.
  • the event of entering a specific area e.g., a residential community
  • the items in the first instruction information mentioned above can be sent by the same signaling message or by different signaling messages.
  • Step 3 The second device executes the measurement process according to the first instruction information, that is, it receives the first signal and the second signal sent by other devices, and measures and obtains the first measurement data and the second measurement data.
  • the second device can send the first/second signal and receive the echo signal to perform the measurement.
  • the other device can be the first device, that is, the first device sends a first signal/second signal, and the second device receives the first signal/second signal and measures it.
  • Step 4 The second device sends measurement data to the first device, the measurement data including at least one of the following:
  • the measurement result i.e. the value of the measurement quantity required by the above measurement quantity format, includes the measurement quantity used for inference of the AI unit or the measurement quantity used for monitoring of the AI unit, or the measurement quantity used for both.
  • Explanatory information associated with the measurement results includes at least one of the following:
  • the timestamp information can be absolute time (UTC) or time relative to a certain reference time point, such as wireless frame number, half-frame number, subframe number, TTI sequence number, slot sequence number, sub-slot sequence number, symbol sequence number, etc.; wherein, the slot sequence number or symbol sequence number can also refer to the sequence number of the starting slot or symbol relative to the sensing coherent processing time.
  • UTC absolute time
  • the slot sequence number or symbol sequence number can also refer to the sequence number of the starting slot or symbol relative to the sensing coherent processing time.
  • the sensing mode identifier corresponding to the measurement result (including at least one of the six sensing modes in the background technology);
  • the identifier of the AI unit to which the measurement result is input is used to indicate which AI unit the measurement result is input to;
  • Performance metrics including but not limited to RSRP, RSRQ, SINR, and perception performance metrics, are considered valid when a perception performance metric exceeds a preset threshold.
  • Step 5 The first device receives measurement data and inputs the corresponding measurement results (measurement results used for reasoning by the first AI unit) into the corresponding AI unit to obtain perception data, which may be the first perception result.
  • the first device obtains reference data based on the corresponding measurement results (measurement results used for monitoring by the first AI unit) through a non-AI method (parameter estimation algorithm), which is the reference data obtained by the second device through a non-AI method, specifically the second perception result.
  • a non-AI method parameter estimation algorithm
  • the first device obtains monitoring result information based on the sensed data and reference data, including:
  • the perception accuracy information of the first AI unit is calculated, that is, the error information or error statistics of the perception data output by the first AI unit compared with the reference data.
  • the perception accuracy information includes at least one of the following:
  • Distance accuracy information speed accuracy information, angle accuracy information, positioning accuracy information, detection accuracy information, and recognition accuracy information.
  • the aforementioned monitoring results may also include at least one of the following:
  • the validity information of the first AI unit is determined based on the perception accuracy information. For example, if the error information of more than X perception results in N perception result calculations exceeds a preset threshold, the AI unit is considered to be ineffective. Or, for target positioning, if the root mean square error of the position coordinates exceeds a preset threshold when the number of perception result calculations reaches a preset value within a given time, the AI unit is considered to be ineffective.
  • the performance level information of the first AI unit is that different AI units have different preset performance levels. Different performance levels are associated with different perception accuracy levels. The preset performance level reached by the AI unit is determined based on the perception accuracy information.
  • This embodiment mainly describes how to obtain reference data based on non-AI parameter estimation.
  • the first AI unit is deployed on the second device side, which is a base station or a terminal, and the first device is a sensing network function, a base station, or a terminal.
  • the first device obtains first measurement data output by the first AI unit from the second device, and second measurement data used to calculate reference data through parameter estimation. The specific process is shown in Figure 12.
  • Step 1 The first device (taking SensingMF as an example) acquires sensing requirement information, specifically the same as in Implementation Example 1.
  • Step 2 The first device sends a second instruction message (which is all or part of the target information in the above embodiments) to the second device to instruct the second device to acquire measurement data and/or report measurement data, that is, to request the first measurement data output by the AI unit and the second measurement data for monitoring from the second device.
  • the second instruction message includes at least one of the following:
  • the signal configuration information is the same as in Implementation Example 1;
  • Measurement configuration information includes at least one of the following:
  • the measured signal resource indication such as the identifier of the first signal or the identifier of the second signal
  • the measurement quantity includes the measurement quantity associated with the first measurement data (perception data output by the first AI unit) and the measurement quantity associated with the second measurement data (for monitoring of the AI unit).
  • the measurement quantity associated with the first measurement data is the perception result (see Embodiment 1) or the basic measurement quantity; the measurement quantity associated with the second measurement data is the basic measurement quantity or spectral information, or the perception result.
  • the measurement format is the same as in Example 1;
  • Measurement method indication includes at least one of the following:
  • the identifier of the AI unit used indicates which AI unit the second device uses to calculate the perception result
  • the accuracy information, performance level, or confidence level of the AI unit used.
  • Step 3 The second device executes the measurement process (i.e., measures the first signal) according to the second instruction information to obtain a first intermediate measurement result, and further processes the first intermediate measurement result based on the first AI unit to obtain sensing data, specifically the value of the sensing result or basic measurement quantity.
  • the first intermediate measurement result is the input data of the first AI unit, including at least one of channel information, spectrum information, and basic measurement quantity, as described in Embodiment 1.
  • the second device executes the measurement process (i.e., measures the second signal) according to the second instruction information to obtain the second intermediate measurement result, i.e. the second measurement data, including the values of basic measurement quantities (time delay value, Doppler value, angle value, etc.) or spectral information.
  • the second device obtains the perception result, i.e., reference data, such as the position and number of targets, based on the second intermediate measurement result using a non-AI parameter estimation method.
  • the second device may obtain the basic measurement quantity based on the second intermediate measurement result using a non-AI parameter estimation method.
  • the first signal and the second signal can be the same signal, and the first intermediate measurement result and the second intermediate measurement result can be the same.
  • the second device obtains spectral information by measuring the same signal, and then inputs the spectral information into the AI unit for inference to obtain the perception result or the value of the basic measurement quantity, i.e., the first measurement data. Then, the spectral information is further processed by a non-AI parameter estimation method to obtain the value of the basic measurement quantity as the second measurement data, or the perception result is obtained by further calculating the value of the basic measurement quantity as the second measurement data.
  • Step 4 The second device reports the first measurement data and the second measurement data to the first device, including at least one of the following:
  • the measurement result i.e. the value of the measurement quantity required by the measurement quantity format, includes the measurement result obtained through inference by the AI unit and the measurement result used for monitoring by the AI unit;
  • Explanatory information related to the perception or localization results includes at least one of the following:
  • the identifier of the AI unit used indicates which AI unit's reasoning result the perception or localization result belongs to;
  • Step 5 The first device receives the first measurement data and obtains the perception data output by the first AI unit.
  • the first measurement data may include perception data (the first AI unit on the second device side outputs perception data), and the perception data is the first perception result; or, the first measurement data may include the value of a basic measurement quantity (the first AI unit on the second device side outputs the value of the basic measurement quantity), and the first device calculates the first perception result based on the value of the basic measurement quantity.
  • the first device receives the second measurement data and obtains reference data, which may specifically be a second perception result.
  • reference data which may specifically be a second perception result.
  • the first device obtains the second perception result based on the values of the basic measurements using a non-AI method (parameter estimation algorithm).
  • the perception result included in the second measurement data is the second perception result obtained by the second device using a non-AI method.
  • the first device obtains monitoring result information based on the first and second sensing results, as in Embodiment 1.
  • Step 6 The first device sends the monitoring result information to the second device, or sends subsequent operation instructions for the first AI unit, such as deactivation, switching, fine-tuning, retraining, etc.
  • This embodiment mainly describes the acquisition of reference data based on sensors.
  • the first AI unit is deployed on the second device side or the first device side.
  • the specific process can be referred to in Embodiment 1; when the first AI unit is deployed on the second device side, the specific process can be referred to in Embodiment 2.
  • the first device sends a third instruction message (which may be all or part of the target information in the above embodiments) to the second device to instruct the second device to acquire measurement data and/or report measurement data, that is, to request the second device the first measurement data input (or output) by the first AI unit and the second measurement data for monitoring by the first AI unit.
  • the third instruction message includes at least one of the following:
  • the signal configuration information includes the configuration information of a first signal, which is a signal used to acquire input data from the AI unit.
  • Measurement configuration information includes at least one of the following:
  • the measured signal resource indication such as the first signal identifier
  • the measurement quantity is a measurement quantity associated with the inference of the first AI unit, including the measurement quantity corresponding to the measurement data input by the first AI unit (when the first AI unit is on the first device side), or the measurement quantity corresponding to the measurement data output by the first AI unit (when the first AI unit is on the second device side).
  • the sensor type indication indicates which sensor the second device uses to acquire the perception data for monitoring the first AI unit.
  • the sensor type includes at least one of the following: visible light camera, infrared camera, GNSS, lidar, millimeter-wave radar, thermometer, hygrometer, barometer, gyroscope, accelerometer, magnetometer, gravity sensor, sonar, and rain gauge.
  • the measurement format refers to the format of the measurement data input by the first AI unit (when the first AI unit is on the first device side) or the format of the measurement data output by the first AI unit (when the first AI unit is on the second device side).
  • This also includes the sensor's data format, such as binary or ASCII formats, or data formats associated with a specific sensor type, such as YUV or RGB for image sensors (cameras).
  • a specific sensor type such as YUV or RGB for image sensors (cameras).
  • Accelerometer Measures the acceleration applied to a device, in a format including acceleration along the x-axis, y-axis, and z-axis. Further, it can be categorized into results that include gravitational acceleration, results that do not include gravitational acceleration, results that include bias compensation, and results that do not include bias compensation.
  • Gyroscope measures the rate of rotation (radians per second) around the x, y, and z axes of a device. Similar to an accelerometer, it can be represented as a three-dimensional vector.
  • Magnetometer Monitors changes in the Earth's magnetic field, measuring the intensity of the geomagnetic field along each of the three coordinate axes (in microtesla), in a format shown. Typically, this sensor is not used directly but is combined with other sensors to obtain rotation angle information.
  • Rotation vector sensor can be obtained by combining different sensors to acquire the terminal angle, which is in the format of (the angle of rotation of the terminal around the x, y, and z axes, or the rotation angle relative to the East-North-Up (northeast sky)/North-East-Down (northeast earth) coordinate axes).
  • the measurement method is indicated in the same way as in Embodiment 1 (when the AI unit is on the first device side) or in the same way as in Embodiment 2 (when the AI unit is on the second device side).
  • the second device executes the measurement procedure according to the third instruction information and reports measurement data to the first device, wherein the measurement data includes at least one of the following:
  • the measurement result i.e. the value of the measurement quantity required by the measurement quantity format, includes the measurement result obtained through inference by the AI unit and the measurement result used for monitoring by the AI unit, i.e. the perception data from the sensor output.
  • Explanatory information related to the measurement results includes at least one of the following:
  • the identifier of the reference data type associated with monitoring i.e., an indication that the reference data comes from the sensor
  • the sensor outputs accuracy information
  • Coordinate system relationships the relationship between the sensor's corresponding coordinate system and the global coordinate system, such as the transformation parameters from the local coordinate system to the global coordinate system, that is, the rotation angles of the local coordinate system relative to the global coordinate system: ⁇ (bearing angle), ⁇ (downward tilt angle) and ⁇ (tilt angle).
  • Embodiment 1 when the first AI unit is on the first device side
  • Embodiment 2 when the first AI unit is on the second device side
  • This embodiment mainly describes monitoring based on specific devices or specific sensing modes to obtain reference data.
  • the first AI unit is deployed on the second device side or the first device side.
  • the specific process can be referred to in Embodiment 1; when the first AI unit is deployed on the second device side, the specific process can be referred to in Embodiment 2.
  • the first device In addition to requesting the first measurement data input (or output) of the first AI unit from the second device, the first device also requests the second measurement data for monitoring the first AI unit from the third device, wherein the third device is a device that meets specific conditions or supports acquiring perception data through a specific perception mode (including at least one of the six perception modes in the background art).
  • the specific conditions include at least one of the following:
  • the third device is a stationary device
  • the third device is a base station
  • the third device's supported transmit power, number of supported antennas or array aperture, and available resources such as signal bandwidth for sensing signals exceed a preset threshold.
  • the clock synchronization error (including timing synchronization or frequency synchronization) between the third device and the second device is less than a preset threshold.
  • the third device and the first device can be the same device.
  • the first device acquires measurement data for inference of the first AI unit based on the perception mode of transmitting and receiving perception signals with other devices, and acquires measurement data for monitoring the first AI unit based on the perception mode of spontaneously receiving perception signals.
  • the third device and the second device can be the same device.
  • the first device receives perception signals sent by the second device and performs measurements to acquire measurement data for inference of the first AI unit, and sends it to the second device.
  • the second device acquires measurement data for monitoring the first AI unit based on the perception mode of spontaneously receiving perception signals.
  • This embodiment mainly describes the sensing performance indicators.
  • the aforementioned perceived performance metrics may include at least one of the following:
  • Sensing metrics related to received power, as well as interference or noise power are related to received power, as well as interference or noise power.
  • the aforementioned sensing indicators related to received power may include: a first indicator, which is used to indicate the received power of the sensing target associated path.
  • the first indicator mentioned above may be the linear average value (in W) of the received power of the path associated with the sensed target in the channel response measured from the target signal over the resource unit carrying the target signal.
  • This resource unit may be a time-domain or frequency-domain resource unit. Using a linear average value makes the received power more accurate and reliable. It should be noted that the embodiments of this application do not limit the received power to a linear average value. For example, in some embodiments, it may also be the median received power, the lowest received power, or the highest received power.
  • the target signal mentioned above can be the first signal or the second signal in the above embodiments, such as a dedicated signal for sensing services, or a communication signal such as a reference signal or a synchronization signal.
  • the aforementioned perception metrics related to interference or noise power include at least one of the following:
  • the second indicator is the sum of a first linear average and a second linear average.
  • the first linear average is the linear average of the power of paths other than the path associated with the sensing target in the channel response of the target signal on the target resource.
  • the second linear average is the linear average of the interference or noise power from signals other than the target signal on the first resource.
  • the second indicator is equal to the difference between the total received power and the first indicator, where the total received power is the total received power of the first device on the target resource.
  • the third indicator is the linear average of the interference or noise power from signals other than the target signal on the second resource, or the third indicator is equal to the difference between the total received power and the received power of the target signal, where the total received power is the total received power of the first device on the target resource.
  • the fourth indicator is the linear average power of the paths other than the path associated with the sensing target in the channel response of the target signal on the target resource; or, the fourth indicator is equal to the difference between the received power of the target signal and the first indicator mentioned above.
  • the first indicator is used to indicate the received power of the target signal along the path associated with the sensing target
  • the target resource is the transmission resource of the target signal
  • the first resource includes the target resource or at least one resource other than the target resource
  • the second resource includes the target resource or at least one resource other than the target resource.
  • the other paths mentioned above can be all or part of the paths in the target signal other than those associated with the perceived target.
  • Other signals besides the target signal mentioned above can refer to all or part of the signals detected by the first device on the first resource, excluding the target signal.
  • the aforementioned first resource including the target resource or at least one resource other than the target resource, means that the first resource includes at least one of the following:
  • the target resource and at least one other resource besides the target resource.
  • the aforementioned second resource including the target resource or at least one resource other than the target resource, means that the second resource includes at least one of the following:
  • the target resource and at least one other resource besides the target resource.
  • At least one resource other than the target resource can refer to at least one resource other than the target resource among the resources that the first device needs to detect or receive signals from, such as resources configured by higher-layer signaling or resources that the first device has predetermined to detect or receive signals from.
  • the aforementioned interference or noise power includes the sum of interference power and noise power, or interference power or noise power.
  • the total received power of the first device on the target resource may include the received power of signals from the serving cell and non-serving cells on the target resource, adjacent channel interference power, and thermal noise power, etc. Furthermore, the total received power may also be a linear average of the total received power of the first device on the target resource (in W).
  • the received power of the target signal mentioned above refers to the reference signal received power (RSRP) of the target signal.
  • RSRP reference signal received power
  • Second indicator Total received power - First indicator.
  • the aforementioned sensing metrics related to received power, and also related to interference or noise power include at least one of the following:
  • the fifth indicator is equal to the quotient obtained by dividing the first indicator by the second indicator.
  • the sixth indicator is equal to the quotient obtained by dividing the first indicator by the third indicator.
  • the seventh indicator is equal to the quotient obtained by dividing the first indicator by the fourth indicator.
  • the eighth index is equal to the product of the quotient obtained by dividing the first index by the total received power and the target coefficient
  • the total received power is the total received power of the first device on the target resource.
  • the first, second, third, and fourth indicators mentioned above are the same as those described in the above implementation method, and will not be repeated here. It should be noted that, when at least one of the fifth, sixth, seventh, and eighth indicators is included, the perception-related indicators in the embodiments of this application may or may not include the first, second, third, and fourth indicators mentioned above.
  • the receiving power and interference or noise can be taken into account when determining the measurement switching, so as to make the measurement switching more reliable.
  • the aforementioned sensing metrics related to received power, and also related to interference or noise power may include at least one of the following:
  • Indicators related to perceived SINR perceived SNR, perceived signal interference ratio (SIR), and perceived RSRQ.
  • the path associated with the perceived target satisfies at least one of the following:
  • the parameter meets the first preset threshold, or the parameter is within the first preset range.
  • the parameters meet the preset modulation rules
  • the parameter difference with the first arrival path meets the second preset threshold, or the parameter difference with the first arrival path is within the second preset range;
  • the parameter difference with the reference path meets the third preset threshold, or the parameter difference with the reference path is within the third preset range.
  • the above parameters may include at least one of the following:
  • the above parameter difference may include at least one of the following:
  • Amplitude difference power difference, intensity difference, energy difference, phase difference, Doppler difference, time delay difference, and angle difference.
  • the first preset threshold, the first preset interval range, the second preset threshold, the second preset interval range, the third preset threshold, and the third preset interval range can be agreed upon by the protocol or configured on the network side. Alternatively, these preset thresholds or preset interval ranges can be determined by the receiving device based on prior sensing information or sensing requirements.
  • the parameters satisfying the first preset threshold can be defined as the parameters exceeding or equaling the first preset threshold.
  • the parameter difference with the first path satisfying the second preset threshold can be defined as the parameter difference with the first path exceeding or equaling the second preset threshold.
  • the parameter difference with the reference path satisfying the third preset threshold can be defined as the parameter difference with the reference path exceeding or equaling the third preset threshold.
  • the sensing service is moving target detection
  • the path with a Doppler greater than zero needs to be detected as the path associated with the sensing target
  • the sensing target is a vehicle, with a default vehicle speed of 40km/h to 120km/h
  • the path within the corresponding speed range needs to be detected as the path associated with the sensing target
  • the distance between the sensing target area and the sensing signal transceiver needs to meet specific requirements
  • the path within the corresponding time delay range needs to be detected as the path associated with the sensing target
  • the sensing service is respiratory monitoring
  • the normal breathing rate can be determined based on the person's gender and age (e.g., 15 to 30 breaths/minute, which can be used as prior information for sensing, and the corresponding Doppler range of 0.25 to 0.5Hz can be calculated).
  • the aforementioned first-arrival path can be a line-of-sight (LOS) path, specifically the path from which the target signal first reaches the receiver.
  • the aforementioned reference path can be a path reflected by a known target, such as a path reflected by a reconfigurable intelligence surface (RIS), backscatter, or other known passive targets.
  • RIS reconfigurable intelligence surface
  • the aforementioned preset modulation rules can be agreed upon by the protocol or configured on the network side.
  • Specific modulation rules are the modulation rules of tags, backscatter devices, or RIS, that is, the path associated with the sensed target can be a path that has been modulated and reflected by tags, backscatter devices, or RIS.
  • the path associated with the perceived target can be determined in multiple ways, which can improve the flexibility of determining the path associated with the perceived target, and can also improve the accuracy of determining the path associated with the perceived target by combining multiple methods.
  • a set of paths can be determined, including paths whose amplitude, power, intensity, or energy exceeds a certain threshold, as shown in Figure 13.
  • This set of paths includes paths 0, 1, 2, and 3.
  • the path associated with the sensed target is then determined from this set based on at least one of the aforementioned criteria, thereby reducing computational complexity.
  • the first indicator is calculated in the following way:
  • the first device transforms it to the first dimension and determines the path associated with the perceived target in the first dimension. Then, it calculates the power of the path associated with the perceived target as a first indicator. If the path associated with the perceived target includes multiple paths, the sum of the powers of the multiple paths is calculated as the first indicator.
  • the first dimension includes one of the following:
  • the first dimension is the delay-Doppler dimension.
  • f 0,1,2,...,N-1 represents the frequency domain sampling points (e.g., subcarrier index)
  • s 0,1,2,...,P-1 represents the spatial domain sampling points (antenna index or port index).
  • sensing path Method for determining the path associated with the sensed target (referred to as the sensing path) in the channel response obtained from the measurement of the target signal:
  • the paths in the path set include those whose amplitude, power, intensity, or energy exceeds a certain threshold after the channel response is transformed to the first dimension.
  • a certain threshold can be set to be higher than a noise threshold or a noise interference threshold, or as agreed upon by the protocol.
  • This step (determining the path set) is optional; it can be done solely based on the next step to determine the paths associated with the sensing target.
  • the path that satisfies the first condition is selected from the set of paths or from all paths of the target signal, and is used as the path associated with the perceived target.
  • the first condition includes at least one of the following:
  • the amplitude, power, intensity, or energy of the noise exceeds a preset threshold or falls within a preset range, such as a preset threshold that exceeds 5 times the noise threshold.
  • the Doppler amplitude of the path exceeds the preset threshold or falls within the preset range;
  • the path delay exceeds a preset threshold or falls within a preset range
  • the angle of the radius exceeds the preset threshold or falls within the preset range
  • the difference in amplitude/power/intensity/energy between the path and the first-reach path (e.g., the LOS path) or the reference path exceeds a preset threshold or is within a preset range.
  • the reference path can be a path reflected by a known target (e.g., RIS/Backscatter/other known passive targets, etc.).
  • the time delay difference between the path and the first path (e.g., the LOS path) or the reference path exceeds a preset threshold or falls within a preset range;
  • the angle difference between the diameter and the first-arrival diameter (e.g., the LOS diameter) or the reference diameter exceeds a preset threshold or falls within a preset range.
  • the amplitude, power, intensity, energy, or phase of the path satisfies a specific modulation rule, which is the modulation rule of the Tag/Backscatter device or RIS. That is, the path associated with the sensed target can be a path modulated and reflected by the Tag/Backscatter device or RIS.
  • the first condition of each of the above can also be based on the statistical results over a period of time; for example, the proportion of the above indicators (such as Doppler of the path, delay of the path, etc.) exceeding the preset threshold or falling within the preset range within the preset time window reaches the preset proportion, or the number of times the above indicators (such as Doppler of the path, delay of the path, etc.) exceed the preset threshold or fall within the preset range within the preset time window reaches the preset number.
  • the proportion of the above indicators such as Doppler of the path, delay of the path, etc.
  • the preset threshold or set range is sent to the receiving device by other devices, and determined by those devices based on prior sensing information or sensing requirements.
  • the preset threshold or preset range can be agreed upon by a protocol, or it can be determined by the receiving device based on prior sensing information or sensing requirements.
  • prior information for perception or perception needs includes the following information:
  • the sensing services or sensing service types may include, for example, detecting the presence of a target, positioning, speed detection, distance detection, angle detection, acceleration detection, material analysis, composition analysis, shape detection, category classification, radar cross section (RCS) detection, polarization scattering characteristic detection, fall detection, intrusion detection, quantity counting, indoor positioning, gesture recognition, lip reading, gait recognition, facial expression recognition, face recognition, respiration monitoring, heart rate monitoring, pulse monitoring, humidity/brightness/temperature/atmospheric pressure monitoring, air quality monitoring, weather condition monitoring, environmental reconstruction, and terrain and landform analysis.
  • the sensing services include building/vegetation distribution detection, pedestrian or vehicle flow detection, crowd density and vehicle density detection, etc.
  • the sensing service types can be classified according to certain characteristics, such as by function (detection-type sensing services, including intrusion detection and fall detection), parameter estimation-type sensing services (distance, angle, and speed calculation), and recognition-type sensing services (action recognition and identity recognition). They can also be classified by sensing range (near-range, medium-range, and long-range sensing), by sensing precision (coarse-grained sensing and fine-grained sensing), by power consumption/energy consumption, and by resource usage.
  • function detection-type sensing services, including intrusion detection and fall detection
  • parameter estimation-type sensing services distance, angle, and speed calculation
  • recognition-type sensing services action recognition and identity recognition
  • sensing range near-range, medium-range, and long-range sensing
  • sensing precision coarse-grained sensing and fine-grained sensing
  • power consumption/energy consumption and by resource usage.
  • the corresponding normal breathing rate can be determined based on the person's gender and age (e.g., males: 13-21 breaths/minute, females: 15-20 breaths/minute; adults: 12-20 breaths/minute, children: approximately 30-40 breaths/minute), which can be used as prior information for sensing.
  • Target area refers to the location area of the perceived object, or the location area that needs to be imaged or reconstructed; for example, a preset range of time delay for determining the path associated with the perceived target based on the approximate location/distance of the perceived object.
  • Sensing object type Sensing objects are classified according to their possible motion characteristics. Each sensing object type contains information such as the typical motion velocity range, motion acceleration range, and typical RCS range of the sensing object.
  • the number of perceived targets for example, the number of perceived targets can be obtained from the camera's perception results as a priori information.
  • paths 0, 1, 2, and 3 are paths in the path set, where paths 2 and 3 are sensing paths that satisfy the first condition (e.g., their time delay meets a preset threshold), and paths 0 and 1 are paths associated with other scatterers.
  • Figure 13 shows a multipath diagram of the channel response in the first dimension (time delay dimension, Doppler dimension, azimuth dimension, or elevation dimension), where the horizontal axis represents the first dimension and the vertical axis represents the normalized amplitude, power, intensity, or energy.
  • the reference point for the first indicator can be the antenna connector of the receiving device, such as the terminal.
  • the first indicator measured and reported by the receiving device cannot be lower than the indicator of any single receiving channel.
  • the first indicator measured for a certain receiving channel needs to be obtained by measuring the combined signal on multiple antenna elements corresponding to that receiving channel.
  • the second method for calculating the first indicator can be as follows:
  • the received power of the path associated with the perceived target it can also be the power of the path associated with the perceived target in the first dimension and...
  • the difference is used as the first indicator, where N1 represents the number of paths associated with the perceived target.
  • N1 represents the number of paths associated with the perceived target.
  • Method 1 for calculating the received power of the target signal can be as follows:
  • the received power of the target signal can be obtained by the receiving device after obtaining the channel response H(k), transforming it to the first dimension, determining the path set in the first dimension, and then calculating the sum of the power of all paths in the path set.
  • Method 2 for calculating the received power of the target signal can be as follows:
  • the received power of the target signal can also be the sum of the powers of all paths in the first-dimensional path set.
  • Y(k) is the received signal corresponding to the target signal
  • k 0, 1, 2, ..., K-1 represents the resource unit index
  • K is the number of resource units.
  • the second indicator can be calculated as follows:
  • the first filtering process is used to eliminate noise and interference in the first dimension, as well as paths associated with non-perceived targets. For example, the first filtering process sets the amplitude, power, intensity, or energy of other paths in Figure 13, excluding those associated with perceived targets, to zero.
  • the channel response H filter1 (k) after the first filtering process does not contain noise and interference, nor paths associated with non-perceived targets, but only paths associated with perceived targets.
  • the third indicator can be calculated in the following way:
  • the second filtering process described above can be noise interference suppression processing on the first dimension (e.g., setting the amplitude, power, intensity, or energy of other paths besides the path set to zero in Figure 6), or minimum mean squared error (MMSE) filtering.
  • the channel response H filter2 (k) after the second filtering process does not contain noise and interference, but only the paths in the path set.
  • the third indicator can be calculated in the following way:
  • the third index P ⁇ 2 is calculated, that is Where N represents the number of sampling points in the first dimension.
  • the receiving device identifies multiple sensing targets, or if the receiving device obtains the number of sensing targets based on prior sensing information or sensing requirements, the following methods are available:
  • Method 1 Calculate the perception-related indicators (also called target indicators) for each sensing target separately. For example, in Figure 4, determine the paths associated with each sensing target, and then calculate the perception-related indicators for each sensing target.
  • Method 2 Calculate a perception-related index for multiple perception targets. For example, in Figure 6, determine the paths associated with any perception target, and then define these paths as paths associated with the perception target; this is equivalent to treating multiple perception targets as a virtual perception target, and then calculating the perception-related index corresponding to this virtual perception target.
  • the AI unit monitoring method provided in this application can be executed by an AI unit monitoring device.
  • This application uses an AI unit monitoring device executing the AI unit monitoring method as an example to illustrate the AI unit monitoring device provided in this application.
  • the AI unit monitoring device can be a communication device or a component within a communication device, such as a chip.
  • the communication device can be a terminal, a network-side device, or a server, etc.
  • the terminal can be, but is not limited to, the type of terminal 11 listed above
  • the network-side device can be, but is not limited to, the type of network-side device 12 listed above. This application does not impose specific limitations.
  • the AI unit monitoring device may include a receiving module, a transmitting module, and a processing module. These modules can be implemented in software or hardware.
  • the processing module can be implemented by a processor.
  • the processor may include a general-purpose processor, a special-purpose processor, such as a Central Processing Unit (CPU), a microprocessor, a Digital Signal Processor (DSP), an Artificial Intelligence (AI) processor, a Graphics Processing Unit (GPU), an Application Specific Integrated Circuit (ASIC), a Network Processor (NP), a Field Programmable Gate Array (FPGA), or other programmable logic devices, gate circuits, transistors, discrete hardware components, etc.
  • the receiving and transmitting modules may be implemented by a communication interface, which may include one or more of the following: a transceiver, pins, circuits, a bus, and a radio frequency unit.
  • the AI unit monitoring device 1400 when the AI unit monitoring device is a terminal or network-side device, or a component in a terminal or network-side device, the AI unit monitoring device 1400 includes:
  • Processing module 1401 is used to acquire the perception data output by the first AI unit
  • the processing module is also used to acquire reference data for monitoring the first AI unit
  • the processing module is further configured to determine the monitoring result of the first AI unit based on the reference data and the perception data output by the first AI unit.
  • the reference data includes at least one of the following:
  • the perception data output by the second AI unit has a processing capability that is stronger than that of the first AI unit.
  • the perception data acquired through non-AI methods includes at least one of the following:
  • Perception data acquired through target perception devices acquired through target perception devices.
  • acquiring the perception data output by the first AI unit includes:
  • the first measurement data is used as input to the first AI unit for inference to obtain the perception data output by the first AI unit, and the first measurement data includes the first measurement result;
  • the first measurement data includes the perception data output by the first AI unit.
  • the first measurement data further includes: first explanatory information, which includes at least one of the following:
  • the first measurement data includes timestamp information, perception mode information, indication information for indicating that the first measurement result is used for AI unit inference, information of the AI unit corresponding to the first measurement data, performance index information of the first measurement data, indication information for indicating that the first measurement data includes perception data output by the first AI unit, reference data type information associated with the first measurement data, accuracy information corresponding to the first measurement data, performance level information corresponding to the first measurement data, and confidence level information corresponding to the first measurement data.
  • obtaining reference data for monitoring the first AI unit includes:
  • the second measurement data is used to determine the reference data, and the second measurement data includes the second measurement result; or,
  • the second measurement data includes the reference data.
  • the second measurement data further includes second explanatory information, which includes at least one of the following:
  • the second measurement data includes timestamp information, perception mode information, indication information for monitoring the second measurement result, reference data type information associated with the second measurement data, performance index information of the second measurement data, indication information for indicating that the second measurement data includes the reference data, information of the AI unit corresponding to the second measurement data, accuracy information corresponding to the second measurement data, coordinate system information of the second measurement data, and sensor type information corresponding to the second measurement data.
  • the monitoring results include at least one of the following:
  • Effectiveness information for performance level information, and perception accuracy information
  • the perception accuracy information includes at least one of the following:
  • Distance accuracy information speed accuracy information, angle accuracy information, positioning accuracy information, detection accuracy information, and recognition accuracy information.
  • the device further includes:
  • a transmitting module is configured to transmit target information, the target information including at least one of the following: signal configuration information, measurement configuration information, and indication information; wherein the indication information is used to indicate at least one of the following:
  • the first measurement data is used to be input into the first AI unit for reasoning to obtain the perception data output by the first AI unit, and the first measurement data includes the first measurement result;
  • the second measurement data is used to determine the reference data, and the second measurement data includes the second measurement result.
  • the signal configuration information includes at least one of the following:
  • the configuration information of the first signal and the configuration information of the second signal wherein the first signal is a signal used to acquire input data of the first AI unit, and the second signal is a signal used to acquire the reference data, and the first signal and the second signal satisfy at least one of the following:
  • the frequency domain resource length of the second signal is greater than that of the first signal
  • the frequency domain resource spacing of the second signal is smaller than that of the first signal
  • the time-domain resource length of the second signal is greater than that of the first signal
  • the time-domain resource unit of the second signal is smaller than the time-domain resource interval of the first signal
  • the transmission power of the second signal is greater than the transmission power of the first signal
  • the first signal is a subset of the second signal.
  • the measurement configuration information includes at least one of the following:
  • the measurement quantity of the first measurement data includes at least one of the following:
  • Received signal channel information, spectrum information, and basic measurement quantities
  • the measurement quantity of the second measurement data includes at least one of the following:
  • Received signal channel information, spectrum information, and basic measurement quantities.
  • the format information of the first measurement data is used to indicate at least one of the following:
  • channel information The dimensions of channel information, the range of spectral information, the types of basic measurements, and the format of sensor data;
  • the format information of the second measurement data is used to indicate at least one of the following:
  • the first device sends target information including at least one of the following:
  • the first device sends all or part of the target information to the second device;
  • the first device sends all or part of the target information to the third device;
  • the second device is a device for providing at least one of the first measurement data and the second measurement data
  • the third device is a device for providing the second measurement data.
  • the second device satisfies at least one of the following conditions:
  • the supported transmit power exceeds the preset threshold
  • the number of supported antennas exceeds a preset threshold
  • Supported array aperture exceeds preset threshold
  • the amount of resources available for sensing exceeds a preset threshold
  • the clock synchronization error between the device and the third device is less than a preset threshold
  • the third device satisfies at least one of the following conditions:
  • the supported transmit power exceeds the preset threshold
  • the number of supported antennas exceeds a preset threshold
  • Supported array aperture exceeds preset threshold
  • the amount of resources available for sensing exceeds a preset threshold
  • the clock synchronization error between the device and the second device is less than a preset threshold.
  • the processing module 1403 is further configured to determine a target operation for the first AI unit based on the monitoring results, the target operation including at least one of the following:
  • the aforementioned AI unit monitoring device helps to improve or ensure perception performance.
  • the signal monitoring device provided in this application embodiment can implement the various processes implemented in the method embodiment of FIG6 and achieve the same technical effect. To avoid repetition, it will not be described again here.
  • the AI unit monitoring device 1500 when the AI unit monitoring device is a terminal or network-side device, or a component in a terminal or network-side device, the AI unit monitoring device 1500 includes:
  • the sending module 1501 is used to send data, wherein the sent data includes at least one of the following:
  • the first measurement data is used to be input into the first AI unit for reasoning to obtain the perception data output by the first AI unit, and the first measurement data includes the first measurement result; or, the first measurement data includes the perception data output by the first AI unit.
  • the second measurement data is used to determine a reference signal, and the second measurement data includes a second measurement result; or, the second measurement data includes reference data used to monitor the first AI unit.
  • the reference data includes at least one of the following:
  • the perception data output by the second AI unit has a processing capability that is stronger than that of the first AI unit.
  • the perception data acquired through non-AI methods includes at least one of the following:
  • Perception data acquired through target perception devices acquired through target perception devices.
  • the first measurement data further includes: first explanatory information, which includes at least one of the following:
  • the first measurement data includes timestamp information, perception mode information, indication information for indicating that the first measurement result is used for AI unit inference, information of the AI unit corresponding to the first measurement data, performance index information of the first measurement data, indication information for indicating that the first measurement data includes perception data output by the first AI unit, reference data type information associated with the first measurement data, accuracy information corresponding to the first measurement data, performance level information corresponding to the first measurement data, and confidence level information corresponding to the first measurement data.
  • the second measurement data further includes second explanatory information, which includes at least one of the following:
  • the second measurement data includes timestamp information, perception mode information, indication information for monitoring the second measurement result, reference data type information associated with the second measurement data, performance index information of the second measurement data, indication information for indicating that the second measurement data includes the reference data, information of the AI unit corresponding to the second measurement data, accuracy information corresponding to the second measurement data, coordinate system information of the second measurement data, and sensor type information corresponding to the second measurement data.
  • the device further includes:
  • a receiving module is configured to receive target information, the target information including at least one of the following: signal configuration information, measurement configuration information, and indication information; wherein the indication information is used to indicate at least one of the following:
  • the signal configuration information includes at least one of the following:
  • the configuration information of the first signal and the configuration information of the second signal wherein the first signal is a signal used to acquire input data of the first AI unit, and the second signal is a signal used to acquire the reference data, and the first signal and the second signal satisfy at least one of the following:
  • the frequency domain resource length of the second signal is greater than that of the first signal
  • the frequency domain resource interval of the second signal is smaller than that of the first signal
  • the time-domain resource length of the second signal is greater than that of the first signal
  • the time-domain resource unit of the second signal is smaller than the time-domain resource interval of the first signal
  • the transmission power of the second signal is greater than the transmission power of the first signal
  • the first signal is a subset of the second signal.
  • the measurement configuration information includes at least one of the following:
  • the measurement quantity of the first measurement data includes at least one of the following:
  • Received signal channel information, spectrum information, and basic measurement quantities
  • the measurement quantity of the second measurement data includes at least one of the following:
  • Received signal channel information, spectrum information, and basic measurement quantities.
  • the format information of the first measurement data is used to indicate at least one of the following:
  • channel information The dimensions of channel information, the range of spectral information, the types of basic measurements, and the format of sensor data;
  • the format information of the second measurement data is used to indicate at least one of the following:
  • the device further includes:
  • the processing module is configured to perform a target operation based on the target information, wherein the target operation includes at least one of the following:
  • the second device satisfies at least one of the following conditions:
  • the supported transmit power exceeds the preset threshold
  • the number of supported antennas exceeds a preset threshold
  • Supported array aperture exceeds preset threshold
  • the amount of resources available for sensing exceeds a preset threshold
  • the clock synchronization error between the device and the third device is less than a preset threshold, wherein the third device is a device used to provide the second measurement data.
  • the aforementioned AI unit monitoring device helps to improve or ensure perception performance.
  • the signal monitoring device provided in this application embodiment can implement the various processes implemented in the method embodiment of FIG8 and achieve the same technical effect. To avoid repetition, it will not be described again here.
  • this application embodiment also provides a communication device 1600, including a processor 1601 and a memory 1602.
  • the memory 1602 stores a program or instructions that can run on the processor 1601.
  • the communication device 1600 is a first device
  • the program or instructions are executed by the processor 1601
  • the communication device 1600 is a second device
  • the program or instructions are executed by the processor 1601
  • they implement the various steps of the AI unit monitoring method embodiment on the second device side described above, and achieve the same technical effect. To avoid repetition, this will not be repeated here.
  • This application also provides a device, which is a first device, including a processor and a communication interface.
  • the communication interface is coupled to the processor, and the processor is used to run programs or instructions to implement the steps in the method embodiment shown in FIG6.
  • This device embodiment corresponds to the above-described first device-side method embodiment. All implementation processes and methods of the above-described method embodiments can be applied to this terminal embodiment and achieve the same technical effect.
  • This device may be the AI unit monitoring device shown in FIG14.
  • This application embodiment also provides a device, including a processor and a communication interface, wherein the processor or the communication interface is used to acquire perception data output by a first AI unit; acquire reference data for monitoring the first AI unit; and the processor is used to determine the monitoring result of the first AI unit based on the reference data and the perception data output by the first AI unit.
  • Figure 17 is a schematic diagram of the hardware structure of a device that implements an embodiment of this application.
  • the device 1700 includes, but is not limited to, at least some of the following components: radio frequency unit 1701, network module 1702, audio output unit 1703, input unit 1704, sensor 1705, display unit 1706, user input unit 1707, interface unit 1708, memory 1709, and processor 1710.
  • device 1700 may also include a power supply (such as a battery) for powering various components.
  • the power supply may be logically connected to processor 1710 through a power management system, thereby enabling functions such as managing charging, discharging, and power consumption through the power management system.
  • the device structure shown in Figure 17 does not constitute a limitation on the device.
  • the device may include more or fewer components than shown, or combine certain components, or have different component arrangements, which will not be elaborated here.
  • the input unit 1704 may include a graphics processor 17041 and a microphone 17042.
  • the graphics processor 17041 processes image data of still images or videos obtained by an image capture device (such as a camera) in video capture mode or image capture mode.
  • the display unit 1706 may include a display panel 17061, which may be configured in the form of a liquid crystal display, an organic light-emitting diode, or the like.
  • the user input unit 1707 includes at least one of a touch panel 17071 and other input devices 17072.
  • the touch panel 17071 is also called a touch screen.
  • the touch panel 17071 may include a touch detection device and a touch controller.
  • Other input devices 17072 may include, but are not limited to, physical keyboards, function keys (such as volume control buttons, power buttons, etc.), trackballs, mice, and joysticks, which will not be described in detail here.
  • the radio frequency unit 1701 can transmit it to the processor 1710 for processing; in addition, the radio frequency unit 1701 can send uplink data to the network-side device.
  • the radio frequency unit 1701 includes, but is not limited to, antennas, amplifiers, transceivers, couplers, low-noise amplifiers, duplexers, etc.
  • the memory 1709 can be used to store software programs or instructions, as well as various data.
  • the memory 1709 may primarily include a first storage area for storing programs or instructions and a second storage area for storing data.
  • the first storage area may store the operating system, application programs or instructions required for at least one function (such as sound playback, image playback, etc.).
  • the memory 1709 may include volatile memory or non-volatile memory.
  • the non-volatile memory may be read-only memory (ROM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), or flash memory.
  • Volatile memory can be random access memory (RAM), static random access memory (SRAM), dynamic random access memory (DRAM), synchronous dynamic random access memory (SDRAM), double data rate synchronous dynamic random access memory (DDRSDRAM), enhanced synchronous dynamic random access memory (ESDRAM), synchronous link dynamic random access memory (SLDRAM), and direct memory bus RAM (DRRAM).
  • RAM random access memory
  • SRAM static random access memory
  • DRAM dynamic random access memory
  • SDRAM synchronous dynamic random access memory
  • DDRSDRAM double data rate synchronous dynamic random access memory
  • ESDRAM enhanced synchronous dynamic random access memory
  • SLDRAM synchronous link dynamic random access memory
  • DRRAM direct memory bus RAM
  • the memory 1709 in this embodiment includes, but is not limited to, these and any other suitable types of memory.
  • Processor 1710 may include one or more processing units; optionally, processor 1710 integrates an application processor and a modem processor, wherein the application processor mainly handles operations involving the operating system, user interface, and applications, and the modem processor mainly handles wireless communication signals, such as a baseband processor. It is understood that the aforementioned modem processor may also not be integrated into processor 1710.
  • the aforementioned device is used as the first device, and the first device is a terminal for illustrative purposes.
  • the radio frequency unit 1701 or the processor 1710 is used to acquire the perception data output by the first AI unit.
  • the radio frequency unit 1701 or the processor 1710 is used to acquire reference data for monitoring the first AI unit;
  • the processor 1710 is configured to determine the monitoring result of the first AI unit based on the reference data and the perception data output by the first AI unit.
  • the reference data includes at least one of the following:
  • the perception data output by the second AI unit has a processing capability that is stronger than that of the first AI unit.
  • the perception data acquired through non-AI methods includes at least one of the following:
  • Perception data acquired through target perception devices acquired through target perception devices.
  • acquiring the perception data output by the first AI unit includes:
  • the first measurement data is used as input to the first AI unit for inference to obtain the perception data output by the first AI unit, and the first measurement data includes the first measurement result;
  • the first measurement data includes the perception data output by the first AI unit.
  • the first measurement data further includes: first explanatory information, which includes at least one of the following:
  • the first measurement data includes timestamp information, perception mode information, indication information for indicating that the first measurement result is used for AI unit inference, information of the AI unit corresponding to the first measurement data, performance index information of the first measurement data, indication information for indicating that the first measurement data includes perception data output by the first AI unit, reference data type information associated with the first measurement data, accuracy information corresponding to the first measurement data, performance level information corresponding to the first measurement data, and confidence level information corresponding to the first measurement data.
  • obtaining reference data for monitoring the first AI unit includes:
  • the second measurement data is used to determine the reference data, and the second measurement data includes the second measurement result; or,
  • the second measurement data includes the reference data.
  • the second measurement data further includes second explanatory information, which includes at least one of the following:
  • the second measurement data includes timestamp information, perception mode information, indication information for monitoring the second measurement result, reference data type information associated with the second measurement data, performance index information of the second measurement data, indication information for indicating that the second measurement data includes the reference data, information of the AI unit corresponding to the second measurement data, accuracy information corresponding to the second measurement data, coordinate system information of the second measurement data, and sensor type information corresponding to the second measurement data.
  • the monitoring results include at least one of the following:
  • Effectiveness information for performance level information, and perception accuracy information
  • the perception accuracy information includes at least one of the following:
  • Distance accuracy information speed accuracy information, angle accuracy information, positioning accuracy information, detection accuracy information, and recognition accuracy information.
  • the radio frequency unit 1701 is also used for:
  • the target information including at least one of the following: signal configuration information, measurement configuration information, and indication information; wherein, the indication information is used to indicate at least one of the following:
  • the first measurement data is used to be input into the first AI unit for reasoning to obtain the perception data output by the first AI unit, and the first measurement data includes the first measurement result;
  • the second measurement data is used to determine the reference data, and the second measurement data includes the second measurement result.
  • the signal configuration information includes at least one of the following:
  • the configuration information of the first signal and the configuration information of the second signal wherein the first signal is a signal used to acquire input data of the first AI unit, and the second signal is a signal used to acquire the reference data, and the first signal and the second signal satisfy at least one of the following:
  • the frequency domain resource length of the second signal is greater than that of the first signal
  • the frequency domain resource interval of the second signal is smaller than that of the first signal
  • the time-domain resource length of the second signal is greater than that of the first signal
  • the time-domain resource unit of the second signal is smaller than the time-domain resource interval of the first signal
  • the transmission power of the second signal is greater than the transmission power of the first signal
  • the first signal is a subset of the second signal.
  • the measurement configuration information includes at least one of the following:
  • the measurement quantity of the first measurement data includes at least one of the following:
  • Received signal channel information, spectrum information, and basic measurement quantities
  • the measurement quantity of the second measurement data includes at least one of the following:
  • Received signal channel information, spectrum information, and basic measurement quantities.
  • the format information of the first measurement data is used to indicate at least one of the following:
  • channel information The dimensions of channel information, the range of spectral information, the types of basic measurements, and the format of sensor data;
  • the format information of the second measurement data is used to indicate at least one of the following:
  • the target information to be sent includes at least one of the following:
  • the second device is a device for providing at least one of the first measurement data and the second measurement data
  • the third device is a device for providing the second measurement data.
  • the second device satisfies at least one of the following conditions:
  • the supported transmit power exceeds the preset threshold
  • the number of supported antennas exceeds a preset threshold
  • Supported array aperture exceeds preset threshold
  • the amount of resources available for sensing exceeds a preset threshold
  • the clock synchronization error between the device and the third device is less than a preset threshold
  • the third device satisfies at least one of the following conditions:
  • the supported transmit power exceeds the preset threshold
  • the number of supported antennas exceeds a preset threshold
  • Supported array aperture exceeds preset threshold
  • the amount of resources available for sensing exceeds a preset threshold
  • the clock synchronization error between the device and the second device is less than a preset threshold.
  • processor 1710 is also used for:
  • a target operation is determined for the first AI unit, and the target operation includes at least one of the following:
  • the aforementioned equipment helps to improve or ensure sensing performance.
  • This application embodiment also provides a device, including a processor and a communication interface, wherein the communication interface is used to send data, and the sent data includes at least one of the following: sending first measurement data and sending second measurement data; wherein the first measurement data is used to be input to a first AI unit for inference to obtain perception data output by the first AI unit, and the first measurement data includes a first measurement result; or, the first measurement data includes perception data output by the first AI unit; the second measurement data is used to determine a reference signal, and the second measurement data includes a second measurement result; or, the second measurement data includes reference data, and the reference data is used to monitor the first AI unit.
  • This application also provides a device, which is a second device, including a processor and a communication interface.
  • the communication interface is coupled to the processor, and the processor is used to run programs or instructions to implement the steps of the method embodiment shown in FIG8.
  • This device embodiment corresponds to the above-described second device-side method embodiment. All implementation processes and methods of the above-described method embodiments can be applied to this device embodiment and can achieve the same technical effect.
  • this application embodiment also provides a device, which is a second device, and may be the signal transmitting device shown in FIG15.
  • the device 1800 includes: an antenna 1801, a radio frequency device 1802, a baseband device 1803, a processor 1804, and a memory 1805.
  • the antenna 1801 is connected to the radio frequency device 1802.
  • the radio frequency device 1802 receives information through the antenna 1801 and sends the received information to the baseband device 1803 for processing.
  • the baseband device 1803 processes the information to be transmitted and sends it to the radio frequency device 1802, which processes the received information and then transmits it through the antenna 1801.
  • the methods executed by the device in the above embodiments can be implemented in the baseband device 1803, which includes a baseband processor.
  • the baseband device 1803 may include at least one baseband board, on which multiple chips are disposed, as shown in FIG18.
  • One of the chips is, for example, a baseband processor, which is connected to the memory 1805 via a bus interface to call the program in the memory 1805 and execute the network device operation shown in the above method embodiment.
  • the device may also include a network interface 1806, such as a Common Public Radio Interface (CPRI).
  • a network interface 1806 such as a Common Public Radio Interface (CPRI).
  • CPRI Common Public Radio Interface
  • the device 1800 in this application embodiment further includes: instructions or programs stored in memory 1805 and executable on processor 1804.
  • Processor 1804 calls the instructions or programs in memory 1805 to execute the methods executed by each module shown in FIG15 and achieve the same technical effect. To avoid repetition, it will not be described in detail here.
  • the second device described above is used as a network-side device for illustration.
  • Radio frequency device 1802 is used to transmit data, the transmitted data including at least one of the following:
  • the first measurement data is used to be input into the first AI unit for reasoning to obtain the perception data output by the first AI unit, and the first measurement data includes the first measurement result; or, the first measurement data includes the perception data output by the first AI unit.
  • the second measurement data is used to determine a reference signal, and the second measurement data includes a second measurement result; or, the second measurement data includes reference data used to monitor the first AI unit.
  • the reference data includes at least one of the following:
  • the perception data output by the second AI unit has a processing capability that is stronger than that of the first AI unit.
  • the perception data acquired through non-AI methods includes at least one of the following:
  • Perception data acquired through target perception devices acquired through target perception devices.
  • the first measurement data further includes: first explanatory information, which includes at least one of the following:
  • the first measurement data includes timestamp information, perception mode information, indication information for indicating that the first measurement result is used for AI unit inference, information of the AI unit corresponding to the first measurement data, performance index information of the first measurement data, indication information for indicating that the first measurement data includes perception data output by the first AI unit, reference data type information associated with the first measurement data, accuracy information corresponding to the first measurement data, performance level information corresponding to the first measurement data, and confidence level information corresponding to the first measurement data.
  • the second measurement data further includes second explanatory information, which includes at least one of the following:
  • the second measurement data includes timestamp information, perception mode information, indication information for monitoring the second measurement result, reference data type information associated with the second measurement data, performance index information of the second measurement data, indication information for indicating that the second measurement data includes the reference data, information of the AI unit corresponding to the second measurement data, accuracy information corresponding to the second measurement data, coordinate system information of the second measurement data, and sensor type information corresponding to the second measurement data.
  • the radio frequency device 1802 is also used for:
  • Receive target information including at least one of the following: signal configuration information, measurement configuration information, and indication information; wherein, the indication information is used to indicate at least one of the following:
  • the signal configuration information includes at least one of the following:
  • the configuration information of the first signal and the configuration information of the second signal wherein the first signal is a signal used to acquire input data of the first AI unit, and the second signal is a signal used to acquire the reference data, and the first signal and the second signal satisfy at least one of the following:
  • the frequency domain resource length of the second signal is greater than that of the first signal
  • the frequency domain resource interval of the second signal is smaller than that of the first signal
  • the time-domain resource length of the second signal is greater than that of the first signal
  • the time-domain resource unit of the second signal is smaller than the time-domain resource interval of the first signal
  • the transmission power of the second signal is greater than the transmission power of the first signal
  • the first signal is a subset of the second signal.
  • the measurement configuration information includes at least one of the following:
  • the measurement quantity of the first measurement data includes at least one of the following:
  • Received signal channel information, spectrum information, and basic measurement quantities
  • the measurement quantity of the second measurement data includes at least one of the following:
  • Received signal channel information, spectrum information, and basic measurement quantities.
  • the format information of the first measurement data is used to indicate at least one of the following:
  • channel information The dimensions of channel information, the range of spectral information, the types of basic measurements, and the format of sensor data;
  • the format information of the second measurement data is used to indicate at least one of the following:
  • the radio frequency device 1802 is also used for:
  • a target operation is performed, wherein the target operation includes at least one of the following:
  • the second device satisfies at least one of the following conditions:
  • the supported transmit power exceeds the preset threshold
  • the number of supported antennas exceeds a preset threshold
  • Supported array aperture exceeds preset threshold
  • the amount of resources available for sensing exceeds a preset threshold
  • the clock synchronization error between the device and the third device is less than a preset threshold, wherein the third device is a device used to provide the second measurement data.
  • the aforementioned equipment helps to improve or ensure sensing performance.
  • this application embodiment also provides a network-side device, which is a first device.
  • the network-side device 1900 includes: a processor 1901, a network interface 1902, and a memory 1903.
  • the network interface 1902 is, for example, a common public radio interface (CPRI).
  • CPRI common public radio interface
  • the network-side device 1900 in this application embodiment further includes: instructions or programs stored in memory 1903 and executable on processor 1901.
  • Processor 1901 calls the instructions or programs in memory 1903 to execute the methods executed by each module shown in FIG15 and achieve the same technical effect. To avoid repetition, it will not be described in detail here.
  • the first device described above is used as an example of a core network device.
  • the network interface 1902 or the processor 1901 is used to acquire the perception data output by the first AI unit;
  • the network interface 1902 or the processor 1901 is used to acquire reference data for monitoring the first AI unit
  • Processor 1901 is used to determine the monitoring result of the first AI unit based on the reference data and the perception data output by the first AI unit.
  • the reference data includes at least one of the following:
  • the perception data output by the second AI unit has a processing capability that is stronger than that of the first AI unit.
  • the perception data acquired through non-AI methods includes at least one of the following:
  • Perception data acquired through target perception devices acquired through target perception devices.
  • acquiring the perception data output by the first AI unit includes:
  • the first measurement data is used as input to the first AI unit for inference to obtain the perception data output by the first AI unit, and the first measurement data includes the first measurement result;
  • the first measurement data includes the perception data output by the first AI unit.
  • the first measurement data further includes: first explanatory information, which includes at least one of the following:
  • the first measurement data includes timestamp information, perception mode information, indication information for indicating that the first measurement result is used for AI unit inference, information of the AI unit corresponding to the first measurement data, performance index information of the first measurement data, indication information for indicating that the first measurement data includes perception data output by the first AI unit, reference data type information associated with the first measurement data, accuracy information corresponding to the first measurement data, performance level information corresponding to the first measurement data, and confidence level information corresponding to the first measurement data.
  • obtaining reference data for monitoring the first AI unit includes:
  • the second measurement data is used to determine the reference data, and the second measurement data includes the second measurement result; or,
  • the second measurement data includes the reference data.
  • the second measurement data further includes second explanatory information, which includes at least one of the following:
  • the second measurement data includes timestamp information, perception mode information, indication information for monitoring the second measurement result, reference data type information associated with the second measurement data, performance index information of the second measurement data, indication information for indicating that the second measurement data includes the reference data, information of the AI unit corresponding to the second measurement data, accuracy information corresponding to the second measurement data, coordinate system information of the second measurement data, and sensor type information corresponding to the second measurement data.
  • the monitoring results include at least one of the following:
  • Effectiveness information for performance level information, and perception accuracy information
  • the perception accuracy information includes at least one of the following:
  • Distance accuracy information speed accuracy information, angle accuracy information, positioning accuracy information, detection accuracy information, and recognition accuracy information.
  • network interface 1902 is also used for:
  • the target information including at least one of the following: signal configuration information, measurement configuration information, and indication information; wherein, the indication information is used to indicate at least one of the following:
  • the first measurement data is used to be input into the first AI unit for reasoning to obtain the perception data output by the first AI unit, and the first measurement data includes the first measurement result;
  • the second measurement data is used to determine the reference data, and the second measurement data includes the second measurement result.
  • the signal configuration information includes at least one of the following:
  • the configuration information of the first signal and the configuration information of the second signal wherein the first signal is a signal used to acquire input data of the first AI unit, and the second signal is a signal used to acquire the reference data, and the first signal and the second signal satisfy at least one of the following:
  • the frequency domain resource length of the second signal is greater than that of the first signal
  • the frequency domain resource interval of the second signal is smaller than that of the first signal
  • the time-domain resource length of the second signal is greater than that of the first signal
  • the time-domain resource unit of the second signal is smaller than the time-domain resource interval of the first signal
  • the transmission power of the second signal is greater than the transmission power of the first signal
  • the first signal is a subset of the second signal.
  • the measurement configuration information includes at least one of the following:
  • the measurement quantity of the first measurement data includes at least one of the following:
  • Received signal channel information, spectrum information, and basic measurement quantities
  • the measurement quantity of the second measurement data includes at least one of the following:
  • Received signal channel information, spectrum information, and basic measurement quantities.
  • the format information of the first measurement data is used to indicate at least one of the following:
  • channel information The dimensions of channel information, the range of spectral information, the types of basic measurements, and the format of sensor data;
  • the format information of the second measurement data is used to indicate at least one of the following:
  • the target information to be sent includes at least one of the following:
  • the second device is a device for providing at least one of the first measurement data and the second measurement data
  • the third device is a device for providing the second measurement data.
  • the second device satisfies at least one of the following conditions:
  • the supported transmit power exceeds the preset threshold
  • the number of supported antennas exceeds the preset threshold
  • Supported array apertures exceed a preset threshold
  • the amount of resources available for sensing exceeds a preset threshold
  • the clock synchronization error between the device and the third device is less than a preset threshold
  • the third device satisfies at least one of the following conditions:
  • the supported transmit power exceeds the preset threshold
  • the number of supported antennas exceeds a preset threshold
  • Supported array aperture exceeds preset threshold
  • the amount of resources available for sensing exceeds a preset threshold
  • the clock synchronization error between the device and the second device is less than a preset threshold.
  • processor 1901 is also used for:
  • a target operation is determined for the first AI unit, and the target operation includes at least one of the following:
  • the aforementioned equipment helps to improve or ensure sensing performance.
  • This application also provides a readable storage medium storing a program or instructions.
  • the program or instructions When the program or instructions are executed by a processor, they implement the various processes of the AI unit monitoring method embodiments described above and achieve the same technical effect. To avoid repetition, they will not be described again here.
  • the processor mentioned above is the processor in the terminal described in the above embodiments.
  • the readable storage medium includes computer-readable storage media, such as computer read-only memory (ROM), random access memory (RAM), magnetic disk, or optical disk.
  • ROM computer read-only memory
  • RAM random access memory
  • magnetic disk magnetic disk
  • optical disk optical disk
  • the readable storage medium may be a non-transient readable storage medium.
  • This application embodiment also provides a chip, which includes a processor and a communication interface.
  • the communication interface is coupled to the processor.
  • the processor is used to run programs or instructions to implement the various processes of the above-described AI unit monitoring method embodiment and can achieve the same technical effect. To avoid repetition, it will not be described again here.
  • chip mentioned in the embodiments of this application may also be referred to as a system-on-a-chip, system chip, chip system, or system-on-a-chip, etc.
  • This application also provides a computer program/program product, which is stored in a storage medium and executed by at least one processor to implement the various processes of the AI unit monitoring method embodiment described above, and can achieve the same technical effect. To avoid repetition, it will not be described again here.
  • This application also provides a wireless communication system, including a first device and a second device.
  • the first device can be used to execute the steps of the AI unit monitoring method on the first device side as provided in this application
  • the second device can be used to execute the steps of the AI unit monitoring method on the second device side as provided in this application.

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • General Health & Medical Sciences (AREA)
  • Computing Systems (AREA)
  • Biomedical Technology (AREA)
  • Biophysics (AREA)
  • Computational Linguistics (AREA)
  • Data Mining & Analysis (AREA)
  • Evolutionary Computation (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Molecular Biology (AREA)
  • Artificial Intelligence (AREA)
  • General Engineering & Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Mathematical Physics (AREA)
  • Software Systems (AREA)
  • Health & Medical Sciences (AREA)
  • Computer Networks & Wireless Communication (AREA)
  • Signal Processing (AREA)
  • Mobile Radio Communication Systems (AREA)

Abstract

本申请公开了一种AI单元监控方法、装置及设备，属于通信技术领域，本申请实施例的AI单元监控方法包括：第一设备获取第一AI单元输出的感知数据；所述第一设备获取用于监控所述第一AI单元的参考数据；所述第一设备基于所述参考数据和所述第一AI单元输出的感知数据，确定所述第一AI单元的监控结果。

Description

AI单元监控方法、装置及设备
相关申请的交叉引用
本申请主张在2024年05月14日在中国提交的中国专利申请No.202410597731.2的优先权，其全部内容通过引用包含于此。
技术领域
本申请属于通信技术领域，具体涉及一种人工智能(Artificial Intelligence，AI)单元监控方法、装置及设备。
背景技术
在一些通信系统中支持感知，且支持通过AI单元推理感知数据。在一些相关技术中，往往是在训练完AI单元后，就一直使用该AI单元推理感知数据。而AI单元一般存在泛化问题，例如：在一些感知场景中或随着时间的推移，其处理能力可能不能满足感知需求。因此，一直使用该AI单元推理感知数据可能会容易使得感知数据准确性下降，导致感知性能容易变差。
发明内容
本申请实施例提供一种AI单元监控方法、装置及设备，能够解决一直使用AI单元推理感知数据可能会容易使得感知数据准确性下降导致的感知性能容易变差的问题。
第一方面，提供了一种AI单元监控方法，包括：
第一设备获取第一AI单元输出的感知数据；
所述第一设备获取用于监控所述第一AI单元的参考数据；
所述第一设备基于所述参考数据和所述第一AI单元输出的感知数据，确定所述第一AI单元的监控结果。
第二方面，提供了一种AI单元监控方法，包括：
第二设备发送数据，所述发送数据包括如下至少一项：
发送第一测量数据、发送第二测量数据；
其中，所述第一测量数据用于输入到第一AI单元进行推理得到所述第一AI单元输出的感知数据，且所述第一测量数据包括第一测量结果；或者，所述第一测量数据包括所述第一AI单元输出的感知数据；
所述第二测量数据用于确定参考信号，且所述第二测量数据包括第二测量结果；或者，所述第二测量数据包括参考数据，所述参考数据用于监控所述第一AI单元。
第三方面，提供了一种AI单元监控装置，包括：
处理模块，用于获取第一AI单元输出的感知数据；
所述处理模块，还用于获取用于监控所述第一AI单元的参考数据；
所述处理模块，还用于基于所述参考数据和所述第一AI单元输出的感知数据，确定所述第一AI单元的监控结果。
第四方面，提供了一种AI单元监控装置，包括：
发送模块，用于发送数据，所述发送数据包括如下至少一项：
发送第一测量数据、发送第二测量数据；
其中，所述第一测量数据用于输入到第一AI单元进行推理得到所述第一AI单元输出的感知数据，且所述第一测量数据包括第一测量结果；或者，所述第一测量数据包括所述第一AI单元输出的感知数据；
所述第二测量数据用于确定参考信号，且所述第二测量数据包括第二测量结果；或者，所述第二测量数据包括参考数据，所述参考数据用于监控所述第一AI单元。
第五方面，提供了一种AI单元监控装置，所述装置被配置为执行如本申请实施例提供的第一设备侧的AI单元监控方法的步骤，或者实现如本申请实施例提供的第二设备侧的AI单元监控方法的步骤。
第六方面，提供了一种设备，该终端包括处理器和存储器，所述存储器存储可在所述处理器上运行的程序或指令，所述程序或指令被所述处理器执行时实现如本申请实施例提供的第一设备侧的AI单元监控方法的步骤。
第七方面，提供了一种设备，包括处理器及通信接口，其中，所述处理器或者所述通信接口用于获取第一AI单元输出的感知数据；获取用于监控所述第一AI单元的参考数据；所述处理器用于基于所述参考数据和所述第一AI单元输出的感知数据，确定所述第一AI单元的监控结果。
第八方面，提供了一种设备，该设备包括处理器和存储器，所述存储器存储可在所述处理器上运行的程序或指令，所述程序或指令被所述处理器执行时实现如本申请实施例提供的第二设备侧的AI单元监控方法的步骤。
第九方面，提供了一种设备，包括处理器及通信接口，其中，所述通信接口用于发送数据，所述发送数据包括如下至少一项：发送第一测量数据、发送第二测量数据；其中，所述第一测量数据用于输入到第一AI单元进行推理得到所述第一AI单元输出的感知数据，且所述第一测量数据包括第一测量结果；或者，所述第一测量数据包括所述第一AI单元输出的感知数据；所述第二测量数据用于确定参考信号，且所述第二测量数据包括第二测量结果；或者，所述第二测量数据包括参考数据，所述参考数据用于监控所述第一AI单元。
第十方面，提供了一种可读存储介质，所述可读存储介质上存储程序或指令，所述程序或指令被处理器执行时实现如本申请实施例提供的第一设备侧的AI单元监控方法的步骤，或者实现如本申请实施例提供的第二设备侧的AI单元监控方法的步骤。
第十一方面，提供了一种无线通信系统，包括：第一设备及第二设备，所述第一设备可用于执行如本申请实施例提供的第一设备侧的AI单元监控方法的步骤，所述第二设备可用于执行如本申请实施例提供的第二设备侧的AI单元监控方法的步骤。
第十二方面，提供了一种芯片，所述芯片包括处理器和通信接口，所述通信接口和所述处理器耦合，所述处理器用于运行程序或指令，实现如本申请实施例提供的第一设备侧的AI单元监控方法，或实现如本申请实施例提供的第二设备侧的AI单元监控方法。
第十三方面，提供了一种计算机程序/程序产品，所述计算机程序/程序产品被存储在存储介质中，所述计算机程序/程序产品被至少一个处理器执行以实现如本申请实施例提供的第一设备侧的AI单元监控方法的步骤，或者所述计算机程序/程序产品被至少一个处理器执行以实现如本申请实施例提供的第二设备侧的AI单元监控方法的步骤。
在本申请实施例中，第一设备获取第一AI单元输出的感知数据；所述第一设备获取用于监控所述第一AI单元的参考数据；所述第一设备基于所述参考数据和所述第一AI单元输出的感知数据，确定所述第一AI单元的监控结果。这样由于基于参考数据和第一AI单元输出的感知数据确定第一AI单元的监控结果，从而实现对第一AI单元的监控，可以防止第一AI单元的可靠性变差，从而可以避免一直使用AI单元推理感知数据使得感知数据准确性下降导致的感知性能容易变差的问题，有利于提升或者保证感知性能。
附图说明
图1是本申请实施例可应用的一种无线通信系统的框图；
图2是本申请实施例提供的一种测量的场景示意图；
图3是本申请实施例提供的另一种测量的场景示意图；
图4是本申请实施例提供的一种神经网络的示意图；
图5是本申请实施例提供的一种神经元的示意图；
图6是本申请实施例提供的一种AI单元监控方法的流程图；
图7是本申请实施例提供的一种信号资源的示意图；
图8是本申请实施例提供的另一种AI单元监控方法的流程图；
图9是本申请实施例提供的一种AI单元监控方法的示意图；
图10是本申请实施例提供的一种时延谱信息子集截取的示意图；
图11是本申请实施例提供的另一种时延谱信息子集截取的示意图；
图12是本申请实施例提供的一种AI单元监控方法的示意图；
图13是本申请实施例提供的一种信号径检测的示意图；
图14是本申请实施例提供的一种AI单元监控装置的示意图；
图15是本申请实施例提供的另一种AI单元监控装置的示意图；
图16是本申请实施例提供的一种设备的结构图；
图17是本申请实施例提供的另一种设备的结构图；
图18是本申请实施例提供的另一种设备的结构图；
图19是本申请实施例提供的另一种设备的结构图。
具体实施方式
下面将结合本申请实施例中的附图，对本申请实施例中的技术方案进行清楚描述，显然，所描述的实施例是本申请一部分实施例，而不是全部的实施例。基于本申请中的实施例，本领域普通技术人员所获得的所有其他实施例，都属于本申请保护的范围。
本申请的术语“第一”、“第二”等是用于区别类似的对象，而不用于描述特定的顺序或先后次序。应该理解这样使用的术语在适当情况下可以互换，以便本申请的实施例能够以除了在这里图示或描述的那些以外的顺序实施，且“第一”、“第二”所区别的对象通常为一类，并不限定对象的个数，例如第一对象可以是一个，也可以是多个。此外，本申请中的“或”表示所连接对象的至少其中之一。例如“A或B”的保护范围至少涵盖三种方案，即，方案一：包括A且不包括B；方案二：包括B且不包括A；方案三：既包括A又包括B。此外，术语“A和/或B”、“A和B中的至少一项”、“A或B中的至少一项”也分别至少涵盖上述三种方案。字符“/”一般表示前后关联对象是一种“或”的关系。
本申请的术语“指示”既可以是一个直接的指示(或者说显式的指示)，也可以是一个间接的指示(或者说隐含的指示)。其中，直接的指示可以理解为，发送方在发送的指示中明确告知了接收方具体的信息、需要执行的操作或请求结果等内容；间接的指示可以理解为，接收方根据发送方发送的指示确定对应的信息，或者进行判断并根据判断结果确定需要执行的操作或请求结果等。
值得指出的是，本申请实施例所描述的技术不限于长期演进型(Long Term Evolution，LTE)/LTE的演进(LTE-Advanced，LTE-A)系统，还可用于其他无线通信系统，诸如码分多址(Code Division Multiple Access，CDMA)、时分多址(Time Division Multiple Access，TDMA)、频分多址(Frequency Division Multiple Access，FDMA)、正交频分多址(Orthogonal Frequency Division Multiple Access，OFDMA)、单载波频分多址(Single-carrier Frequency-Division Multiple Access，SC-FDMA)或其他系统。本申请实施例中的术语“系统”和“网络”常被可互换地使用，所描述的技术既可用于以上提及的系统和无线电技术，也可用于其他系统和无线电技术。以下描述出于示例目的描述了新空口(New Radio，NR)系统，并且在以下大部分描述中使用NR术语，但是这些技术也可应用于NR系统以外的系统，如第6代(6th Generation，6G)通信系统。
图1示出本申请实施例可应用的一种无线通信系统的框图。无线通信系统包括终端11和网络侧设备12。其中，终端11可以是手机、平板电脑(Tablet Personal Computer)、膝上型电脑(Laptop Computer)、笔记本电脑、个人数字助理(Personal Digital Assistant，PDA)、掌上电脑、上网本、超级移动个人计算机(Ultra-mobile Personal Computer，UMPC)、移动上网装置(Mobile Internet Device，MID)、增强现实(Augmented Reality，AR)、虚拟现实(Virtual Reality，VR)设备、机器人、可穿戴式设备(Wearable Device)、飞行器(flight vehicle)、车载用户设备(Vehicle User Equipment，VUE)、船载设备、行人用户设备(Pedestrian User Equipment，PUE)、智能家居(具有无线通信功能的家居设备，如冰箱、电视、洗衣机或者家具等)、游戏机、个人计算机(Personal Computer，PC)、柜员机或者自助机等终端侧设备。可穿戴式设备包括：智能手表、智能手环、智能耳机、智能眼镜、智能首饰(智能手镯、智能手链、智能戒指、智能项链、智能脚镯、智能脚链等)、智能腕带、智能服装等。其中，车载设备也可以称为车载终端、车载控制器、车载模块、车载部件、车载芯片或车载单元等。需要说明的是，在本申请实施例并不限定终端11的具体类型。网络侧设备12可以包括接入网设备或核心网设备，其中，接入网设备也可以称为无线接入网(Radio Access Network,RAN)设备、无线接入网功能或无线接入网单元。接入网设备可以包括基站、无线局域网(Wireless Local Area Network，WLAN)接入点(Access Point，AP)或无线保真(Wireless Fidelity，WiFi)节点等。其中，基站可被称为节点B(Node B，NB)、演进节点B(Evolved Node B，eNB)、下一代节点B(the next generation Node B，gNB)、新空口节点B(New Radio Node B，NR Node B)、接入点、中继站(Relay Base Station，RBS)、服务基站(Serving Base Station，SBS)、基收发机站(Base Transceiver Station，BTS)、无线电基站、无线电收发机、基本服务集(Basic Service Set，BSS)、扩展服务集(Extended Service Set，ESS)、家用B节点(home Node B，HNB)、家用演进型B节点(home evolved Node B)、发送接收点(Transmit/Receive Point，TRP)或所属领域中其他某个合适的术语，只要达到相同的技术效果，所述基站不限于特定技术词汇，需要说明的是，在本申请实施例中仅以NR系统中的基站为例进行介绍，并不限定基站的具体类型。
核心网设备也可以称为核心网节点、核心网功能或核心网网元等，其包含但不限于如下至少一项：移动管理实体(Mobility Management Entity，MME)、接入移动管理功能(Access and Mobility Management Function，AMF)、会话管理功能(Session Management Function，SMF)、用户平面功能(User Plane Function，UPF)、策略控制功能(Policy Control Function，PCF)、策略与计费规则功能单元(Policy and Charging Rules Function，PCRF)、边缘应用服务发现功能(Edge Application Server Discovery Function，EASDF)、统一数据管理(Unified Data Management，UDM)、统一数据仓储(Unified Data Repository，UDR)、归属用户服务器(Home Subscriber Server，HSS)、集中式网络配置(Centralized network configuration，CNC)、网络存储功能(Network Repository Function，NRF)、网络开放功能(Network Exposure Function，NEF)、本地NEF(Local NEF，或L-NEF)、绑定支持功能(Binding Support Function，BSF)、应用功能(Application Function，AF)、位置管理功能(Location Management Function，LMF)、网关的移动位置中心(Gateway Mobile Location Centre，GMLC)、网络数据分析功能(Network Data Analytics Function，NWDAF)等。需要说明的是，在本申请实施例中仅以NR系统中的核心网设备为例进行介绍，并不限定核心网设备的具体类型，如果在后续协议版本(例如6G)中本申请实施例提到的核心网设备的名称发生变化，也在本申请的保护范围内。
可选的，核心网设备可以由一个设备中的一个或多个功能模块实现，也可以由多个设备共同实现，本申请实施例对此不作具体限定。可以理解的是，上述功能模块既可以是硬件设备中的网络元件，也可以是在专用硬件上运行的软件功能模块，或者是平台(例如，云平台)上实例化的虚拟化功能模块。
在一些实施例中，网络侧设备和终端除了具备通信能力外可以具备感知能力，感知能力，即具备感知能力的一个或多个设备，能够通过无线信号的发送和接收，来感知目标物体的方位、距离、速度等信息，或者对目标物体、事件或环境等进行检测、跟踪、识别、成像等。一些感知功能与应用场景如表1所示：
表1
需要说明的是，上述表1所示的感知类别仅是一个举例说明，本申请实施例中对感知测量的类别并不作限定。
另外，本申请实施例可以应用于通信感知一体化场景，其中，通信感知一体化是指在同一系统中通过频谱共享与硬件共享，实现通信和感知功能一体化设计，系统在进行信息传递的同时，能够感知方位、距离、速度等信息，对目标设备或事件进行检测、跟踪、识别，通信系统与感知系统相辅相成，实现整体性能上的提升并带来更好的服务体验。
例如：通信与雷达的一体化属于典型的通信感知一体化(通信感知融合)应用，且通信与雷达系统融合能够带来许多优势，例如节约成本、减小尺寸、降低功耗、提升频谱效率、减小互干扰等，从而提升系统整体性能。
本申请实施例中，根据感知信号发送节点和接收节点的不同，可以包括但不限于图2所示的6种感知链路。需要说明的是，图2中每种感知链路都是以一个发送节点和一个接收节点进行举例说明，实际系统中，根据不同的感知需求可以选择不同的感知链路，每种感知链路的发送节点和接收节点可以有一个或多个，且实际感知系统可以包括多种不同的感知链路。且图2中的感知目标以人和车作为例子，且假设人和车均没有携带或安装信号收/发设备，实际场景的感知目标将更加丰富。
感知链路1：基站自发自收感知。该方式下基站发送感知信号，并通过接收该感知信号的回波来获得感知结果；
感知链路2：基站间空口感知。该方式下基站2接收基站1发送的感知信号，获得感知结果。
感知链路3：上行空口感知。该方式下基站接收终端发送的感知信号，获得感知结果。
感知链路4：下行空口感知。该方式下终端接收基站发送的感知信号，获得感知结果。
感知链路5：终端自发自收感知。该方式下终端发送感知信号，并通过接收该感知信号的回波来获得感知结果。
感知链路6：终端间旁链路(Sidelink)感知。例如，终端2接收终端1发送的感知信号，获得感知结果，或者终端1接收终端2发送的感知信号，获得感知结果。
在一些实施例中，无线接入网设备和终端、不同终端之间的信令传输以是通过无线资源控制(Radio Resource Control，RRC)信令或媒体接入控制控制单元(Medium Access Control Control Element，MAC CE)或层1信令或其他新定义感知信令；感知网络功能和终端之间的信令传输可以是通过非接入层(Non-Access-Stratum，NAS)信令(经AMF转发)或通过RRC信令或MAC CE或层1信令或其他新定义感知信令；感知网络功能和基站之间的交互可以是利用AMF通过N2接口转发给无线接入网；或者核心网感知网络功能发送给UPF，UPF通过N3接口发送给无线接入网；或者通过新定义的接口发送给无线接入网(如基站)；无线接入网设备间的信令传输可以是通过Xn接口。
在一些实施例中，感知网络功能也可以叫做感知网元或者感知管理功能(Sensing Management Function，Sensing MF)，可以处于无线接入网侧或核心网侧，即可以是无线接入网设备或者核心网设备，具体可以是指核心网或RAN中负责感知请求处理、感知资源调度、感知信息交互、感知数据处理等至少一项功能的网络节点，可以是基于移动通信网络中AMF或LMF升级，也可以是其他网络节点或新定义的网络节点，具体的，感知网络功能/感知网元的功能特性可以包括以下至少一项：
与无线信号发送设备或无线信号测量设备(包括目标终端或者目标终端的服务基站或者目标区域关联的基站)进行目标信息交互，其中，目标信息包括感知处理请求，感知能力，感知辅助数据，感知测量量类型，感知资源配置信息等，以获得无线信号测量设备发送目标感知结果或感知测量量(上行测量量或下行测量量)的值；其中，无线信号也可以称作感知信号。
根据感知业务的类型、感知业务消费者信息、所需的感知服务质量(Quality of Service，QoS)要求信息、无线信号发送设备的感知能力、无线信号测量设备的感知能力等因素来决定使用的感知方法，该感知方法可以包括：无线接入网设备A发无线接入网设备B收，或者无线接入网设备发终端收，或者无线接入网设备A自发自收，或者终端发无线接入网设备收，或者终端自发自收，或者终端A发终端B收等。
根据感知业务的类型、感知业务消费者的信息、所需的感知QoS要求信息、无线信号发送设备的感知能力、无线信号测量设备的感知能力等因素，来决定为感知业务服务的感知设备，其中，感知设备包括无线信号发送设备或无线信号测量设备。
管理感知业务所需资源的整体协调和调度，如对无线接入网设备或终端的感知资源进行相应的配置；
对感知测量量的值进行数据处理，或进行计算获得感知结果。还可以验证感知结果，估计感知精度等。
在一些实施例中，LMF是5G核心网中提供控制面定位的核心网元，完成5G网络中位置信息的计算和反馈，提供定位过程管理、终端能力获取、辅助数据提供、终端位置估计等功能，具体提供如下至少一项功能：
支持终端的位置计算；
获得来自终端的下行链路位置测量或位置估计；
从下一代无线接入网(Next Generation Radio Access Network，NG RAN)获得上行链路位置测量。；
从NG RAN获得非终端相关的辅助数据。
LMF支持小区标识(CELL Identifier，CELL ID)、上行时间差到达(Uplink Time Difference of Arrival，UL-TDOA)、辅助全球导航卫星系统(Assisting Global Navigation Satellite System，A-GNSS等中高精度定位方法。
在一些实施例中，雷达按发射机和接收机是否分置可分为单基地雷达和双/多基地雷达，双基地雷达一般要求发射和接收天线距离很远，与雷达作用距离可比拟。其中，外辐射源雷达是双基地雷达的一种特例，利用相关的电磁波探测理论技术与信号处理技术，获取第三方(例如通信基站)发射的非合作电磁信号，实现对目标的探测、定位、跟踪和识别，又叫做无源雷达、双/多基地无源雷达、被动雷达、非合作照射源雷达或非合作无源探测系统。
其中，双基地雷达感知结果计算一般需要基于参考信道(直达径)信号和监测信道(反射径)信号，典型的双基地雷达架构示意图如图3所示。其中，RT为信号发端(Tx)到目标距离，RR为信号接收端(Tx)到目标距离，L为基线距离，θT为目标相对于信号发送端的角度，θR(θR1、θR2)为目标相对于信号接收端的角度，β为双基地角。
AI目前在各个领域获得了广泛的应用，将人工智能融入无线通信网络，显著提升吞吐量、时延以及用户容量等技术指标是未来的无线通信网络的重要任务。AI模块有多种实现方式，例如神经网络、决策树、支持向量机、贝叶斯分类器等。
在一些实施例中，以神经网络为例进行说明，但是并不限定AI模块的具体类型。一个神经网络的示意图如图4所示。其中，神经网络由神经元组成，神经元的示意图如图5所示。其中a1,a2,…aK为输入，w为权值(乘性系数)，b为偏置(加性系数)，σ(.)为激活函数。常见的激活函数包括Sigmoid、tanh、线性整流函数(Rectified Linear Unit，ReLU)等等。
在一些实施例中，神经网络的参数通过梯度优化算法进行优化。梯度优化算法是一类最小化或者最大化目标函数(有时候也叫损失函数)的算法，而目标函数往往是模型参数和数据的数学组合。例如给定数据X和其对应的标签Y，我们构建一个神经网络模型f(.)，有了模型后，根据输入x就可以得到预测输出f(x)，并且可以计算出预测值和真实值之间的差距(f(x)-Y)，这个就是损失函数。目的是找到合适的W,b使上述的损失函数的值达到最小，损失值越小，则说明模型越接近于真实情况。
优化算法可以是基于误差反向传播(error Back Propagation，BP)算法。BP算法的基本思想是，学习过程由信号的正向传播与误差的反向传播两个过程组成。正向传播时，输入样本从输入层传入，经各隐层逐层处理后，传向输出层。若输出层的实际输出与期望的输出不符，则转入误差的反向传播阶段。误差反传是将输出误差以某种形式通过隐层向输入层逐层反传，并将误差分摊给各层的所有单元，从而获得各层单元的误差信号，此误差信号即作为修正各单元权值的依据。这种信号正向传播与误差反向传播的各层权值调整过程，是周而复始地进行的。权值不断调整的过程，也就是网络的学习训练过程。此过程一直进行到网络输出的误差减少到可接受的程度，或进行到预先设定的学习次数为止。
优化算法还可以有梯度下降(Gradient Descent)、随机梯度下降(Stochastic Gradient Descent，SGD)、小批量梯度下降(mini-batch gradient descent)、动量法(Momentum)、Nesterov(发明者的名字，具体为带动量的随机梯度下降)、自适应梯度下降(ADAptive GRADient descent，Adagrad)、Adadelta、均方根误差降速(root mean square prop，RMSprop)、自适应动量估计(Adaptive Moment Estimation，Adam)等。
这些优化算法在误差反向传播时，都是根据损失函数得到的误差/损失，对当前神经元求导数/偏导，加上学习速率、之前的梯度/导数/偏导等影响，得到梯度，将梯度传给上一层。
本申请实施例中，AI单元也可称为AI模型、机器学习(machine learning，ML)模型、ML单元、AI结构、AI功能、AI特性、机器学习模型、神经网络、神经网络函数、神经网络功能等，或者AI单元也可以是指能够实现与AI相关的特定的算法、公式、处理流程、能力等的处理单元，或者AI单元可以是针对特定数据集的处理方法、算法、功能、模块或单元，或者AI单元可以是运行在图形处理器(graphics processing unit，GPU)、神经处理单元(Neural Processing Unit，NPU)、张量处理单元即(Tensor Processing Unit，TPU)、专用集成电路(application specific integrated circuit，ASIC)等AI/ML相关硬件上的处理方法、算法、功能、模块或单元，本申请实施例对此不做具体限定。可选地，上述特定数据集包括AI单元的输入和/或输出。
在一些实施例中，AI单元的标识可以是AI模型标识、AI结构标识、AI算法标识，或者AI单元关联的特定数据集的标识，或者AI/ML相关的特定场景、环境、信道特征、设备的标识，或者AI/ML相关的功能、特性、能力或模块的标识，本申请实施例对此不做具体限定。
下面结合附图，通过一些实施例及其应用场景对本申请实施例提供的AI单元监控方法、装置及设备进行详细地说明。
请参见图6，图6是本申请实施例提供的一种AI单元监控方法的流程图，如图6所示，包括以下步骤：
步骤601、第一设备获取第一AI单元输出的感知数据。
上述第一设备可以是终端或者网络侧设备，如可以是无线接入网设备或者核心网设备。
上述第一AI单元为感知关联的AI单元，如第一AI单元输出的感知数据为感知结果或者用于确定感知结果的中间数据。
上述第一AI单元可以是部署在上述第一设备上的AI单元，如上述感知数据为上述第一设备使用上述第一AI单元进行推理得到的感知数据。或者，上述第一AI单元可以是部署在其他设备上的AI单元，如上述感知数据为第一设备接收其他设备发送的第一AI单元输出的感知数据。
本申请实施例中，对第一AI单元部署的设备不作限定，第一AI单元可以是部署在上述第一设备上，或者第一AI单元部署在其他设备的，例如：第一AI单元可以部署在核心网感知网络功能或定位管理功能，或部署在基站或终端。其中，该基站或终端包括负责上述感知相关的信号发送或接收，以及感知数据处理的基站或终端，或者不负责上述感知相关的信号发送或接收，仅负责感知数据处理的基站或终端。
上述第一AI单元输出的感知数据也可以称作通过第一AI单元推理得到的感知数据，具体可以是感知结果或者用于确定感知结果的中间数据。
步骤602、所述第一设备获取用于监控所述第一AI单元的参考数据。
上述第一设备获取用于监控所述第一AI单元的参考数据可以是第一设备通过测量、计算或推理等方式获取的参考数据。或者，第一设备获取用于监控所述第一AI单元的参考数据可以是第一设备接收其他设备发送的参考数据。
上述参考数据可以理解为上述第一AI单元的真实标签或者参考标签，即作为上述第一AI单元输出的感知数据的参考。
本申请实施例中步骤601和步骤602的执行顺序不作限定，例如：步骤601可以是在步骤602之前执行，或者步骤601和步骤602同时执行，步骤601可以是在步骤602之后执行。
步骤603、所述第一设备基于所述参考数据和所述第一AI单元输出的感知数据，确定所述第一AI单元的监控结果。
上述第一设备基于所述参考数据和所述第一AI单元输出的感知数据，确定所述第一AI单元的监控结果可以是将参考数据与上述感知数据进行比较，确定二者的差异，进而确定第一AI单元的监控结果。或者可以是计算参考数据和第一AI单元输出的感知数据之间的损失值，基于该损失值确定第一AI单元的监控结果等，具体不作限定。
在本申请实施例中，第一设备获取第一AI单元输出的感知数据；所述第一设备获取用于监控所述第一AI单元的参考数据；所述第一设备基于所述参考数据和所述第一AI单元输出的感知数据，确定所述第一AI单元的监控结果。这样由于基于参考数据和第一AI单元输出的感知数据确定第一AI单元的监控结果，从而实现对第一AI单元的监控，可以防止第一AI单元的可靠性变差，从而可以避免一直使用AI单元推理感知数据使得感知数据准确性下降导致的感知性能容易变差的问题，有利于提升或者保证感知性能。
另外，由于基于所述参考数据和所述第一AI单元输出的感知数据，确定所述第一AI单元的监控结果，这样实现了对第一AI单元的监控，从而可以确保第一AI单元有效运行，进而保证基于第一AI的感知性能可靠。
由于第一设备基于所述参考数据和所述第一AI单元输出的感知数据，确定所述第一AI单元的监控结果，这样可以针对第一AI单元存在的泛化问题进行性能监控，以避免AI单元性能发生显著下降。
作为一种可选的实施方式，所述参考数据包括如下至少一项：
通过非AI的方式获取的感知数据；
第二AI单元输出的感知数据，所述第二AI单元的处理能力强于所述第一AI单元的处理能力。
上述非AI的方式是指不通过AI推理的方式。
在一些实施方式中，上述通过非AI的方式获取的感知数据包括如下至少一项：
通过估计算法获取的感知数据；
通过传感器获取的感知数据；
通过目标感知模式获取的感知数据；
通过目标感知设备获取的感知数据。
上述估计算法可以是非AI的参数估计算法，例如：采用无线感知获取测量数据，但不将获取的测量数据输入到AI单元得到感知数据，而是通过其他方式计算得到感知数据，例如根据信道信息通过谱估计或参数估计算法计算的感知数据作为基于第一AI单元推理得到的感知数据的参考数据。如通过二维快速傅里叶变换(2D-Fast Fourier Transform，2D-FFT)、三维快速傅里叶变换(3D-Fast Fourier Transform，3D-FFT)、多重信号分类(Multiple Signal Classification，MUSIC)、ESPRIT算法计算的感知数据作为基于第一AI单元推理得到的感知数据的参考数据。
上述通过传感器获取的感知数据可以是通过可见光摄像头、红外摄像头、全球导航卫星系统(Global Navigation Satellite System，GNSS)、激光雷达、毫米波雷达、温度计、湿度计、气压计、陀螺仪、加速度计、磁力计、重力传感器、声呐、雨量计等传感器获取的感知数据，以将传感器获取的感知数据作为无线感知中基于第一AI单元推理得到的感知数据的参考数据。
上述目标感知模式可以是单基地感知模式或双基地感知模式，如将单基地感知模式获取的感知数据作为双基地感知模式下通过第一AI单元推理得到的感知数据的参考数据。
上述目标感知设备可以是特定设备，如基站或者终端，如将基站获取的感知数据作为终端通过第一AI单元推理得到的感知数据的参考数据。
上述第二AI单元输出的感知数据可以理解为通过能力更强的AI单元输出的感知数据作为第一AI单元推理得到的感知数据的参考数据。
需要说明的是，上述参考数据可以是上述第一设备基于上述方式获取的参考数据，或者可以是第一设备接收其他设备发送的参考数据，即其他设备基于上述方式获取参考数据并发送给第一设备。
在上述一种可选的实施方式中可以实现获取多种方式的参考数据，以提高AI单元监控的灵活性，且也可以满足更多业务或者场景的需求。
作为一种可选的实施方式，所述第一设备获取第一AI单元输出的感知数据包括：
所述第一设备接收第一测量数据；
其中，所述第一测量数据用于输入到所述第一AI单元进行推理得到所述第一AI单元输出的感知数据，且所述第一测量数据包括第一测量结果；或者
所述第一测量数据包括所述第一AI单元输出的感知数据。
上述接收第一测量数据可以是接收第二设备发送的测量数据，即上述测量数据是第二设备获取的，第二设备可以是终端或者网络侧设备。
其中，上述第一测量数据用于输入到所述第一AI单元进行推理得到所述第一AI单元输出的感知数据可以理解为该感知数据是基于上述第一测量数据进行AI推理得到的。即上述第一设备将上述第一测量数据输入到上述第一AI单元进行推理得到上述感知数据。
上述第一测量结果为执行测量流程得到测量结果，可以是信道信息、谱信息、时延、多普勒、角度或强度等测量结果，且具体可以是特定测量量的值，或者满足特定格式要求的测量量的值等。
上述第一测量数据包括上述第一测量结果可以理解为至少将第一测量结果输入到上述第一AI单元进行推理。上述第一测量结果可以是第二设备执行测量流程得到的测量结果。上述第一测量结果可以是信道信息、谱信息或基本测量量等测量结果。
上述第一测量数据包括所述第一AI单元输出的感知数据可以理解为该感知数据是第一设备接收其他设备发送的，即第一AI单元部署在其他设备，如部署在第二设备上，如第二设备执行测量流程，得到测量结果，再将该测量结果输入到第一AI单元进行推理得到感知数据。
在上述实施方式中，可以实现支持多种方式获取第一AI单元输出的感知数据，以提高AI单元监控的灵活性。
可选地，所述第一测量数据还包括：第一说明信息，所述第一说明信息包括如下至少一项：
所述第一测量数据的时间戳信息、所述第一测量数据的感知模式信息、用于指示所述第一测量结果用于AI单元推理的指示信息、所述第一测量数据对应的AI单元的信息、所述第一测量数据的性能指标信息、用于指示所述第一测量数据包括第一AI单元输出的感知数据的指示信息、所述第一测量数据关联的参考数据类型信息、所述第一测量数据对应的精度信息、所述第一测量数据对应的性能等级信息、所述第一测量数据对应的置信度水平信息。
其中，上述时间戳信息可以是绝对时间，如国际标准时间(International Standard Time，UTC)，或相对于某个参考时间点的时间，例如无线帧号，半帧号，子帧号，传输时间间隔(Transmission Time Interval，TTI)序号，时隙序号，子时隙(sub-slot)序号，符号序号等；其中，时隙序号或符号序号也可以是指相对于感知相干处理时间的起始slot或符号的序号。
感知模式信息用于指示第一测量数据对应的感知模式，如图2所示的感知模式中的至少一种。
上述用于指示所述第一测量结果用于AI单元推理的指示信息可以理解为该指示信息用于指示上述第一测量结果为用于AI单元携带的测量结果，或者用于指示第一测量结果为用于AI推理和AI单元监控的公共的测量结果，即第一测量结果可以用于AI推理和AI单元监控。
上述第一测量数据对应的AI单元的信息用于指示上述第一测量数据输入到的AI单元的标识，如模型标识(model ID)，具体用于指示该第一测量数据为输入到哪一个AI单元的测量数据。
上述性能指标信息可以包括但不限于参考数据信号接收功率(Reference Signal Received Power，RSRP)、参考信号接收质量(Reference Signal Received Quality，RSRQ)、信号与干扰加噪声比(Signal to Interference plus Noise Ratio，SINR)、感知性能指标。
通过性能指标信息可以更好地指示测量数据性能，这样第一设备可以选择地基于第一测量数据进行AI单元，更好地保障感知性能，如在性能指标信息表示第一测量数据的性能指标高于预设门限的情况下基于第一测量数据进行AI推理得到上述感知数据。
上述参考数据类型信息可以是通过非AI的方式获取的感知数据或第二AI单元输出的感知数据。
上述精度信息可以表示上述第一测量结果的精度，具体可以是具体精度或者精度等级。
上述性能等级信息可以表示上述第一测量结果的性能等级，如指示第一测量结果的性能指标的等级。
上述置信度水平信息可以表示上述第一测量结果的置信度水平。
上述实施方式中，由于第一测量数据还包括上述第一说明信息，这样在基于第一测量数据进行AI推理，可以提高第一AI单元输出的感知数据的可靠性。一些实施方式中，第一AI单元的输入数据包括上述第一测量结果，还包括上述第一说明信息中的全部或者部分内容，或者在一些实施方式中，第一AI单元的输入数据包括上述第一测量结果，但不包括上述第一说明信息。
需要说明的是，上述第一说明信息中对于第一测量数据的相关信息可以理解为是对第一测量结果的相关信息，如第一测量数据的时间戳信息可以理解为第一测量结果的时间戳信息，所述第一测量数据的感知模式信息可以理解为第一测量结果的感知模式信息，第一测量数据对应的AI单元的信息可以理解为第一测量结果对应的AI单元的信息，第一测量数据的性能指标信息可以理解为第一测量结果的性能指标信息，所述第一测量数据关联的参考数据类型信息可以理解为第一测量结果关联的参考数据类型信息，第一测量数据对应的精度信息可以理解为第一测量结果的精度信息，第一测量数据对应的性能等级信息可以理解为第一测量结果对应的性能等级信息，第一测量数据对应的置信度水平信息可以理解为第一测量结果对应的置信度水平信息。
作为一种可选的实施方式，所述第一设备获取用于监控所述第一AI单元的参考数据包括：
所述第一设备接收第二测量数据；
其中，所述第二测量数据用于确定所述参考数据，且所述第二测量数据包括第二测量结果；或者，
所述第二测量数据包括所述参考数据。
上述第二测量数据用于确定所述参考数据可以理解为第一设备在接收到上述第二测量数据后，基于该测量数据确定上述参考数据，如将第二测量结果输入到第二AI单元进行推理得到上述参考数据，或者基于第二测量结果进行非AI的计算得到上述参考数据。
上述第二测量结果为执行测量流程得到测量结果，可以是信道信息、谱信息、时延、多普勒、角度或强度等测量结果，且具体可以是特定测量量的值，或者满足特定格式要求的测量量的值等。
上述第二测量数据包括所述参考数据可以理解为参考数据是其他设备生成的，第一设备接收其他设备发送的参考数据。
在上述实施方式中，可以实现支持多种方式获取参考数据，以提高AI单元监控的灵活性。
可选地，所述第二测量数据还包括第二说明信息，所述第二说明信息包括如下至少一项：
所述第二测量数据的时间戳信息、所述第二测量数据的感知模式信息、用于所述第二测量结果用于监控的指示信息、所述第二测量数据关联的参考数据类型信息、所述第二测量数据的性能指标信息、用于指示所述第二测量数据包括所述参考数据的指示信息、所述第二测量数据对应的AI单元的信息、所述第二测量数据对应的精度信息、所述第二测量数据的坐标系信息、所述第二测量数据对应的传感器类型信息。
其中，上述时间戳信息、感知模式信息、参考数据类型信息、性能指标信息、精度信息等参见上述实施方式的相应说明，此处不作赘述。在一些实施方式中，当感知性能指标超过预设门限时，认为第二测量数据的测量结果是可以用于第一AI单元的监控的测量结果。
上述坐标系信息可以是传感器对应坐标系与全局坐标系的关系，例如本地坐标系到全局坐标系的转换参数，即本地坐标系相对于全局坐标系的转动角度：α(轴承角)，β(下倾角)和γ(倾斜角)。
上述实施方式中，由于第二测量数据还包括上述第二说明信息，这样在基于第二测量数据确定参考数据，可以提高参考数据的可靠性。一些实施方式中，第二AI单元的输入数据包括上述第二测量结果，还包括上述第二说明信息中的全部或者部分内容，或者在一些实施方式中，第二AI单元的输入数据包括上述第二测量结果，但不包括上述第二说明信息。
需要说明的是，上述第二说明信息中对于第二测量数据的相关信息可以理解为是对第二测量结果的相关信息，如第二测量数据的时间戳信息可以理解为第二测量结果的时间戳信息，所述第二测量数据的感知模式信息可以理解为第二测量结果的感知模式信息，第二测量数据对应的AI单元的信息可以理解为第二测量结果对应的AI单元的信息，第二测量数据的性能指标信息可以理解为第二测量结果的性能指标信息等。
作为一种可选的实施方式，所述监控结果包括如下至少一项：
有效性信息、性能等级信息、感知精度信息；
其中，所述感知精度信息包括如下至少一项：
距离精度信息、速度精度信息、角度精度信息、定位精度信息、检测精度信息、识别精度信息。
其中，上述有效性信息可以是用于表示上述第一AI单元是否有效，或者表示上述第一AI单元是否适用。
上述感知精度信息可以是感知误差信息或者感知误差统计信息。
上述实施方式中，可以实现以多个维度对第一AI单元进行监控，以提高AI单元的监控效果。
在一些实施方式中，上述距离精度信息可以表示目标物体的距离估计结果与其真实距离(即参考数据表示的距离)的接近程度。例如：对于单基地感知，目标物体的距离可以是指目标物体与信号接收设备之间的距离；对于双基地感知，目标物体的距离可以是指目标物体与信号接收设备之间的距离，或者目标物体与信号接收设备以及发送设备之间的距离和，或者目标物体与信号接收设备以及发送设备之间的距离和减去收发设备间的距离，或者目标物体与信号接收设备以及发送设备之间的距离和在双基地角平分线上的投影。
上述距离精度信息可以是根据单次感知结果计算得到的任一目标距离误差信息，即
上述距离精度信息也可以是根据单次感知结果计算得到的全部目标距离误差的和，即
上述距离精度信息也可以是根据多次感知结果计算得到的距离均方根误差，即：
其中，为第i次计算获取的第一AI单元输出的感知数据中第j个目标的距离值，Rij为第i次计算获取的参考数据中第j个目标的距离值。N表示感知结果计算次数(或第二设备上报第一测量数据和第二测量数据的次数)，M表示每次感知结果计算时的目标个数。
上述速度精度信息表示目标物体的速度测量结果与其真实速度(即参考数据表示的速度)的接近程度。对于单基地感知，目标物体的速度可以是指目标物体运动速度在目标与信号接收设备连线上的投影；对于双基地感知，目标物体的速度可以是指目标物体运动速度在双基地角平分线上的投影，或者目标物体运动速度在目标与信号接收设备连线上的投影以及目标物体运动速度在目标与信号发送设备连线上的投影之和。除此之外，还可以是指目标物体旋转速度、动作重复速率(例如呼吸/心跳速率)等。
上述速度精度信息可以是根据单次感知结果计算得到的任一目标速度误差信息，即
上述速度精度信息也可以是根据单次感知结果计算得到的全部目标速度误差的和，即
上述速度精度信息也可以是根据多次感知结果计算得到的速度均方根误差，即：
其中，为第i次计算获取的第一AI单元输出的感知数据中第j个目标的速度值，vij为第i次计算获取的参考数据中第j个目标的速度值。
上述角度精度表示目标物体的角度测量结果与其真实角度(即参考数据表示的角度)的接近程度，包括方位角精度和俯仰角精度。对于单基地感知，目标物体的角度可以是指目标物体相对于信号接收设备的到达角度；对于双基地感知，目标物体的角度可以是指目标物体相对于信号接收设备的到达角度，或者目标物体相对于信号发送设备的离开角度。
上述角度精度信息可以是根据单次感知结果计算得到的任一目标角度误差信息，即
上述角度精度信息也可以是根据单次感知结果计算得到的全部目标角度误差的和，即
上述角度精度信息也可以是根据多次感知结果计算得到的角度均方根误差，即：
其中，为第i次计算获取的第一AI单元输出的感知数据中第j个目标的角度值，θij为第i次计算获取的参考数据中第j个目标的角度值。
上述定位精度信息表示目标物体的位置测量结果与其真实位置(即参考数据表示的位置)的接近程度。其中，目标物体的位置坐标可以由其距离和角度中的至少一项进一步计算得到。
上述定位精度信息可以是根据单次感知结果计算得到的任一目标位置坐标误差信息，即：
上述定位精度信息也可以是根据单次感知结果计算得到的全部目标位置坐标误差的和，即：
上述定位精度信息也可以是根据多次感知结果计算得到的位置坐标均方根误差，即：
其中，为第i次计算获取的第一AI单元输出的感知数据中第j个目标的位置坐标，(xij,yij,zij)为第i次计算获取的参考数据中第j个目标的位置坐标。
上述检测精度信息或者识别精度信息可以表示第一AI单元输出的感知数据与参考数据关于目标的有无、个数、类别的检测/识别的结果一致的次数或不一致的次数；或者上述检测精度信息或者识别精度信息也可以是第一AI单元输出的感知数据与参考数据关于目标的有无、个数、类别的检测/识别的结果一致的次数或不一致的次数占总的感知结果计算次数的比例。
作为一种可选的实施方式，所述方法还包括：
所述第一设备发送目标信息，所述目标信息包括如下至少一项：信号配置信息、测量配置信息、指示信息；其中，所述指示信息用于指示如下至少一项：
获取第一测量数据、获取第二测量数据、上报第一测量数据、上报第二测量数据；
其中，所述第一测量数据用于输入到所述第一AI单元进行推理得到所述第一AI单元输出的感知数据，且所述第一测量数据包括第一测量结果。
上述第一设备发送目标信息可以是向第二设备发送上述目标信息。
上述信号配置信息包括用于获取第一AI单元的输入数据的信号的配置信息，也可以包括用于获取上述参考数据的信号的配置信息。其中，用于获取第一AI单元的输入数据的信号是指感知数据是基于该信号获取的，如对该信号进行测量，得到测量结果，该测量结果为第一AI单元的输入数据。用于获取上述参考数据的信号是指参考数据是基于该信号获取的，如对该信号进行测量，得到测量结果，再基于该测量结果确定参考数据。
在一些实施方式中，用于获取第一AI单元的输入数据的信号和用于获取上述参考数据的信号是同一个信号或者不同信号，或者用于获取第一AI单元的输入数据的信号是用于获取上述参考数据的信号的子集。
上述测量配置信息可以感知数据相关的测量或参考信号的相关测量。
上述指示信息获取第一测量数据、获取第二测量数据或上报第一测量数据、上报第二测量数据是指指示接收到上述目标信息的设备获取第一测量数据、获取第二测量数据或上报第一测量测量数据、上报第二测量数据，如指示第二设备获取第一测量数据、获取第二测量数据或上报第一测量测量数据、上报第二测量数据。
在一些实施方式中，如果指示获取第一测量数据则可以隐式指示上报第一测量数据，如果指示上报第一测量数据可以隐式指示获取第一测量数据，如果指示获取第二测量数据则可以隐式指示上报第二测量数据，如果指示上报第二测量数据可以隐式指示获取第二测量数据。
一些实施方式中，可以是通过上述信号配置信息或测量配置指示来显式或者隐式指示获取测量数据或上报获取第一测量数据、获取第二测量数据、上报第一测量数据、上报第二测量数据。
在上述一种可选的实施方式中，由于发送上述目标信息，这样可以使得接收到目标信息的设备基于该信息进行测量数据获取或上报，进而第一设备能够及时得到上述感知数据和参考数据，以提高AI单元监控性能。
可选地，所述信号配置信息包括如下至少一项：
第一信号的配置信息、第二信号的配置信息，所述第一信号为用于获取所述第一AI单元的输入数据的信号，所述第二信号为用于获取所述参考数据的信号，所述第一信号与所述第二信号满足如下至少一项：
所述第二信号的频域资源长度大于所述第一信号的频域资源长度；
所述第二信号的频域资源间隔小于所述第一信号的频域资源间隔；
所述第二信号的时域资源长度大于所述第一信号的时域资源长度；
所述第二信号的时域资源单元小于所述第一信号的时域资源间隔；
所述第二信号的发射功率大于所述第一信号的发射功率；
所述第一信号为所述第二信号的子集。
其中，上述配置信息可以是信号的用途、类型、资源等配置信息，例如：配置信息包括如下至少一项：
信号资源标识，用于区分不同的信号资源配置；
信号用途，表示该信号是用于通信(例如信道测量、信道估计、同步、承载数据信息等)的信号，用于感知的信号，或者是同时用于通信和感知的信号。具体的，还可以是，用于哪种感知业务的信号，或者是用于哪一类感知业务的信号。感知业务可以包括如下至少一项：检测目标是否存在，定位，速度探测，距离探测、角度探测、加速度探测，材料分析，成分分析，形状检测，类别划分，雷达散射截面积(Radar Cross Section，RCS)检测，极化散射特性检测，跌倒检测，入侵检测，数量统计，室内定位，手势识别，唇语识别，步态识别，表情识别，面部识别，呼吸监测，心率监测，脉搏监测，湿度/亮度/温度/大气压强监测，空气质量监测，天气情况监测，环境重构，地形地貌、建筑/植被分布检测，人流量或车流量检测，人群密度、车辆密度检测等；所述感知业务类型可以是按照一定特征把多个不同的感知业务进行分类，例如按照功能划分为检测类感知业务(例如包括入侵检测、跌倒检测)、参数估计类感知业务(距离、角度、速度计算)、识别类感知业务(动作识别、身份识别)等；还可以是按照感知目标类型分类，包括无人机(Unmanned Aerial Vehicle，UAV)，人(Human)，汽车(Automotive vehicle)，自动导向车(Automated Guided Vehicle，AGV)，公路/铁路上的物体(Objects on roads/railways)；还可以是按照感知的范围(近距离感知、中距离感知、远距离感知)划分，按照感知的精细程度划分(粗粒度感知、精细力度感知等)，按照功耗/能耗划分，按照资源占用划分等；
波形，所述波形例如为正交频分复用(Orthogonal frequency division multiplex，OFDM)，单载波频分多址(Single-carrier Frequency-Division Multiple Access,SC-FDMA)，正交时频空间(Orthogonal Time Frequency Space，OTFS)，调频连续波(Frequency Modulated Continuous Wave，FMCW)，脉冲信号等；
子载波间隔，例如，OFDM系统的子载波间隔30KHz；
保护间隔，保护间隔为从信号结束发送的时刻到该信号的最迟回波信号被接收的时刻之间的时间间隔；该参数正比于最大感知距离；例如，可以通过c/(2Rmax)计算得到，Rmax为最大感知距离(属于感知需求信息)，例如对于自发自收的感知信号，Rmax代表感知信号收发点到信号发射点的最大距离；在某些情况下，OFDM信号循环前缀(CP)可以起到最小保护间隔的作用；c是光速。
起始频域位置，即起始频点，也可以是起始资源单元(Resource Element，RE)、资源块(Resource Block，RB)索引；
起始时域位置，即起始时间点，也可以是起始符号索引、时隙索引、帧索引；
终止频域位置，即终止频点，可以用终止RE、RB索引表示；
终止时域位置，即终止时间点，可以表示用终止RE、RB索引表示；
频域资源长度，即频域带宽，所述频域带宽反比于距离分辨率，每个所述第一信号的频域带宽B≥c/(2ΔR)，其中，c为光速，ΔR为距离分辨率；
时域资源长度，也称为突发(burst)持续时间，时域资源长度反比于多普勒分辨率；
频域资源间隔，表示相邻的信号频域资源单元间隔，可以用RE数或RB数表示，也可以用密度值Density表示，例如Density＝1表示每个RB中有一个RE用于承载信号。所述频域资源间隔反比于最大无模糊距离/时延，其中，对于OFDM系统当子载波采用连续映射时频域间隔等于子载波间隔；所述频域资源单元间隔也可以用梳状映射参数Kcomb表示，例如表示comb1(Kcomb＝1)频域上连续映射，comb2(Kcomb＝2)表示频域上每隔1个子载波进行序列映射(例如信号占子载波0，2，4，…)，comb4(Kcomb＝4)为表示频域上每隔3个子载波进行序列映射(例如信号占子载波0，4，8，…)；
时域资源间隔，时域资源间隔是相邻的两个信号资源单元之间的时间间隔，时域资源间隔与最大无模糊多普勒频移或最大无模糊速度关联；
时域资源特性，如周期性发送，半持续性发送，非周期性发送。
时域burst资源间隔或时域burst发送周期，时域burst资源间隔或时域burst发送周期与感知结果刷新频率关联；
信号功率，如从-20dBm到23dBm每隔2dBm取一个值；
序列信息，包括序列类型信息(ZC序列、PN序列等)，序列生成方式，序列长度等；
信号方向，信号发送的角度信息或波束信息；
准共址(Quasi co-location，QCL)关系，例如信号包括多个资源，每个资源与一个同步信号块(Synchronization Signal Block，SSB)QCL，QCL包括类型A，类型B，类型C或者类型D。
循环前缀(Cyclic Prefix，CP)信息可以包括CP类型或CP长度等，例如常规循环前缀(Normal Cyclic Prefix，NCP)、扩展循环前缀(Extended Cyclic Prefix，ECP)或者新设计的感知测量专用CP等。
其中，上述第一信号为第二信号的子集，可以如图7所示，假设感知信号相干处理时间(即计算一次感知信息所需的信号时域资源长度)为Tp2＝80个slot(也可以是Tp1)，第一信号发送周期为1个slot，第二信号发送周期为0.5个slot，第一信号的频域密度(频域资源单元间隔)为每12个子载波中有一个用于承载第一信号，第二信号在频域占连续的子载波。如此一来，基于第二信号进行测量，相比于基于第一信号进行测量，具有更大的无模糊测距范围和无模糊测速范围，且具有更高的处理增益。
该实施方式中，由于第二信号的资源大于第一信号的资源，或者第一信号为所述第二信号的子集，这样可以使得参考数据的可靠性高于第一AI单元输出的感知数据的可靠性，从而能够更好监控第一AI单元。
需要说明的是，本申请实施例中，并不限定上述第一信号和第二信号满足上述关系，例如：第一信号和第二信号为同一个信号。
需要说明的是，上述信号配置信息包括全部或者部分也可以是协议约定或者预先配置。
可选地，所述测量配置信息包括如下至少一项：
测量的信号资源指示、所述第一测量数据对应的测量用途信息、所述第二测量数据对应的测量用途信息、所述第一测量数据的处理方式信息、所述第二测量数据的处理方式信息、所述第一测量数据对应的AI单元的信息、所述第一测量数据的测量量、所述第二测量数据的测量量、所述第一测量数据的格式信息、所述第二测量数据的格式信息、所述参考数据的类型信息、上报配置、所述第一测量数据是否基于传感器进行测量的指示、所述第二测量数据是否基于传感器进行测量的指示、所述第一测量数据关联的传感器的类型信息、所述第二测量数据关联的传感器的类型信息、所述第一测量数据的测量方式、所述第二测量数据的测量方式。
上述测量的信号资源指示可以指示上述第一信号和第二信号中的至少一项，如指示第一信号和第二信号中至少一项的标识，该标识与信号资源存在映射关系，从而隐式指示第一信号和第二信号中的至少一项的资源，以节约配置开销。
上述第一测量数据对应的测量用途信息可以指示上述第一测量数据用于AI推理。
上述第二测量数据对应的测量用途信息可以指示第二测量数据用于AI单元监控。
在一些实施方式中，不同测量用途的测量量、测量量格式是预先设定的，这样通过用途信息就可以确定测量量或格式，从而节约配置开销。
上述处理方式信息可以指示AI处理方式或非AI处理方式，对于上述第一测量数据采用AI处理方式，对于上述第二测量数据采用非AI处理方式。
在一些实施方式中，不同处理方式对应的测量量或测量量格式是预先设定的。
一些实施方式中，所述第一测量数据的测量量包括如下至少一项：
接收信号、信道信息、谱信息、基本测量量；
一些实施方式中，所述第二测量数据的测量量包括如下至少一项：
接收信号、信道信息、谱信息、基本测量量。
其中，上述信道信息可以是原始信道信息，如包括时域信道响应和频域信道响应)，具体可以包括信道响应的复数结果、幅度/相位、I路/Q路数据中的至少一项。
上述谱信息可以是基于信道信息或接收信号计算得到的谱信息，可以包括以下至少一项：时延(距离)谱，多普勒(速度)谱，角度谱，时延(距离)-多普勒(速度)谱，时延(距离)-角度谱，时延(距离)-多普勒(速度)-角度谱，时间-多普勒谱(微多普勒谱)。
在一些实施方式中，上述谱信息可以是指复数结果，例如时延-多普勒谱指的是2维谱中的时延、多普勒索引以及对应的复数值(含相位信息)；所述谱信息也可以是指功率谱，例如时延-多普勒谱指的是2维谱中的时延、多普勒索引以及对应的功率值(不含相位信息)。
上述基本测量量可以包括以下至少一项：
时延、多普勒、角度、强度、功率。
一些实施方式中，上述基本测量量可以是实际数值的量化结果，也可以是软信息。
一些实施方式中，对于第二测量数据的测量量还可以是参考数据，包括以下至少一项：感知目标的时延、多普勒、角度、距离、速度、朝向、空间位置、加速度、目标是否存在、目标个数、轨迹、手势、动作、表情、生命体征、数量、成像结果、天气、空气质量、形状、材质、成分。
上述格式信息用于表示测量数据的格式，如测量量格式。上述第一测量数据的格式信息表示输入到第一AI单元的测量数据(即第一测量数据)的格式，上述第二测量数据的格式信息表示用于计算参考数据的测量数据(即第二测量数据)的格式。
在一些实施方式中，上述第一测量数据的格式信息用于指示如下至少一项：
信道信息的维度、谱信息的范围、基本测量量的类型、传感器数据的格式。
在一些实施方式中，所述第二测量数据的格式信息用于指示如下至少一项：
信道信息的维度、谱信息的范围、基本测量量的类型。
上述信道信息的维度可以是某个符号上的频域信道响应息(一维信道信息)，或者多个符号上的时频域信道响应信息(二维信道信息)，或者多个符号、多个天线的时频空域信道响应信息(三维信道信息)，以及不同维度的规模，即采样点个数或采样间隔(例如时频域密度)；
谱信息的范围可以用于限制不同谱信息的范围，或者称之为针对谱信息不同维度的截断窗口，例如时延-多普勒谱中特定时延或多普勒范围对应的谱信息子集；也可以指示特定谱的径的个数的上限值或采样点个数的上限值；也可以是指示特定谱的最小粒度，即相邻两个采样点间的间隔(对应感知分辨率)。
上述基本测量量的类型可以是测量量的值的量化方式(例如量化粒度)或者软信息类型，如提供均值+方差，或置信区间和置信度的软信息。
通过上述第一测量数据和第二测量数据的格式信息可以使得测量数据的格式匹配，从而降低AI推理或者参考数据确定的计算量。
上述上报配置可以指示第一测量数据或第二测量数据上报的准则，例如：包括上报的时频域资源配置、上报周期、上报的触发事件中的至少一项。
其中，所述触发事件包括以下至少一项：
进入特定区域(例如小区)的事件；
到达特定时间的事件；
某类测量信号达到一定阈值的事件；
设备从先前位置移动超过一些预定义的(直线)距离的事件；
设备朝向改变超过一些预定义的角度的事件，其中，设备朝向可以是设备的天线、屏幕等器件的朝向；
设备的运动速度超过一些预定义的速度阈值的事件；
设备传感器测量得到的环境信息变化(例如温度/湿度/光照强度)超过一定范围的事件。
通过上述上报配置信息可以使得第二设备能够进行更加可靠的上报。
上述传感器的类型信息可以是传感器的数据格式，例如二进制格式、ASCII格式等，也可以是与特定传感器类型关联的数据格式，例如对于图像传感器(摄像头)，可以是YUV、RGB等。
上述测量方式可以指示采用AI处理或非AI处理，如指示采用AI处理还可以指示AI单元的信息，以及AI单元的精度信息或可以达到的性能等级或置信度水平等信息。
通过上述测量配置信息可以使得第二设备能够进行更加精准地测量，以测量AI单元监控的复杂度。
需要说明的是，上述测量配置信息包括全部或者部分也可以是协议约定或者预先配置。
可选地，所述第一设备发送目标信息包括如下至少一项：
所述第一设备向第二设备发送所述目标信息的全部或者部分；
所述第一设备向第三设备发送所述目标信息的全部或者部分；
其中，所述第二设备为用于提供第一测量数据和第二测量数据中的至少一项的设备，所述第三设备为用于提供第二测量数据的设备。
其中，上述第三设备可以是终端或者网络侧设备。
在一些实施方式中，所述第二设备满足如下至少一项条件：
支持目标感知模式；
位置固定；
支持的发射功率超过预设门限；
支持的天线个数超过预设门限；
支持的阵列孔径超过预设门限；
可分配给用于感知的资源量超过预设门限；
与所述第三设备之间的时钟同步误差小于预设门限。
其中，上述目标感知模式可以是图2所示的感知模式中的至少一种。
上述预设门限可以是协议约定或者网络侧配置的门限。
由于上述第二设备满足上述至少一项条件，这样可以使得第二设备提供的测量数据更加可靠，进而使得AI单元监控更加可靠。
在一些实施方式中，所述第三设备满足如下至少一项条件：
支持目标感知模式；
位置固定；
支持的发射功率超过预设门限；
支持的天线个数超过预设门限；
支持的阵列孔径超过预设门限；
可分配给用于感知的资源量超过预设门限；
与所述第二设备之间的时钟同步误差小于预设门限。
其中，上述目标感知模式可以是图2所示的感知模式中的至少一种。
上述预设门限可以是协议约定或者网络侧配置的门限。
由于上述第三设备满足上述至少一项条件，这样可以使得第三设备提供的测量数据更加可靠，进而使得AI单元监控更加可靠。
作为一种可选的实施方式，所述方法还包括：
所述第一设备基于所述监控结果确定针对所述第一AI单元的目标操作，所述目标操作包括如下至少一项：
去激活、切换、微调、重训练。
上述基于所述监控结果确定针对所述第一AI单元的目标操作可以是在上述监控结果满足预设条件的情况下，针对所述第一AI单元的目标操作，如监控结果表示第一AI单元无效、不适用性能等级较低、精度较低等。
在确定上述目标操作后，如果第一AI单元部署在第一设备，则第一设备执行上述目标操作，如果第一AI单元部署在第二设备，则第一设备向第二设备发送通知，通知第二设备执行上述操作。
该实施方式中，由于确定上述目标操作，这样可以支持对第一AI单元进行去激活、切换、微调、重训练，以提高AI单元的可靠性。
在一些实施方式中，基于在监控结果停止上述AI单元，如使用其他AI单元或者其他方式感知，以提升或者保证感知性能。
在本申请实施例中，第一设备获取第一AI单元输出的感知数据；所述第一设备获取用于监控所述第一AI单元的参考数据；所述第一设备基于所述参考数据和所述第一AI单元输出的感知数据，确定所述第一AI单元的监控结果。这样由于基于参考数据和第一AI单元输出的感知数据确定第一AI单元的监控结果，从而实现对第一AI单元的监控，可以防止第一AI单元的可靠性变差，从而可以避免直使用AI单元推理感知数据使得感知数据准确性下降导致的感知性能容易变差的问题，有利于提升或者保证感知性能。
请参见图8，图8是本申请实施例提供的一种AI单元监控方法的流程图，如图8所示，包括如下步骤：
步骤801、第二设备发送数据，所述发送数据包括如下至少一项：
发送第一测量数据、发送第二测量数据；
其中，所述第一测量数据用于输入到第一AI单元进行推理得到所述第一AI单元输出的感知数据，且所述第一测量数据包括第一测量结果；或者，所述第一测量数据包括所述第一AI单元输出的感知数据；
所述第二测量数据用于确定参考信号，且所述第二测量数据包括第二测量结果；或者，所述第二测量数据包括参考数据，所述参考数据用于监控所述第一AI单元。
可选地，所述参考数据包括如下至少一项：
通过非AI的方式获取的感知数据；
第二AI单元输出的感知数据，所述第二AI单元的处理能力强于所述第一AI单元的处理能力。
可选地，所述通过非AI的方式获取的感知数据包括如下至少一项：
通过估计算法获取的感知数据；
通过传感器获取的感知数据；
通过目标感知模式获取的感知数据；
通过目标感知设备获取的感知数据。
可选地，所述第一测量数据还包括：第一说明信息，所述第一说明信息包括如下至少一项：
所述第一测量数据的时间戳信息、所述第一测量数据的感知模式信息、用于指示所述第一测量结果用于AI单元推理的指示信息、所述第一测量数据对应的AI单元的信息、所述第一测量数据的性能指标信息、用于指示所述第一测量数据包括第一AI单元输出的感知数据的指示信息、所述第一测量数据关联的参考数据类型信息、所述第一测量数据对应的精度信息、所述第一测量数据对应的性能等级信息、所述第一测量数据对应的置信度水平信息。
可选地，所述第二测量数据还包括第二说明信息，所述第二说明信息包括如下至少一项：
所述第二测量数据的时间戳信息、所述第二测量数据的感知模式信息、用于所述第二测量结果用于监控的指示信息、所述第二测量数据关联的参考数据类型信息、所述第二测量数据的性能指标信息、用于指示所述第二测量数据包括所述参考数据的指示信息、所述第二测量数据对应的AI单元的信息、所述第二测量数据对应的精度信息、所述第二测量数据的坐标系信息、所述第二测量数据对应的传感器类型信息。
可选地，所述方法还包括：
所述第二设备接收目标信息，所述目标信息包括如下至少一项：信号配置信息、测量配置信息、指示信息；其中，所述指示信息用于指示如下至少一项：
获取所述第一测量数据、获取所述第二测量数据、上报所述第一测量数据、上报所述第二测量数据。
可选地，所述信号配置信息包括如下至少一项：
第一信号的配置信息、第二信号的配置信息，所述第一信号为用于获取所述第一AI单元的输入数据的信号，所述第二信号为用于获取所述参考数据的信号，所述第一信号与所述第二信号满足如下至少一项：
所述第二信号的频域资源长度大于所述第一信号的频域资源长度；
所述第二信号的频域资源间隔小于所述第一信号的频域资源间隔；
所述第二信号的时域资源长度大于所述第一信号的时域资源长度；
所述第二信号的时域资源单元小于所述第一信号的时域资源间隔；
所述第二信号的发射功率大于所述第一信号的发射功率；
所述第一信号为所述第二信号的子集。
可选地，所述测量配置信息包括如下至少一项：
测量的信号资源指示、所述第一测量数据对应的测量用途信息、所述第二测量数据对应的测量用途信息、所述第一测量数据的处理方式信息、所述第二测量数据的处理方式信息、所述第一测量数据对应的AI单元的信息、所述第一测量数据的测量量、第二测量数据的测量量、所述第一测量数据的格式信息、所述第二测量数据的格式信息、所述参考数据的类型信息、上报配置、所述第一测量数据是否基于传感器进行测量的指示、所述第二测量数据是否基于传感器进行测量的指示、所述第一测量数据关联的传感器的类型信息、所述第二测量数据关联的传感器的类型信息、所述第一测量数据的测量方式、所述第二测量数据的测量方式。
可选地，所述第一测量数据的测量量包括如下至少一项：
接收信号、信道信息、谱信息、基本测量量；
所述第二测量数据的测量量包括如下至少一项：
接收信号、信道信息、谱信息、基本测量量。
可选地，所述第一测量数据的格式信息用于指示如下至少一项：
信道信息的维度、谱信息的范围、基本测量量的类型、传感器数据的格式；
所述第二测量数据的格式信息用于指示如下至少一项：
信道信息的维度、谱信息的范围、基本测量量的类型。
可选地，所述方法还包括：
所述第二设备基于所述目标信息执行目标操作，所述目标操作包括如下至少一项：
执行第一测量，得到所述第一测量结果；
执行第二测量，得到第二测量结果；
执行第三测量，并将所述第三测量的测量结果输入到所述第一AI单元进行推理，得到所述第一AI单元输出的感知数据；
执行第四测量，并基于所述第四测量的测量结果确定所述参考数据。
可选地，所述第二设备满足如下至少一项条件：
支持目标感知模式；
位置固定；
支持的发射功率超过预设门限；
支持的天线个数超过预设门限；
支持的阵列孔径超过预设门限；
可分配给用于感知的资源量超过预设门限；
与第三设备之间的时钟同步误差小于预设门限，所述第三设备为用于提供第二测量数据的设备。
需要说明的是，本实施例作为与图6所示的实施例中对应的第二设备的实施方式，其具体的实施方式可以参见图6所示的实施例的相关说明，以为避免重复说明，本实施例不再赘述。
下面通过多个实施例，对本申请实施例提供的方法进行举例说明：
实施例一：
本实施例主要描述基于非AI的参数估计获取参考数据的监控。
本实施例中，第一AI单元部署在第一设备侧，第一设备为感知网络功能，或基站，或终端，第二设备为基站或终端。第一设备从第二设备获取输入到第一AI单元的第一测量数据，和用于通过参数估计的方法计算参考数据的第二测量数据。具体流程如图9所示。
步骤1、第一设备(以感知网络功能(SensingMF)为例)获取感知需求信息，感知需求信息的来源可以是：
感知需求来自外部应用，此时应用功能(Application Function，AF)发送感知需求信息给NEF，再发送给AMF，AMF选择SensingMF，并将感知需求信息发送给SensingMF；
或者，AF直接发送感知需求信息给SensingMF；
感知需求信息也可以来自基站和/或终端，此时基站和/或终端发送给AMF，AMF选择SensingMF，并将感知需求信息发送给SensingMF；
或者，基站和/或终端直接将感知需求信息发送给SensingMF；
感知需求信息也可以来自核心网网元，核心网网元发送感知需求给AMF；
AMF选择SensingMF，并将感知需求信息发送给SensingMF；
或者核心网网元直接将感知需求信息发送给SensingMF。
其中，经过AMF转发感知需求信息的方式可能但不限于出现在网络中部署了多个感知网元，需要由AMF根据感知对象的位置、感知业务类型或感知QoS要求信息等信息从多个感知网元中选择一个合适的感知网元的场景；不经过AMF转发感知需求的方式可能但不限于出现在，网络中部署了一个或较少的SensingMF的场景。
若第一设备为基站或终端，第一设备获取感知需求信息的方式可以接收感知网络功能发送的感知需求信息；或者该步骤为可选的。
其中，感知需求信息包括以下至少一项：
感知业务或感知业务类型，其中，感知业务或感知业务类型参见上述实施方式的相应说明，此处不作赘述；
感知目标区域，感知目标区域可以是指感知对象可能存在位置区域，或者，需要进行成像或环境重构的位置区域；
感知对象类型，感知对象类型可以是针对感知对象可能的运动特性对感知对象进行分类，每个感知对象类型中包含了典型感知对象的运动速度、运动加速度、典型RCS等信息；
感知QoS，感知QoS可以是对感知目标区域或感知对象进行感知的性能指标，包括以下至少一项：
感知分辨率，可分为：测距分辨率、测角分辨率、测速分辨率、成像分辨率等；
感知精度，可分为：测距精度、测角精度、测速精度、定位精度等；
感知范围，可分为：测距范围、测速范围、测角范围、成像范围等；
感知时延，感知时延可以是从感知信号发送到获得感知结果的时间间隔，或，从感知需求发起到获取感知结果的时间间隔；
感知更新速率，如相邻两次执行感知并获得感知结果的时间间隔；
检测概率，如在感知对象存在的情况下被正确检测出来的概率；
识别概率(用于多元检测场景，表示感知目标处于特定状态或属于特定类别，正确检测到目标状态或类别的概率)；
虚警概率，如在感知对象不存在的情况下错误检测出感知目标的概率；
可感知的最大目标个数。
步骤2、第一设备向第二设备发送第一指示信息(具体可以是上述实施例中的目标信息的全部或者部分信息)，第一指示信息用于指示第二设备获取测量数据和/或上报测量数据，即向第二设备请求用于第一AI单元的推理和监控的测量数据，所述第一指示信息包括以下至少一项：
信号配置信息、测量配置信息。
其中，信号配置信息包括第一信号的配置信息和第二信号的配置信息中的至少一项；所述第一信号为用于获取第一AI单元输入数据的信号，所述第二信号为用于获取参考数据的信号；
其中，第一信号和第二信号为相同的信号，或第一信号和第二信号为不同的信号，且第一信号与第二信号的特征满足以下至少一项：
第二信号的频域资源长度(带宽)大于第一信号的频域资源长度；
第二信号的频域资源单元间隔小于第一信号的频域资源间隔；
第二信号的时域资源长度大于第一信号的时域资源长度；
第二信号的时域资源单元间隔(发送周期)小于第一信号的时域资源间隔；
第二信号的发送功率大于第一信号。
可以理解的是，采用相同的测量方法时，基于第二信号的测量性能要好于基于第一信号的测量性能。
在一些实施方式中，第一信号为第二信号的子集，例如图7所示。感知信号相干处理时间(计算一次感知信息所需的信号时域资源长度)为Tp2＝80个slot(也可以是Tp1)，第一信号发送周期为1个slot，第二信号发送周期为0.5个slot，第一信号的频域密度(频域资源单元间隔)为每12个子载波中有一个用于承载第一信号，第二信号在频域占连续的子载波。如此一来，基于第二信号进行测量，相比于基于第一信号进行测量，具有更大的无模糊测距范围和无模糊测速范围，且具有更高的处理增益。
上述测量配置信息包括以下至少一项：
测量的信号资源指示、测量用途或测量数据处理方法的指示、测量量、测量量格式、与监控关联的参考数据类型、上报配置。
其中，测量的信号资源指示可以第一信号的标识或第二信号的标识；
测量用途或测量数据处理方法的指示包括以下至少一项：该测量为用于AI单元的推理或监控的指示，不同用途的测量量是预先设定的，第二设备根据所述指示即可确定对应的测量量；测量数据后续采用AI处理或非AI处理的指示，可选的，不同处理方法对应的测量量或测量量格式是预先设定的，第二设备根据所述指示即可确定对应的测量量以及测量量格式；对于本实施例，采用AI处理即用于AI单元的推理的测量数据，采用非AI处理即用于AI单元的监控的测量数据；测量数据输入的AI单元的标识信息，即不同AI单元对应的测量量或测量量格式不同，第二设备根据所述指示即可确定对应的测量量以及测量量格式；
测量量可以是第一设备用于第一AI单元输入的测量量，或用于计算参考数据的测量量，包括以下至少一项：接收信号或原始信道信息(包括时域信道响应和频域信道响应)，包括以下至少一项：接收信号或信道响应的复数结果，幅度/相位，I路/Q路数据；基于原始信道信息计算得到的谱信息，包括以下至少一项：时延(距离)谱，多普勒(速度)谱，角度谱，时延(距离)-多普勒(速度)谱，时延(距离)-角度谱，时延(距离)-多普勒(速度)-角度谱，时间-多普勒谱(微多普勒谱)；其中，所述谱信息可以是指复数结果，例如时延-多普勒谱指的是2维谱中的时延、多普勒索引以及对应的复数值(含相位信息)；所述谱信息也可以是指功率谱，例如时延-多普勒谱指的是2维谱中的时延、多普勒索引以及对应的功率值(不含相位信息)。
基本测量量包括以下至少一项：时延、多普勒、角度、强度(功率)；
其中，所述基本测量量可以是实际数值的量化结果，也可以是软信息。
上述测量量可以是AI推理和监控各自对应的测量量，即分别指示第一AI单元推理使用的测量量(例如信道信息或谱信息)和用于基于非AI的参数估计计算参考数据的测量量(例如基本测量量，具体的，例如基于信道信息获取谱信息后，谱信息中超过预设门限的径或采样点对应的时延值或多普勒值或角度值)。
特别的，对于用于计算参考数据的测量量，还可以是感知结果，包括以下至少一项：感知目标的时延、多普勒、角度、距离、速度、朝向、空间位置、加速度、目标是否存在、目标个数、轨迹、手势、动作、表情、生命体征、数量、成像结果、天气、空气质量、形状、材质、成分。所述感知结果即参考数据，即所述基于非AI的参数估计方法计算参考数据是在第二设备侧完成并直接将所述参考数据发送给第一设备的。
上述测量量格式可以是输入到第一AI单元的测量数据(即第一测量数据)的格式，或用于计算参考数据的测量数据(即第二测量数据)的格式，包括以下至少一项：
对于原始信道信息，可以是信道信息的维度，例如某个符号上的频域信道响应息(一维信道信息)，多个符号上的时频域信道响应信息(二维信道信息)，多个符号、多个天线的时频空域信道响应信息(三维信道信息)，以及不同维度的规模，即采样点个数或采样间隔(例如时频域密度)；
对于谱信息，可以是限制不同谱信息的范围(或者称之为针对谱信息不同维度的截断窗口)，例如时延-多普勒谱中特定时延或多普勒范围对应的谱信息子集；也可以指示特定谱的径的个数的上限值或采样点个数的上限值；也可以是指示特定谱的最小粒度，即相邻两个采样点间的间隔(对应感知分辨率)；
上述谱信息子集中包括用于设备定位的部分谱信息和用于感知的部分谱信息，例如时延谱中第N2～N3个采样点或第N2～N3个径的信息，如图10中框线部分。
又例如，上述谱信息子集为时延-多普勒谱中多普勒绝对值小于X1，时延值小于X2的部分。如图11中框线部分。
对于基本测量量，可以是测量量的值的量化方式(例如量化粒度)，软信息类型(例如提供均值+方差，或置信区间和置信度)。
上述测量量格式指的可以是第一AI单元的推理和监控各自对应的测量量格式，例如两者对应的测量量不同，进而测量量格式也不同；也可以是AI单元的推理和监控共用的测量量格式，例如两者对应的测量量相同以及信号配置相同(第一信号和第二信号为同一信号)，进而对应的测量量格式也相同。
与监控关联的参考数据类型包括以下至少一项：
通过非AI的参数估计获取的感知数据；
通过传感器获取的感知数据；
通过特定感知模式获取的感知数据，例如单基地感知模式，基站自发自收感知模式；
通过特定感知设备或特定类型感知设备获取的感知数据，例如基站获取的感知数据，静止设备获取的感知数据；
通过能力更强的AI单元的推理得到感知数据。
上述上报配置为第二设备测量数据上报的准则，包括以下至少一项：
上报的时频域资源配置；
上报周期，可以是以感知的相干处理时间为上报周期，进行感知和定位相关的测量结果上报；
上报的触发条件，可以是预先设定的事件，包括不限于：
某类测量信号达到一定阈值的事件，例如当基于第二信号测量得到的感知性能指标超过预设门限时，上报所述第二测量数据用于AI单元的监控；
进入特定区域(例如小区)的事件；
到达特定时间的事件；
或者设备从先前位置移动超过一些预定义的(直线)距离的事件；
或者设备朝向改变超过一些预定义的角度的事件；
或者设备的运动速度超过一些预定义的速度阈值的事件；
或者设备传感器测量得到的环境信息变化(例如温度/湿度/光照强度)超过一定范围的事件。
需要注意的说明，上述第一指示信息中的各项可以是同一条信令发送的，也可以是分不同信令发送的。
步骤3、第二设备根据所述第一指示信息，执行测量流程，即接收其他设备发送的第一信号和第二信号，并测量得到第一测量数据和第二测量数据；
或者，第二设备自己发送第一信号/第二信号并接收回波信号进行测量；
或者，当第一设备为基站或终端时，所述其他设备可以是第一设备，即第一设备发送第一信号/第二信号，第二设备接收第一信号/第二信号并测量。
步骤4、第二设备向第一设备发送测量数据，所述测量数据包括以下至少一项：
测量结果，即上述测量量格式要求的测量量的值，所述测量量包括用于AI单元的推理的测量量或用于AI单元的监控的测量量，或者两者共用的测量量；
与测量结果关联的说明信息，包括以下至少一项：
时间戳信息，可以是绝对时间UTC，或相对于某个参考时间点的时间，例如无线帧号，半帧号，子帧号，TTI序号，slot序号，sub-slot序号，符号序号等；其中，所述slot序号或符号序号也可以是指相对于感知相干处理时间的起始slot或符号的序号；
测量结果对应的感知模式标识(包括背景技术中6种感知模式中的至少一项)；
表示测量结果为用于AI单元推理的测量结果，或用于监控的测量结果，或公共的测量结果的标识；
测量结果输入到的AI单元的标识，用于指示该测量结果为输入到哪一个AI单元的测量结果；
与监控关联的参考数据类型的标识；
性能指标，包括但不限于RSRP、RSRQ、SINR、感知性能指标，当感知性能指标超过预设门限时，认为该测量结果是可以用于AI单元的监控的测量结果。
步骤5、第一设备接收测量数据，并将对应的测量结果(用于第一AI单元的推理的测量结果)输入到相应的AI单元进行推理得到感知数据，具体可以为第一感知结果；
第一设备基于对应的测量结果(用于第一AI单元的监控的测量结果)通过非AI的方法(参数估计算法)得到参考数据，即为第二设备通过非AI的方法获取的参考数据，具体可以为第二感知结果。
第一设备根据所述感知数据和参考数据，得到监控结果信息，包括：
计算得到第一AI单元的感知精度信息，即与参考数据相比，第一AI单元输出的感知数据的误差信息或误差的统计信息，所述感知精度信息包括以下至少一项：
距离精度信息、速度精度信息、角度精度信息、定位精度信息、检测精度信息、识别精度信息。
其中，上述距离精度信息、速度精度信息、角度精度信息、定位精度信息、检测精度信息、识别精度信息参见上述实施例的相应说明，此处不作赘述。
上述监控结果信息还可以包括如下至少一项：
第一AI单元的有效性信息，即根据所述感知精度信息判断该AI单元是否有效，例如当N次感知结果计算中超过X次感知结果的误差信息大于预设门限时认为该AI单元性能不满足要求，即无效；或者对于目标定位，当给定时间内或者感知结果计算的次数达到预设值时所述位置坐标的均方根误差超过预设门限，则认为该AI单元性能不满足要求，即无效。
第一AI单元的性能等级信息，即预设不同的AI单元的性能等级，不同性能等级与不同的感知精度水平关联，根据所述感知精度信息判断该AI单元达到的预设的性能等级。
实施例二：
本实施例主要描述基于非AI的参数估计获取参考数据。
本实施例中，第一AI单元部署在第二设备侧，第二设备为基站或终端，第一设备为感知网络功能，或基站，或终端。第一设备从第二设备获取第一AI单元输出的第一测量数据，和用于通过参数估计的方法计算参考数据的第二测量数据。具体流程如图12所示。
步骤1、第一设备(以感知网络功能(SensingMF)为例)获取感知需求信息，具体同实施例一。
步骤2、第一设备向第二设备发送第二指示信息(为上述实施例中的目标信息中的全部或者部分内容)，用于指示第二设备获取测量数据，和/或，上报测量数据，即向第二设备请求AI单元的推理输出的第一测量数据，和用于监控的第二测量数据，所述第二指示信息包括以下至少一项：
信号配置信息、测量配置信息。
其中，信号配置信息具体同实施例一；
测量配置信息包括以下至少一项：
测量的信号资源指示，例如第一信号的标识或第二信号的标识；
测量用途或测量数据处理方法的指示，具体同实施例一；
测量量，包括与第一测量数据(第一AI单元输出的感知数据)关联的测量量，和与第二测量数据(用于AI单元的监控)关联的测量量，所述与第一测量数据关联的测量量为感知结果(具体见实施例一)，或基本测量量；所属与第二测量数据关联的测量量为基本测量量或谱信息，或感知结果。
测量量格式，具体同实施例一；
上报配置，具体同实施例一；
与监控关联的参考数据类型，具体同实施例一；
测量方式指示，包括以下至少一项：
采用AI处理或非AI处理的指示，即指示第二设备采用什么方法计算得到感知结果；
使用的AI单元的标识指示，即指示第二设备采用哪个AI单元计算得到感知结果；
使用的AI单元的精度信息或可以达到的性能等级或置信度水平。
步骤3、第二设备根据所述第二指示信息，执行测量流程(即对第一信号进行测量)得到第一中间测量结果，并对所述第一中间测量结果进行基于第一AI单元的进一步处理，得到感知数据，具体为感知结果或基本测量量的值。其中，所述第一中间测量结果为第一AI单元的输入数据，包括信道信息、谱信息、基本测量量中的至少一项，具体同实施例一。
第二设备根据所述第二指示信息，执行测量流程(即对第二信号进行测量)得到第二中间测量结果，即第二测量数据，包括基本测量量的值(时延值、多普勒值、角度值等)或谱信息。
可选地，第二设备基于所述第二中间测量结果，通过非AI的参数估计方法得到感知结果，即参考数据，例如目标的位置、个数等，或者，当所述第二中间测量结果为谱信息时，还可以是，第二设备基于所述第二中间测量结果，通过非AI的参数估计方法得到基本测量量。
其中，第一信号和第二信号可以是相同的信号，第一中间测量结果和第二中间测量结果可以是相同的，例如第二设备通过对同一信号进行测量得到谱信息，然后将所述谱信息输入到AI单元进行推理得到感知结果或基本测量量的值，即第一测量数据，然后通过非AI的参数估计方法对所述谱信息进行进一步处理得到基本测量量的值作为第二测量数据，或者再通过对所述基本测量量的值进行进一步计算得到感知结果作为第二测量数据。
步骤4、第二设备向第一设备上报所述第一测量数据和第二测量数据，包括以下至少一项：
测量结果，即所述测量量格式要求的测量量的值，包括通过AI单元的推理得到的测量结果，和用于AI单元的监控的测量结果；
与感知结果或定位结果相关的说明信息，包括以下至少一项：
时间戳信息；
测量结果对应的感知模式标识；
表示测量结果为AI单元的推理的测量结果的标识；
使用的AI单元的标识，用于指示该感知结果或定位结果为哪一个AI单元的推理结果；
表示测量结果为用于AI单元的监控的测量结果的标识；
与监控关联的参考数据类型的标识；
AI单元推理得到的测量结果的精度信息，或达到的性能等级或置信度水平；
性能指标，同实施例一。
步骤5、第一设备接收所述第一测量数据，并得到第一AI单元输出的感知数据。具体可以是第一测量数据包含感知数据(第二设备侧的第一AI单元输出的是感知数据)，所述感知数据即第一感知结果，或者，所述第一测量数据中包含基本测量量的值(第二设备侧第一AI单元输出的是基本测量量的值)，第一设备根据所述基本测量量的值计算得到第一感知结果；
第一设备接收所述第二测量数据，并得到参考数据，具体可以是第二感知结果，如第二测量数据包含基本测量量的值，第一设备基于基于所述基本测量量的值通过非AI的方法(参数估计算法)得到第二感知结果。可选地，所述第二测量数据包含的感知结果即为第二设备通过非AI的方法获取的第二感知结果。
第一设备根据所述第一感知结果和第二感知结果，得到监控结果信息，同实施例一。
步骤6、第一设备向第二设备发送所述监控结果信息，或发送第一AI单元后续操作指示信息，例如去激活，切换，微调，重训练等。
实施例三：
本实施例主要描述基于传感器获取参考数据。
本实施例中，第一AI单元部署在第二设备侧，或第一设备侧。当第一AI单元部署在第一设备侧时，具体流程可参考实施例一；当第一AI单元部署在第二设备侧时，具体流程可参考实施例二。
其中，第一设备向第二设备发送第三指示信息(可以上述实施例中的目标信息的全部或者部分信息)，用于指示第二设备获取测量数据，和/或，上报测量数据，即向第二设备请求第一AI单元输入(或输出)的第一测量数据，和用于第一AI单元的监控的第二测量数据，所述第三指示信息包括以下至少一项：
信号配置信息、测量配置信息、上报配置。
其中，信号配置信息包括第一信号的配置信息，所述第一信号为用于获取AI单元输入数据的信号；
测量配置信息包括以下至少一项：
测量的信号资源指示，例如第一信号标识；
测量量，所述测量量为第一AI单元推理相关联的测量量，包括第一AI单元输入的测量数据的对应的测量量(当第一AI单元在第一设备侧)，或第一AI单元输出的测量数据对应的测量量(当第一AI单元在第二设备侧)；
是否基于传感器进行测量的指示；
传感器类型指示，即指示第二设备基于哪个传感器获取用于第一AI单元的监控的感知数据，所述传感器类型包括以下至少一项：可见光摄像头、红外摄像头、GNSS、激光雷达、毫米波雷达、温度计、湿度计、气压计、陀螺仪、加速度计、磁力计、重力传感器、声呐、雨量计。
测量量格式，即第一AI单元输入的测量数据的格式(当第一AI单元在第一设备侧)，或第一AI单元输出的测量数据的格式(当第一AI单元在第二设备侧)；
还包括传感器的数据格式，例如二进制格式、ASCII格式等，也可以是与特定传感器类型关联的数据格式，例如对于图像传感器(摄像头)，可以是YUV、RGB等。又例如：
加速度计：测量施加到设备的加速度，格式为，包括沿x轴、y轴、z轴的加速度。进一步地，还可以分为包含重力加速度的结果、不包含重力加速度的结果、包含偏差补偿的结果、不包含偏差补偿的结果。
陀螺仪：测量围绕设备的x、y和z轴的旋转速率(弧度/秒)，与加速度计类似可以表示为三维矢量。
磁力计：监测地球磁场的变化，测量沿三个坐标轴中每个坐标轴的地磁场强度数据(以微特斯拉为单位)，格式为。通常无需直接使用此传感器，而是和其他传感器结合获取旋转角度信息。
旋转矢量传感器：通过不同传感器的组合可以得到旋转矢量传感器，获取终端角度，格式为(分别对应终端绕x、y、z轴旋转的角度，或者相对于East-North-Up(东北天)/North-East-Down(北东地)坐标轴的旋转角度)。
测量方式指示，具体同实施例一(当AI单元在第一设备侧)，或同实施例二(当AI单元在第二设备侧)。
上报配置具体同实施例一。
其中，第二设备根据所述第三指示信息执行测量流程，向第一设备上报测量数据，所述测量数据包括以下至少一项：
测量结果，即所述测量量格式要求的测量量的值，包括通过AI单元的推理得到的测量结果，和用于AI单元的监控的测量结果，即来自传感器输出的感知数据；
与测量结果相关的说明信息，包括以下至少一项：
与监控关联的参考数据类型的标识，即参考数据来自传感器的指示；
获取参考数据的传感器类型信息；
传感器输出精度信息；
坐标系关系，传感器对应坐标系与全局坐标系的关系，例如本地坐标系到全局坐标系的转换参数，即本地坐标系相对于全局坐标系的转动角度：α(轴承角)，β(下倾角)和γ(倾斜角)；
其他与测量结果相关的说明信息同实施例一(当第一AI单元在第一设备侧)或实施例二(当第一AI单元在第二设备侧)。
实施例四：
本实施例主要描述基于特定设备或特定感知模式获取参考数据的监控。
本实施例中，第一AI单元部署在第二设备侧，或第一设备侧。当第一AI单元部署在第一设备侧时，具体流程可参考实施例一；当第一AI单元部署在第二设备侧时，具体流程可参考实施例二。
其中，第一设备除了向第二设备请求第一AI单元输入(或输出)的第一测量数据，还包括，向第三设备请求用于第一AI单元的监控的第二测量数据，其中第三设备为满足特定条件，或者支持通过特定感知模式(包括背景技术中6种感知模式中的至少一项)获取感知数据的设备。
所述特定条件包括以下至少一项：
第三设备为静止设备；
第三设备为基站；
第三设备支持的发射功率、支持的天线个数或阵列孔径、可分配给感知信号的资源例如信号带宽等超过预设门限；
第三设备与第二设备两者之间的时钟同步误差(包括定时同步或频率同步)小于预设门限。
其中，第三设备和第一设备可以是相同的设备，例如第一设备基于与其他设备之间收发感知信号的感知模式获取用于第一AI单元的推理的测量数据，基于自发自收感知信号的感知模式获取用于第一AI单元的监控的测量数据；或者，第三设备和第二设备可以是相同的设备，例如第一设备接收第二设备发送的感知信号并进行测量获取用于第一AI单元的推理的测量数据，并发送给第二设备，第二设备基于自发自收感知信号的感知模式获取用于第一AI单元的监控的测量数据。
实施例五：
本实施例主要描述感知性能指标.
上述感知性能指标可以包括如下至少一项：
接收功率相关的感知指标；
干扰或噪声功率相关的感知指标；
与接收功率相关，以及还与干扰或噪声功率相关的感知指标。
上述接收功率相关的感知指标可以包括：第一指标，所述第一指标用于指示感知目标关联径的接收功率。
在一些实施方式中，上述第一指标可以是对目标信号测量得到的信道响应中与感知目标关联的径的接收功率在承载目标信号的资源单元上的线性平均值(单位为W)，该资源单元是时域或频域资源单元，这样通过线性平均值可以使得接收功率更加准确、可靠。需要说明的是，本申请实施例并不限定接收功率为线性平均值，例如：在一些实施方式中，也可以是取中位数接收功率、最低接收功率或最高接收功率。
其中，上述目标信号可以是上述实施例中的第一信号或第二信号，如用于感知业务的专用信号，或通信信号如参考信号，同步信号等。
上述干扰或噪声功率相关的感知指标包括如下至少一项：
第二指标，所述第二指标为第一线性平均值和第二线性平均值之和，所述第一线性平均值为目标资源上目标信号的信道响应中除感知目标关联的径之外的其他径的功率的线性平均值，所述第二线性平均值为第一资源上来自所述目标信号之外的其他信号的干扰或噪声功率的线性平均值；或者，所述第二指标等于总接收功率与上述第一指标的差值，所述总接收功率为所述第一设备在目标资源上的总接收功率；
第三指标，所述第三指标为第二资源上来自目标信号以外的其他信号的干扰或噪声功率的线性平均值，或者，所述第三指标等于总接收功率与所述目标信号的接收功率的差值，所述总接收功率为所述第一设备在目标资源上的总接收功率；
第四指标，所述第四指标为目标资源上所述目标信号的信道响应中除感知目标关联的径之外的其他径的功率的线性平均值；或者，所述第四指标等于所述目标信号的接收功率与上述第一指标的差值；
其中，所述第一指标用于指示所述目标信号的与感知目标关联的径的接收功率，所述目标资源为所述目标信号的传输资源，所述第一资源包括所述目标资源或除所述目标资源外的至少一个资源，所述第二资源包括所述目标资源或除目标资源外的至少一个资源。
上述其他径可以是目标信号中除上述感知目标关联的径的全部或者部分径。
上述目标信号之外的其他信号可以是指在第一资源上第一设备检测到除目标信号之外的全部或者部分信号。
上述第一资源包括所述目标资源或除所述目标资源外的至少一个资源是指第一资源包括如下至少一项：
目标资源、除所述目标资源外的至少一个资源。
上述第二资源包括所述目标资源或除所述目标资源外的至少一个资源是指第二资源包括如下至少一项：
目标资源、除所述目标资源外的至少一个资源。
其中，上述除目标资源外的至少一个资源可以是指第一设备需要检测或接收信号的资源中除目标资源之外的至少一个资源，如高层信令配置的资源或第一设备预先确定需要检测或接收信号的资源。
上述干扰或噪声功率包括干扰功率和噪声功率之和、干扰功率或者噪声功率。
上述第一设备在所述目标资源上的总接收功率可以包括在目标资源上服务小区和非服务小区的信号的接收功率、邻信道干扰功率和热噪声功率等。且上述总接收功率也可以是第一设备在所述目标资源上的总接收功率的线性平均值(单位为W)。
上述第一设备在所述第一资源上的接收信号强度指示(Received Signal Strength Indication，RSSI)对应的功率可以是总接收功率＝RSSI*K1，K1是系数，K1具体可以是协议约定或者网络侧配置。在一些实施方式中，上述RSSI对应的功率也可以是RSSI，即总接收功率＝RSSI。
上述目标信号的接收功率是指目标信号的参考信号接收功率(Reference Signal Received Power，RSRP)。
上述第二指标等于总接收功率与所述第一指标的差值可以表示为第二指标＝总接收功率-第一指标。
上述第三指标等于总接收功率与所述目标信号的接收功率的差值可以表示为第三指标＝总接收功率-目标信号接收功率。
上述第四指标等于所述目标信号的接收功率与所述第一指标的差值可以表示为第四指标＝目标信号的接收功率-第一指标。
在一些实施方式中，上述与接收功率相关，以及还与干扰或噪声功率相关的感知指标包括如下至少一项：
第五指标，所述第五指标等于上述第一指标除以所述第二指标得到的商；
第六指标，所述第六指标等于上述第一指标除以所述第三指标得到的商；
第七指标，所述第七指标等于上述第一指标除以所述第四指标得到的商；
第八指标，所述第八指标等于上述第一指标除以总接收功率得到的商与目标系数的乘积；
其中，所述总接收功率为所述第一设备在所述目标资源上的总接收功率。
其中，上述第一指标、第二指标、第三指标和第四指标参见上述实施方式，此处不作赘述。需要说明的是，在包括上述第五指标、第六指标、第七指标和第八指标中至少一项指标的情况下，本申请实施例中的感知相关的指标可以包括或者不包括上述第一指标、第二指标、第三指标和第四指标。
上述目标系数可以表示为K2，如第八指标＝K2*第一指标/总接收功率，K2为系数，K2具体可以是协议约定或者网络侧配置。
该实施方式中，通过上述第五指标、第六指标、第七指标或第八指标，可以实现在确定测量切换时考虑到接收功率以及干扰或噪声，以使得测量切换更加可靠。
在一些实施方式中，上述与接收功率相关，以及还与干扰或噪声功率相关的感知指标还可以是包括如下至少一项：
感知SINR相关的指标、感知SNR相关的指标、感知信号干扰比(Signal Interference Ratio，SIR)相关的指标、感知RSRQ相关的指标。
在一些实施方式中，与感知目标关联的径满足如下至少一项：
参数满足第一预设门限，或者，参数处于第一预设区间范围；
参数满足预设调制规则；
与首达径的参数差满足第二预设门限，或者，与首达径的参数差处于第二预设区间范围；
与参考径的参数差满足第三预设门限，或者，与参考径的参数差处于第三预设区间范围。
其中，上述参数可以包括如下至少一项：
幅度、功率、强度、能量、相位、多普勒、时延、角度；
上述参数差可以包括如下至少一项：
幅度差、功率差、强度差、能量差、相位差、多普勒差、时延差、角度差。
第一预设门限、第一预设区间范围、第二预设门限、第二预设区间范围、第三预设门限、第三预设区间范围可以是协议约定或者网络侧配置的，或者，这些预设门限或预设区间范围是接收设备根据感知先验信息或感知需求确定的。上述参数满足第一预设门限可以是参数超过或者等于第一预设门限，上述与首达径的参数差满足第二预设门限可以是与首达径的参数差超过或者等于第二预设门限，上述与参考径的参数差满足第三预设门限可以是与参考径的参数差超过或者等于第三预设门限。
例如：感知业务为动目标检测，则需要检测多普勒大于零的径作为感知目标关联的径；或者对于交通场景感知目标为车，默认车速为40km/h～120km/h，则检测对应的速度范围内(多普勒范围内)的径作为感知目标关联的径；或者感知目标区域与感知信号收发设备的距离需要满足特定要求，则检测对应时延范围内的径作为感知目标关联的径；又或者如果感知业务是呼吸监测，则可以根据人的性别、年龄来判断对应的正常呼吸频率(例如，15～30次/分钟，可以作为感知先验信息，对应可以计算出多普勒范围，0.25～0.5Hz)。
上述首达径可以是视距(Line-of-Sight，LOS)径，具体为目标信号中最先到达接收端的径。上述参考径可以是经过已知目标反射的径，如经过智能超表面(Reconfigurable Intelligence Surface，RIS)、反向散射(Backscatter)或其他已知的无源目标等反射的径。
上述预设调制规则可以是协议约定或者网络侧配置的。特定调制规则为标签(Tag)或Backscatter设备或RIS的调制规则，即感知目标关联的径可以是经过Tag或Backscatter设备或RIS调制并反射的径。
在上述一种可选的实施方式中可以实现通过多种方式确定感知目标关联的径，既可以提高确定感知目标关联的径的灵活性，又可以基于多种方式共同确定以提高确定感知目标关联的径的准确性。
在一些实施方式中，在确定感知目标关联的径之前，还可以确定径集合，该径集合包括幅度、功率、强度或能量超过一定门限的径，如图13所示，径集合包括径0，1，2和3。再在径集合中基于上述至少一项确定感知目标关联的径，以降低计算量。
下面通过一个实施例对本申请实施例的指标计算进行举例说明，需要说明的是，本申请实施例中对各指标的计算不限定，下面实施例仅是一个举例说明。
第一指标的计算方式1如下：
第一设备(如终端)基于发送的目标信号X(k)和目标信号对应的接收信号Y(k)进行信道估计得到信道响应(Channel Response)H(k)＝Y(k)/X(k)，其中k＝0,1,2,…,K-1表示资源单元索引，K为资源单元个数。第一设备获取信道响应H(k)后，将其变换到第一维度，在第一维度中确定感知目标关联的径。然后计算感知目标关联的径的功率作为第一指标，若感知目标关联的径包括多条径，则计算多条径的功率之和作为第一指标。
其中，所述第一维度包括以下之一：
时延维；
多普勒维；
方位角维；
俯仰角维；
时延维、多普勒维、方位角维和俯仰角维中至少两项联合的维度。例如，时延-多普勒维，时延-多普勒-角度维度等；
例如，H(f)为信道响应，其中f＝0,1,2,…,N-1表示频域采样点(例如子载波索引)，则通过对H(f)进行逆傅里叶变换可以将其变换到时延维度(第一维度)；又例如，H(f,t)为信道响应，其中f＝0,1,2,…,N-1表示频域采样点(例如子载波索引)，t＝0,1,2,…,M-1表示时域采样点(例如OFDM符号索引)，则通过对H(f,t)进行沿频域维度的逆傅里叶变换和沿时域维度的傅里叶变换可以将其变换到时延-多普勒维度(第一维度)；又例如，H(f,t,s)为信道响应，其中f＝0,1,2,…,N-1表示频域采样点(例如子载波索引)，t＝0,1,2,…,M-1表示时域采样点(例如OFDM符号索引)，s＝0,1,2,…,P-1表示空域采样点(天线索引或端口索引)，则通过对H(f,t,s)进行沿频域维度的逆傅里叶变换和沿时域维度的傅里叶变换和沿天线域维度的傅里叶变换可以将其变换到时延-多普勒-角度维度(第一维度)。
对目标信号测量得到的信道响应中与感知目标关联的径(简称为感知径)的确定方法：
确定径集合。径集合中的径包括信道响应变换到第一维度后，全部径中幅度、功率、强度或能量超过一定门限的径。例如图6中，径0，1，2，3为径集合的径；一定门限可以设置为高于噪声门限或者高于噪声干扰门限，或者协议约定。其中，这一步(确定径集合)是可选的，可以只根据下一步来确定与感知目标关联的径。
所述径集合中或者从目标信号的所有径中选择满足第一条件的径，作为与感知目标关联的径。第一条件包括以下至少一项：
径的幅度、功率、强度或能量超过预设门限或位于预设区间范围，如预设门限为超过噪声门限的5倍；
径的多普勒超过预设门限或位于预设区间范围；
径的时延超过预设门限或位于预设区间范围；
径的角度超过预设门限或位于预设区间范围；
径与首达径(例如LOS径)或参考径的幅度/功率/强度/能量的差超过预设门限或位于预设区间范围，其中，参考径可以是经过已知目标(例如RIS/Backscatter/其他已知的无源目标等)反射的径；
径与首达径(例如LOS径)或参考径的多普勒差超过预设门限或位于预设区间范围；
径与首达径(例如LOS径)或参考径的时延差超过预设门限或位于预设区间范围；
径与首达径(例如LOS径)或参考径的角度差超过预设门限或位于预设区间范围；
径的幅度、功率、强度、能量或相位满足特定调制规则，特定调制规则为Tag/Backscatter设备或RIS的调制规则，即感知目标关联的径可以是经过Tag/Backscatter设备或RIS调制并反射的径
其中，上述各项第一条件还可以根据一段时间统计的结果；例如，在预设时间窗上述指标(例如径的多普勒，径的时延等)超过预设门限或位于预设区间范围的比例达到预设比例，或者是在预设时间窗上述指标(例如径的多普勒，径的时延等)超过预设门限或位于预设区间范围的次数达到预设次数；
其中，预设门限或设定区间范围是其他设备发送给接收设备的，其他设备根据感知先验信息或感知需求确定的。或者，预设门限或预设区间范围可以是协议约定，或者预设门限或预设区间范围是接收设备根据感知先验信息或感知需求确定的。
其中，感知先验信息或感知需求包括如下信息：
感知业务或感知业务类型，所述感知业务可以是例如检测目标是否存在，定位，速度探测，距离探测、角度探测、加速度探测，材料分析，成分分析，形状检测，类别划分，雷达散射截面积(Radar Cross Section，RCS)检测，极化散射特性检测，跌倒检测，入侵检测，数量统计，室内定位，手势识别，唇语识别，步态识别，表情识别，面部识别，呼吸监测，心率监测，脉搏监测，湿度/亮度/温度/大气压强监测，空气质量监测，天气情况监测，环境重构，地形地貌、建筑/植被分布检测，人流量或车流量检测，人群密度、车辆密度检测等；所述感知业务类型可以是按照一定特征把多个不同的感知业务进行分类，例如按照功能划分为检测类感知业务(例如包括入侵检测、跌倒检测)、参数估计类感知业务(距离、角度、速度计算)、识别类感知业务(动作识别、身份识别)等，还可以是按照感知的范围(近距离感知、中距离感知、远距离感知)划分，按照感知的精细程度划分(粗粒度感知、精细力度感知等)，按照功耗/能耗划分，按照资源占用划分等。如果感知业务是呼吸监测，则可以根据人的性别、年龄来判断对应的正常呼吸频率(例如，男性：13～21次/分钟，女性15～20次/分钟；成人：12～20次/分钟，儿童：约30～40次/分钟)，可以作为感知先验信息；
感知目标区域：是指感知对象的位置区域，或者，需要进行成像或环境重构的位置区域；例如，根据感知对象的大概位置/距离确定感知目标关联的径的时延的预设区间范围；
感知对象类型：针对感知对象可能的运动特性对感知对象进行分类，每个感知对象类型中包含了典型感知对象的运动速度范围、运动加速度范围、典型RCS范围等信息；
感知的目标个数；例如，摄像头感知结果作为一种感知先验信息，可以得到感知的目标个数。
例如图13中，径0，1，2，3为径集合中的径，其中径2,3为满足第一条件(例如其时延满足预设门限)的感知径，径0，1为其他散射体关联的径。
其中图13中，信道响应在第一维度(时延维，多普勒维，方位角维，或俯仰角维)中的多径示意图，其中，横轴为第一维度，纵轴为归一化的幅度、功率、强度或能量。
对于频率范围(frequency range)1,第一指标的参考点(reference point)可以是接收设备如终端的天线连接器(antenna connector)。对于frequency range 1,如果接收设备有多个接收通道，则接收设备测量并上报的第一指标不能低于任意一个单接收通道的指标。对于frequency range 2，某个接收通道测得的第一指标需要对该接收通道对应的多个天线单元上的合并信号进行测量得到。
第一指标的计算方式2可以如下：
计算感知目标关联的径的接收功率时，还可以是第一维度中感知目标关联的径的功率与的差值作为第一指标，其中N1表示感知目标关联的径的个数。为第一维度中径集合之外的多条径的平均功率。
目标信号的接收功率的计算方式1可以如下：
目标信号的接收功率可以是接收设备获得信道响应(Channel Response)H(k)后，将其变换到第一维度，在第一维度中确定径集合，然后计算所述径集合中的全部径的功率和。
目标信号的接收功率的计算方式2可以如下：
目标信号的接收功率还可以是第一维度中径集合中的全部径的功率和与的差值，其中N2表示径集合中的径的个数。
总接收功率的计算方式：
总接收功率
其中，Y(k)为目标信号对应的接收信号，k＝0,1,2,…,K-1表示资源单元索引，K为资源单元个数。
第二指标的计算方式可以如下：
将信道响应H(k)经过第一滤波处理得到Hfilter1(k)，然后根据Hfilter1(k)和目标信号X(k)计算得到第一滤波处理后的接收信号Yfilter1(k)，即Yfilter1(k)＝Hfilter1(k)X(k)。然后将接收信号Y(k)减去第一滤波处理后的接收信号Yfilter1(k)从而得到干扰和噪声信号Yσ1(k)，即Yσ1(k)＝Y(k)-Yfilter1(k)，然后计算得到第二指标：
其中，所述第一滤波处理用于消除第一维度上的噪声和干扰以及非感知目标关联的径，例如，第一滤波处理将图13中除感知目标关联的径以外的其他径的幅度、功率、强度或能量置零。经过第一滤波处理后的信道响应Hfilter1(k)中不包含噪声和干扰以及非感知目标关联的径，仅包含感知目标关联的径。
第三指标的计算方式1可以如下：
将信道响应H(k)经过第二滤波处理得到Hfilter2(k)，然后根据Hfilter2(k)和目标信号X(k)计算得到第二滤波处理后的接收信号Yfilter2(k)，即Yfilter2(k)＝Hfilter2(k)X(k)。然后将接收信号Y(k)减去第二滤波处理后的接收信号Yfilter2(k)从而得到干扰和噪声信号Yσ2(k)，即Yσ2(k)＝Y(k)-Yfilter2(k)，然后计算得到第三指标：
上述第二滤波处理可以是第一维度上的噪声干扰抑制处理(例如图6中除径集合外的其他径的幅度、功率、强度或能量置零)，或者最小均方误差(Minimum Mean Squared Error，MMSE)滤波。经过第二滤波处理后的信道响应Hfilter2(k)中不包含噪声和干扰，仅包含径集合中的径。
第三指标的计算方式2可以如下：
根据第一维度中径集合之外的多条径的平均功率计算得到第三指标Pσ2，即 其中N表示第一维度采样点个数。
需要说明的是，如果接收设备判断出多个感知目标，或者接收设备根据感知先验信息或感知需求得到感知目标的数目，则有以下几种方法：
方法1：分别计算每个感知目标的感知相关的指标(也可以称作目标指标)。例如在图4中分别确定关联到每个感知目标的径，然后分别计算每个感知目标对应的各项感知相关的指标；此时计算某一感知目标(如感知目标A)对应的第二指标时，有两种方法：即：感知目标A的第二指标＝总接收功率-感知目标A的第一指标；或者，感知目标A的第二指标＝总接收功率-感知目标A的第一指标-感知目标B的第一指标；(假设一共有两个感知目标：A和B)；类似的，第四指标的计算方式也有两种：感知目标A的第四指标＝目标信号的RSRP-感知目标A的第一指标；或者，感知目标A的第四指标＝目标信号的RSRP-感知目标A的第一指标-感知目标B的第一指标；(假设一共有两个感知目标：A和B)
方法2：针对多个感知目标计算一个感知相关的指标。例如在图6中确定关联到任一感知目标的径，然后将这些径都确定为与感知目标关联的径；相当于将多个感知目标视为一个虚拟的感知目标，然后计算该虚拟的感知目标对应的感知相关的指标。
需要说明的是，上述计算方式仅是举例说明，本申请实施例对指标的具体计算方式不作限定。
本申请实施例提供的AI单元监控方法，执行主体可以为AI单元监控装置。本申请实施例中以AI单元监控装置执行AI单元监控方法为例，说明本申请实施例提供的AI单元监控装置。
本申请实施例提供一种AI单元监控装置，作为一种示例，AI单元监控装置可以是通信设备或通信设备中的部件，例如芯片。该通信设备可以是终端、网络侧设备或服务器等。示例性的，终端可以包括但不限于上述所列举的终端11的类型，网络侧设备可以包括但不限于上述所列举的网络侧设备12的类型，本申请实施例不作具体限定。
AI单元监控装置可以包括接收模块、发送模块和处理模块。其中，接收模块、发送模块和处理模块可以是通过软件实现，也可以通过硬件实现。当通过硬件实现时，处理模块可以由处理器实现，示例性的，处理器可以包括通用处理器、专用处理器等，例如包括中央处理单元(Central Processing Unit，CPU)、微处理器、数字信号处理器(Digital Signal Processor，DSP)、人工智能(Artificial Intelligent，AI)处理器、图形处理器(Graphics Processing Unit，GPU)、专用集成电路(Application Specific Integrated Circuit，ASIC)、网络处理器(Network Processor，NP)、现场可编程门阵列(Field Programmable Gate Array，FPGA)或者其他可编程逻辑器件、门电路、晶体管、分立硬件组件等。接收模块和发送模块可以由通信接口实现，通信接口可以包括收发器、管脚、电路、总线、射频单元等其中一种或多种。
具体的，参见图14，当AI单元监控装置为终端或者网络侧设备，或者终端或者网络侧设备中的部件时，AI单元监控装置1400包括：
处理模块1401，用于获取第一AI单元输出的感知数据；
所述处理模块，还用于获取用于监控所述第一AI单元的参考数据；
所述处理模块，还用于基于所述参考数据和所述第一AI单元输出的感知数据，确定所述第一AI单元的监控结果。
可选地，所述参考数据包括如下至少一项：
通过非AI的方式获取的感知数据；
第二AI单元输出的感知数据，所述第二AI单元的处理能力强于所述第一AI单元的处理能力。
可选地，所述通过非AI的方式获取的感知数据包括如下至少一项：
通过估计算法获取的感知数据；
通过传感器获取的感知数据；
通过目标感知模式获取的感知数据；
通过目标感知设备获取的感知数据。
可选地，获取第一AI单元输出的感知数据包括：
接收第一测量数据；
其中，所述第一测量数据用于输入到所述第一AI单元进行推理得到所述第一AI单元输出的感知数据，且所述第一测量数据包括第一测量结果；或者
所述第一测量数据包括所述第一AI单元输出的感知数据。
可选地，所述第一测量数据还包括：第一说明信息，所述第一说明信息包括如下至少一项：
所述第一测量数据的时间戳信息、所述第一测量数据的感知模式信息、用于指示所述第一测量结果用于AI单元推理的指示信息、所述第一测量数据对应的AI单元的信息、所述第一测量数据的性能指标信息、用于指示所述第一测量数据包括第一AI单元输出的感知数据的指示信息、所述第一测量数据关联的参考数据类型信息、所述第一测量数据对应的精度信息、所述第一测量数据对应的性能等级信息、所述第一测量数据对应的置信度水平信息。
可选地，获取用于监控所述第一AI单元的参考数据包括：
接收第二测量数据；
其中，所述第二测量数据用于确定所述参考数据，且所述第二测量数据包括第二测量结果；或者，
所述第二测量数据包括所述参考数据。
可选地，所述第二测量数据还包括第二说明信息，所述第二说明信息包括如下至少一项：
所述第二测量数据的时间戳信息、所述第二测量数据的感知模式信息、用于所述第二测量结果用于监控的指示信息、所述第二测量数据关联的参考数据类型信息、所述第二测量数据的性能指标信息、用于指示所述第二测量数据包括所述参考数据的指示信息、所述第二测量数据对应的AI单元的信息、所述第二测量数据对应的精度信息、所述第二测量数据的坐标系信息、所述第二测量数据对应的传感器类型信息。
可选地，所述监控结果包括如下至少一项：
有效性信息、性能等级信息、感知精度信息；
其中，所述感知精度信息包括如下至少一项：
距离精度信息、速度精度信息、角度精度信息、定位精度信息、检测精度信息、识别精度信息。
可选地，所述装置还包括：
发送模块，用于发送目标信息，所述目标信息包括如下至少一项：信号配置信息、测量配置信息、指示信息；其中，所述指示信息用于指示如下至少一项：
获取第一测量数据、获取第二测量数据、上报第一测量数据、上报第二测量数据；
其中，所述第一测量数据用于输入到所述第一AI单元进行推理得到所述第一AI单元输出的感知数据，且所述第一测量数据包括第一测量结果；
所述第二测量数据用于确定所述参考数据，且所述第二测量数据包括第二测量结果。
可选地，所述信号配置信息包括如下至少一项：
第一信号的配置信息、第二信号的配置信息，所述第一信号为用于获取所述第一AI单元的输入数据的信号，所述第二信号为用于获取所述参考数据的信号，所述第一信号与所述第二信号满足如下至少一项：
所述第二信号的频域资源长度大于所述第一信号的频域资源长度；
所述第二信号的频域资源间隔小于所述第一信号的频域资源间隔；
所述第二信号的时域资源长度大于所述第一信号的时域资源长度；
所述第二信号的时域资源单元小于所述第一信号的时域资源间隔；
所述第二信号的发射功率大于所述第一信号的发射功率；
所述第一信号为所述第二信号的子集。
可选地，所述测量配置信息包括如下至少一项：
测量的信号资源指示、所述第一测量数据对应的测量用途信息、所述第二测量数据对应的测量用途信息、所述第一测量数据的处理方式信息、所述第二测量数据的处理方式信息、所述第一测量数据对应的AI单元的信息、所述第一测量数据的测量量、所述第二测量数据的测量量、所述第一测量数据的格式信息、所述第二测量数据的格式信息、所述参考数据的类型信息、上报配置、所述第一测量数据是否基于传感器进行测量的指示、所述第二测量数据是否基于传感器进行测量的指示、所述第一测量数据关联的传感器的类型信息、所述第二测量数据关联的传感器的类型信息、所述第一测量数据的测量方式、所述第二测量数据的测量方式。
可选地，所述第一测量数据的测量量包括如下至少一项：
接收信号、信道信息、谱信息、基本测量量；
所述第二测量数据的测量量包括如下至少一项：
接收信号、信道信息、谱信息、基本测量量。
可选地，所述第一测量数据的格式信息用于指示如下至少一项：
信道信息的维度、谱信息的范围、基本测量量的类型、传感器数据的格式；
所述第二测量数据的格式信息用于指示如下至少一项：
信道信息的维度、谱信息的范围、基本测量量的类型。
可选地，所述第一设备发送目标信息包括如下至少一项：
所述第一设备向第二设备发送所述目标信息的全部或者部分；
所述第一设备向第三设备发送所述目标信息的全部或者部分；
其中，所述第二设备为用于提供第一测量数据和第二测量数据中的至少一项的设备，所述第三设备为用于提供第二测量数据的设备。
可选地，所述第二设备满足如下至少一项条件：
支持目标感知模式；
位置固定；
支持的发射功率超过预设门限；
支持的天线个数超过预设门限；
支持的阵列孔径超过预设门限；
可分配给用于感知的资源量超过预设门限；
与所述第三设备之间的时钟同步误差小于预设门限；
或，所述第三设备满足如下至少一项条件：
支持目标感知模式；
位置固定；
支持的发射功率超过预设门限；
支持的天线个数超过预设门限；
支持的阵列孔径超过预设门限；
可分配给用于感知的资源量超过预设门限；
与所述第二设备之间的时钟同步误差小于预设门限。
可选地，处理模块1403还用于基于所述监控结果确定针对所述第一AI单元的目标操作，所述目标操作包括如下至少一项：
去激活、切换、微调、重训练。
上述AI单元监控装置有利于提升或者保证感知性能。
本申请实施例提供的信号监听装置能够实现图6的方法实施例实现的各个过程，并达到相同的技术效果，为避免重复，这里不再赘述。
具体的，参见图15，当AI单元监控装置为终端或者网络侧设备，或者终端或者网络侧设备中的部件时，AI单元监控装置1500包括：
发送模块1501，用于发送数据，所述发送数据包括如下至少一项：
发送第一测量数据、发送第二测量数据；
其中，所述第一测量数据用于输入到第一AI单元进行推理得到所述第一AI单元输出的感知数据，且所述第一测量数据包括第一测量结果；或者，所述第一测量数据包括所述第一AI单元输出的感知数据；
所述第二测量数据用于确定参考信号，且所述第二测量数据包括第二测量结果；或者，所述第二测量数据包括参考数据，所述参考数据用于监控所述第一AI单元。
可选地，所述参考数据包括如下至少一项：
通过非AI的方式获取的感知数据；
第二AI单元输出的感知数据，所述第二AI单元的处理能力强于所述第一AI单元的处理能力。
可选地，所述通过非AI的方式获取的感知数据包括如下至少一项：
通过估计算法获取的感知数据；
通过传感器获取的感知数据；
通过目标感知模式获取的感知数据；
通过目标感知设备获取的感知数据。
可选地，所述第一测量数据还包括：第一说明信息，所述第一说明信息包括如下至少一项：
所述第一测量数据的时间戳信息、所述第一测量数据的感知模式信息、用于指示所述第一测量结果用于AI单元推理的指示信息、所述第一测量数据对应的AI单元的信息、所述第一测量数据的性能指标信息、用于指示所述第一测量数据包括第一AI单元输出的感知数据的指示信息、所述第一测量数据关联的参考数据类型信息、所述第一测量数据对应的精度信息、所述第一测量数据对应的性能等级信息、所述第一测量数据对应的置信度水平信息。
可选地，所述第二测量数据还包括第二说明信息，所述第二说明信息包括如下至少一项：
所述第二测量数据的时间戳信息、所述第二测量数据的感知模式信息、用于所述第二测量结果用于监控的指示信息、所述第二测量数据关联的参考数据类型信息、所述第二测量数据的性能指标信息、用于指示所述第二测量数据包括所述参考数据的指示信息、所述第二测量数据对应的AI单元的信息、所述第二测量数据对应的精度信息、所述第二测量数据的坐标系信息、所述第二测量数据对应的传感器类型信息。
可选地，所述装置还包括：
接收模块，用于接收目标信息，所述目标信息包括如下至少一项：信号配置信息、测量配置信息、指示信息；其中，所述指示信息用于指示如下至少一项：
获取所述第一测量数据、获取所述第二测量数据、上报所述第一测量数据、上报所述第二测量数据。
可选地，所述信号配置信息包括如下至少一项：
第一信号的配置信息、第二信号的配置信息，所述第一信号为用于获取所述第一AI单元的输入数据的信号，所述第二信号为用于获取所述参考数据的信号，所述第一信号与所述第二信号满足如下至少一项：
所述第二信号的频域资源长度大于所述第一信号的频域资源长度；
所述第二信号的频域资源间隔小于所述第一信号的频域资源间隔；
所述第二信号的时域资源长度大于所述第一信号的时域资源长度；
所述第二信号的时域资源单元小于所述第一信号的时域资源间隔；
所述第二信号的发射功率大于所述第一信号的发射功率；
所述第一信号为所述第二信号的子集。
可选地，所述测量配置信息包括如下至少一项：
测量的信号资源指示、所述第一测量数据对应的测量用途信息、所述第二测量数据对应的测量用途信息、所述第一测量数据的处理方式信息、所述第二测量数据的处理方式信息、所述第一测量数据对应的AI单元的信息、所述第一测量数据的测量量、所述第二测量数据的测量量、所述第一测量数据的格式信息、所述第二测量数据的格式信息、所述参考数据的类型信息、上报配置、所述第一测量数据是否基于传感器进行测量的指示、所述第二测量数据是否基于传感器进行测量的指示、所述第一测量数据关联的传感器的类型信息、所述第二测量数据关联的传感器的类型信息、所述第一测量数据的测量方式、所述第二测量数据的测量方式。
可选地，所述第一测量数据的测量量包括如下至少一项：
接收信号、信道信息、谱信息、基本测量量；
所述第二测量数据的测量量包括如下至少一项：
接收信号、信道信息、谱信息、基本测量量。
可选地，所述第一测量数据的格式信息用于指示如下至少一项：
信道信息的维度、谱信息的范围、基本测量量的类型、传感器数据的格式；
所述第二测量数据的格式信息用于指示如下至少一项：
信道信息的维度、谱信息的范围、基本测量量的类型。
可选地，所述装置还包括：
处理模块，用于基于所述目标信息执行目标操作，所述目标操作包括如下至少一项：
执行第一测量，得到所述第一测量结果；
执行第二测量，得到第二测量结果；
执行第三测量，并将所述第三测量的测量结果输入到所述第一AI单元进行推理，得到所述第一AI单元输出的感知数据；
执行第四测量，并基于所述第四测量的测量结果确定所述参考数据。
可选地，所述第二设备满足如下至少一项条件：
支持目标感知模式；
位置固定；
支持的发射功率超过预设门限；
支持的天线个数超过预设门限；
支持的阵列孔径超过预设门限；
可分配给用于感知的资源量超过预设门限；
与第三设备之间的时钟同步误差小于预设门限，所述第三设备为用于提供第二测量数据的设备。
上述AI单元监控装置有利于提升或者保证感知性能。
本申请实施例提供的信号监听装置能够实现图8的方法实施例实现的各个过程，并达到相同的技术效果，为避免重复，这里不再赘述。
如图16所示，本申请实施例还提供一种通信设备1600，包括处理器1601和存储器1602，存储器1602上存储有可在所述处理器1601上运行的程序或指令，例如，该通信设备1600为第一设备时，该程序或指令被处理器1601执行时实现上述第一设备侧的AI单元监控方法实施例的各个步骤，且能达到相同的技术效果。该通信设备1600为第二设备时，该程序或指令被处理器1601执行时实现上述第二设备侧的AI单元监控方法实施例的各个步骤，且能达到相同的技术效果，为避免重复，这里不再赘述。
本申请实施例还提供一种设备，该设备为第一设备，包括处理器和通信接口，所述通信接口和所述处理器耦合，所述处理器用于运行程序或指令，实现如图6所示方法实施例中的步骤。该设备实施例与上述第一设备侧方法实施例对应，上述方法实施例的各个实施过程和实现方式均可适用于该终端实施例中，且能达到相同的技术效果。该设备可以是图14所示的AI单元监控装置。
本申请实施例还提供一种设备，包括处理器及通信接口，其中，所述处理器或者所述通信接口用于获取第一AI单元输出的感知数据；获取用于监控所述第一AI单元的参考数据；所述处理器用于基于所述参考数据和所述第一AI单元输出的感知数据，确定所述第一AI单元的监控结果。
具体地，图17为实现本申请实施例的一种设备的硬件结构示意图。
该设备1700包括但不限于：射频单元1701、网络模块1702、音频输出单元1703、输入单元1704、传感器1705、显示单元1706、用户输入单元1707、接口单元1708、存储器1709以及处理器1710等中的至少部分部件。
本领域技术人员可以理解，设备1700还可以包括给各个部件供电的电源(比如电池)，电源可以通过电源管理系统与处理器1710逻辑相连，从而通过电源管理系统实现管理充电、放电以及功耗管理等功能。图17中示出的设备结构并不构成对设备的限定，设备可以包括比图示更多或更少的部件，或者组合某些部件，或者不同的部件布置，在此不再赘述。
应理解的是，本申请实施例中，输入单元1704可以包括图形处理器17041和麦克风17042，图形处理器17041对在视频捕获模式或图像捕获模式中由图像捕获装置(如摄像头)获得的静态图片或视频的图像数据进行处理。显示单元1706可包括显示面板17061，可以采用液晶显示器、有机发光二极管等形式来配置显示面板17061。用户输入单元1707包括触控面板17071以及其他输入设备17072中的至少一种。触控面板17071，也称为触摸屏。触控面板17071可包括触摸检测装置和触摸控制器两个部分。其他输入设备17072可以包括但不限于物理键盘、功能键(比如音量控制按键、开关按键等)、轨迹球、鼠标、操作杆，在此不再赘述。
本申请实施例中，射频单元1701接收来自网络侧设备的下行数据后，可以传输给处理器1710进行处理；另外，射频单元1701可以向网络侧设备发送上行数据。通常，射频单元1701包括但不限于天线、放大器、收发器、耦合器、低噪声放大器、双工器等。
存储器1709可用于存储软件程序或指令以及各种数据。存储器1709可主要包括存储程序或指令的第一存储区和存储数据的第二存储区，其中，第一存储区可存储操作系统、至少一个功能所需的应用程序或指令(比如声音播放功能、图像播放功能等)等。此外，存储器1709可以包括易失性存储器或非易失性存储器。其中，非易失性存储器可以是只读存储器(Read-Only Memory，ROM)、可编程只读存储器(Programmable ROM，PROM)、可擦除可编程只读存储器(Erasable PROM，EPROM)、电可擦除可编程只读存储器(Electrically EPROM，EEPROM)或闪存。易失性存储器可以是随机存取存储器(Random Access Memory，RAM)，静态随机存取存储器(Static RAM，SRAM)、动态随机存取存储器(Dynamic RAM，DRAM)、同步动态随机存取存储器(Synchronous DRAM，SDRAM)、双倍数据速率同步动态随机存取存储器(Double Data Rate SDRAM，DDRSDRAM)、增强型同步动态随机存取存储器(Enhanced SDRAM，ESDRAM)、同步连接动态随机存取存储器(Synch link DRAM，SLDRAM)和直接内存总线随机存取存储器(Direct Rambus RAM，DRRAM)。本申请实施例中的存储器1709包括但不限于这些和任意其它适合类型的存储器。
处理器1710可包括一个或多个处理单元；可选的，处理器1710集成应用处理器和调制解调处理器，其中，应用处理器主要处理涉及操作系统、用户界面和应用程序等的操作，调制解调处理器主要处理无线通信信号，如基带处理器。可以理解的是，上述调制解调处理器也可以不集成到处理器1710中。
该实施例中，以上述设备为第一设备，第一设备为终端进行举例说明。
其中，射频单元1701或者处理器1710，用于获取第一AI单元输出的感知数据；
射频单元1701或者处理器1710，用于获取用于监控所述第一AI单元的参考数据；
处理器1710，用于基于所述参考数据和所述第一AI单元输出的感知数据，确定所述第一AI单元的监控结果。
可选地，所述参考数据包括如下至少一项：
通过非AI的方式获取的感知数据；
第二AI单元输出的感知数据，所述第二AI单元的处理能力强于所述第一AI单元的处理能力。
可选地，所述通过非AI的方式获取的感知数据包括如下至少一项：
通过估计算法获取的感知数据；
通过传感器获取的感知数据；
通过目标感知模式获取的感知数据；
通过目标感知设备获取的感知数据。
可选地，所述获取第一AI单元输出的感知数据包括：
接收第一测量数据；
其中，所述第一测量数据用于输入到所述第一AI单元进行推理得到所述第一AI单元输出的感知数据，且所述第一测量数据包括第一测量结果；或者
所述第一测量数据包括所述第一AI单元输出的感知数据。
可选地，所述第一测量数据还包括：第一说明信息，所述第一说明信息包括如下至少一项：
所述第一测量数据的时间戳信息、所述第一测量数据的感知模式信息、用于指示所述第一测量结果用于AI单元推理的指示信息、所述第一测量数据对应的AI单元的信息、所述第一测量数据的性能指标信息、用于指示所述第一测量数据包括第一AI单元输出的感知数据的指示信息、所述第一测量数据关联的参考数据类型信息、所述第一测量数据对应的精度信息、所述第一测量数据对应的性能等级信息、所述第一测量数据对应的置信度水平信息。
可选地，所述获取用于监控所述第一AI单元的参考数据包括：
接收第二测量数据；
其中，所述第二测量数据用于确定所述参考数据，且所述第二测量数据包括第二测量结果；或者，
所述第二测量数据包括所述参考数据。
可选地，所述第二测量数据还包括第二说明信息，所述第二说明信息包括如下至少一项：
所述第二测量数据的时间戳信息、所述第二测量数据的感知模式信息、用于所述第二测量结果用于监控的指示信息、所述第二测量数据关联的参考数据类型信息、所述第二测量数据的性能指标信息、用于指示所述第二测量数据包括所述参考数据的指示信息、所述第二测量数据对应的AI单元的信息、所述第二测量数据对应的精度信息、所述第二测量数据的坐标系信息、所述第二测量数据对应的传感器类型信息。
可选地，所述监控结果包括如下至少一项：
有效性信息、性能等级信息、感知精度信息；
其中，所述感知精度信息包括如下至少一项：
距离精度信息、速度精度信息、角度精度信息、定位精度信息、检测精度信息、识别精度信息。
可选地，射频单元1701还用于：
发送目标信息，所述目标信息包括如下至少一项：信号配置信息、测量配置信息、指示信息；其中，所述指示信息用于指示如下至少一项：
获取第一测量数据、获取第二测量数据、上报第一测量数据、上报第二测量数据；
其中，所述第一测量数据用于输入到所述第一AI单元进行推理得到所述第一AI单元输出的感知数据，且所述第一测量数据包括第一测量结果；
所述第二测量数据用于确定所述参考数据，且所述第二测量数据包括第二测量结果。
可选地，所述信号配置信息包括如下至少一项：
第一信号的配置信息、第二信号的配置信息，所述第一信号为用于获取所述第一AI单元的输入数据的信号，所述第二信号为用于获取所述参考数据的信号，所述第一信号与所述第二信号满足如下至少一项：
所述第二信号的频域资源长度大于所述第一信号的频域资源长度；
所述第二信号的频域资源间隔小于所述第一信号的频域资源间隔；
所述第二信号的时域资源长度大于所述第一信号的时域资源长度；
所述第二信号的时域资源单元小于所述第一信号的时域资源间隔；
所述第二信号的发射功率大于所述第一信号的发射功率；
所述第一信号为所述第二信号的子集。
可选地，所述测量配置信息包括如下至少一项：
测量的信号资源指示、所述第一测量数据对应的测量用途信息、所述第二测量数据对应的测量用途信息、所述第一测量数据的处理方式信息、所述第二测量数据的处理方式信息、所述第一测量数据对应的AI单元的信息、所述第一测量数据的测量量、所述第二测量数据的测量量、所述第一测量数据的格式信息、所述第二测量数据的格式信息、所述参考数据的类型信息、上报配置、所述第一测量数据是否基于传感器进行测量的指示、所述第二测量数据是否基于传感器进行测量的指示、所述第一测量数据关联的传感器的类型信息、所述第二测量数据关联的传感器的类型信息、所述第一测量数据的测量方式、所述第二测量数据的测量方式。
可选地，所述第一测量数据的测量量包括如下至少一项：
接收信号、信道信息、谱信息、基本测量量；
所述第二测量数据的测量量包括如下至少一项：
接收信号、信道信息、谱信息、基本测量量。
可选地，所述第一测量数据的格式信息用于指示如下至少一项：
信道信息的维度、谱信息的范围、基本测量量的类型、传感器数据的格式；
所述第二测量数据的格式信息用于指示如下至少一项：
信道信息的维度、谱信息的范围、基本测量量的类型。
可选地，所述发送目标信息包括如下至少一项：
向第二设备发送所述目标信息的全部或者部分；
向第三设备发送所述目标信息的全部或者部分；
其中，所述第二设备为用于提供第一测量数据和第二测量数据中的至少一项的设备，所述第三设备为用于提供第二测量数据的设备。
可选地，所述第二设备满足如下至少一项条件：
支持目标感知模式；
位置固定；
支持的发射功率超过预设门限；
支持的天线个数超过预设门限；
支持的阵列孔径超过预设门限；
可分配给用于感知的资源量超过预设门限；
与所述第三设备之间的时钟同步误差小于预设门限；
或，所述第三设备满足如下至少一项条件：
支持目标感知模式；
位置固定；
支持的发射功率超过预设门限；
支持的天线个数超过预设门限；
支持的阵列孔径超过预设门限；
可分配给用于感知的资源量超过预设门限；
与所述第二设备之间的时钟同步误差小于预设门限。
可选地，处理器1710还用于：
基于所述监控结果确定针对所述第一AI单元的目标操作，所述目标操作包括如下至少一项：
去激活、切换、微调、重训练。
上述设备有利于提升或者保证感知性能。
可以理解，本实施例中提及的各实现方式的实现过程可以参照AI单元监控方法实施例的相关描述，并达到相同或相应的技术效果，为避免重复，在此不再赘述。
需要说明的是，上述设备也可以实现图8所示的方法中的步骤，或者可以实现图15所示的各模块执行的方法。
本申请实施例还提供了一种设备，包括处理器及通信接口，其中，所述通信接口用于发送数据，所述发送数据包括如下至少一项：发送第一测量数据、发送第二测量数据；其中，所述第一测量数据用于输入到第一AI单元进行推理得到所述第一AI单元输出的感知数据，且所述第一测量数据包括第一测量结果；或者，所述第一测量数据包括所述第一AI单元输出的感知数据；所述第二测量数据用于确定参考信号，且所述第二测量数据包括第二测量结果；或者，所述第二测量数据包括参考数据，所述参考数据用于监控所述第一AI单元。
本申请实施例还提供一种设备，该设备为第二设备，包括处理器和通信接口，所述通信接口和所述处理器耦合，所述处理器用于运行程序或指令，实现如图8所示的方法实施例的步骤。该设备实施例与上述第二设备侧方法实施例对应，上述方法实施例的各个实施过程和实现方式均可适用于该设备实施例中，且能达到相同的技术效果。
具体地，本申请实施例还提供了一种设备，该设备为第二设备，该设备可以是图15所示的信号发送装置。如图18所示，该设备1800包括：天线1801、射频装置1802、基带装置1803、处理器1804和存储器1805。天线1801与射频装置1802连接。在上行方向上，射频装置1802通过天线1801接收信息，将接收的信息发送给基带装置1803进行处理。在下行方向上，基带装置1803对要发送的信息进行处理，并发送给射频装置1802，射频装置1802对收到的信息进行处理后经过天线1801发送出去。
以上实施例中设备执行的方法可以在基带装置1803中实现，该基带装置1803包括基带处理器。
基带装置1803例如可以包括至少一个基带板，该基带板上设置有多个芯片，如图18所示，其中一个芯片例如为基带处理器，通过总线接口与存储器1805连接，以调用存储器1805中的程序，执行以上方法实施例中所示的网络设备操作。
该设备还可以包括网络接口1806，该接口例如为通用公共无线接口(Common Public Radio Interface，CPRI)。
具体地，本申请实施例的设备1800还包括：存储在存储器1805上并可在处理器1804上运行的指令或程序，处理器1804调用存储器1805中的指令或程序执行图15所示各模块执行的方法，并达到相同的技术效果，为避免重复，故不在此赘述。
本实施例中，以上述第二设备为网络侧设备进行举例说明。
射频装置1802，用于发送数据，所述发送数据包括如下至少一项：
发送第一测量数据、发送第二测量数据；
其中，所述第一测量数据用于输入到第一AI单元进行推理得到所述第一AI单元输出的感知数据，且所述第一测量数据包括第一测量结果；或者，所述第一测量数据包括所述第一AI单元输出的感知数据；
所述第二测量数据用于确定参考信号，且所述第二测量数据包括第二测量结果；或者，所述第二测量数据包括参考数据，所述参考数据用于监控所述第一AI单元。
可选地，所述参考数据包括如下至少一项：
通过非AI的方式获取的感知数据；
第二AI单元输出的感知数据，所述第二AI单元的处理能力强于所述第一AI单元的处理能力。
可选地，所述通过非AI的方式获取的感知数据包括如下至少一项：
通过估计算法获取的感知数据；
通过传感器获取的感知数据；
通过目标感知模式获取的感知数据；
通过目标感知设备获取的感知数据。
可选地，所述第一测量数据还包括：第一说明信息，所述第一说明信息包括如下至少一项：
所述第一测量数据的时间戳信息、所述第一测量数据的感知模式信息、用于指示所述第一测量结果用于AI单元推理的指示信息、所述第一测量数据对应的AI单元的信息、所述第一测量数据的性能指标信息、用于指示所述第一测量数据包括第一AI单元输出的感知数据的指示信息、所述第一测量数据关联的参考数据类型信息、所述第一测量数据对应的精度信息、所述第一测量数据对应的性能等级信息、所述第一测量数据对应的置信度水平信息。
可选地，所述第二测量数据还包括第二说明信息，所述第二说明信息包括如下至少一项：
所述第二测量数据的时间戳信息、所述第二测量数据的感知模式信息、用于所述第二测量结果用于监控的指示信息、所述第二测量数据关联的参考数据类型信息、所述第二测量数据的性能指标信息、用于指示所述第二测量数据包括所述参考数据的指示信息、所述第二测量数据对应的AI单元的信息、所述第二测量数据对应的精度信息、所述第二测量数据的坐标系信息、所述第二测量数据对应的传感器类型信息。
可选地，射频装置1802还用于：
接收目标信息，所述目标信息包括如下至少一项：信号配置信息、测量配置信息、指示信息；其中，所述指示信息用于指示如下至少一项：
获取所述第一测量数据、获取所述第二测量数据、上报所述第一测量数据、上报所述第二测量数据。
可选地，所述信号配置信息包括如下至少一项：
第一信号的配置信息、第二信号的配置信息，所述第一信号为用于获取所述第一AI单元的输入数据的信号，所述第二信号为用于获取所述参考数据的信号，所述第一信号与所述第二信号满足如下至少一项：
所述第二信号的频域资源长度大于所述第一信号的频域资源长度；
所述第二信号的频域资源间隔小于所述第一信号的频域资源间隔；
所述第二信号的时域资源长度大于所述第一信号的时域资源长度；
所述第二信号的时域资源单元小于所述第一信号的时域资源间隔；
所述第二信号的发射功率大于所述第一信号的发射功率；
所述第一信号为所述第二信号的子集。
可选地，所述测量配置信息包括如下至少一项：
测量的信号资源指示、所述第一测量数据对应的测量用途信息、所述第二测量数据对应的测量用途信息、所述第一测量数据的处理方式信息、所述第二测量数据的处理方式信息、所述第一测量数据对应的AI单元的信息、所述第一测量数据的测量量、所述第二测量数据的测量量、所述第一测量数据的格式信息、所述第二测量数据的格式信息、所述参考数据的类型信息、上报配置、所述第一测量数据是否基于传感器进行测量的指示、所述第二测量数据是否基于传感器进行测量的指示、所述第一测量数据关联的传感器的类型信息、所述第二测量数据关联的传感器的类型信息、所述第一测量数据的测量方式、所述第二测量数据的测量方式。
可选地，所述第一测量数据的测量量包括如下至少一项：
接收信号、信道信息、谱信息、基本测量量；
所述第二测量数据的测量量包括如下至少一项：
接收信号、信道信息、谱信息、基本测量量。
可选地，所述第一测量数据的格式信息用于指示如下至少一项：
信道信息的维度、谱信息的范围、基本测量量的类型、传感器数据的格式；
所述第二测量数据的格式信息用于指示如下至少一项：
信道信息的维度、谱信息的范围、基本测量量的类型。
可选地，射频装置1802还用于：
基于所述目标信息执行目标操作，所述目标操作包括如下至少一项：
执行第一测量，得到所述第一测量结果；
执行第二测量，得到第二测量结果；
执行第三测量，并将所述第三测量的测量结果输入到所述第一AI单元进行推理，得到所述第一AI单元输出的感知数据；
执行第四测量，并基于所述第四测量的测量结果确定所述参考数据。
可选地，所述第二设备满足如下至少一项条件：
支持目标感知模式；
位置固定；
支持的发射功率超过预设门限；
支持的天线个数超过预设门限；
支持的阵列孔径超过预设门限；
可分配给用于感知的资源量超过预设门限；
与第三设备之间的时钟同步误差小于预设门限，所述第三设备为用于提供第二测量数据的设备。
上述设备有利于提升或者保证感知性能。
可以理解，本实施例中提及的各实现方式的实现过程可以参照AI单元监控方法实施例的相关描述，并达到相同或相应的技术效果，为避免重复，在此不再赘述。
需要说明的是，上述设备也可以实现图6所示的方法中的步骤，或者可以实现图14所示的各模块执行的方法。
具体地，本申请实施例还提供了一种网络侧设备，该设备为第一设备。如图19所示，该网络侧设备1900包括：处理器1901、网络接口1902和存储器1903。其中，网络接口1902例如为通用公共无线接口(common public radio interface，CPRI)。
具体地，本申请实施例的网络侧设备1900还包括：存储在存储器1903上并可在处理器1901上运行的指令或程序，处理器1901调用存储器1903中的指令或程序执行图15所示各模块执行的方法，并达到相同的技术效果，为避免重复，故不在此赘述。
本实施例中，以上述第一设备为核心网设备进行举例说明。
其中，网络接口1902或者处理器1901，用于获取第一AI单元输出的感知数据；
网络接口1902或者处理器1901，用于获取用于监控所述第一AI单元的参考数据；
处理器1901，用于基于所述参考数据和所述第一AI单元输出的感知数据，确定所述第一AI单元的监控结果。
可选地，所述参考数据包括如下至少一项：
通过非AI的方式获取的感知数据；
第二AI单元输出的感知数据，所述第二AI单元的处理能力强于所述第一AI单元的处理能力。
可选地，所述通过非AI的方式获取的感知数据包括如下至少一项：
通过估计算法获取的感知数据；
通过传感器获取的感知数据；
通过目标感知模式获取的感知数据；
通过目标感知设备获取的感知数据。
可选地，所述获取第一AI单元输出的感知数据包括：
接收第一测量数据；
其中，所述第一测量数据用于输入到所述第一AI单元进行推理得到所述第一AI单元输出的感知数据，且所述第一测量数据包括第一测量结果；或者
所述第一测量数据包括所述第一AI单元输出的感知数据。
可选地，所述第一测量数据还包括：第一说明信息，所述第一说明信息包括如下至少一项：
所述第一测量数据的时间戳信息、所述第一测量数据的感知模式信息、用于指示所述第一测量结果用于AI单元推理的指示信息、所述第一测量数据对应的AI单元的信息、所述第一测量数据的性能指标信息、用于指示所述第一测量数据包括第一AI单元输出的感知数据的指示信息、所述第一测量数据关联的参考数据类型信息、所述第一测量数据对应的精度信息、所述第一测量数据对应的性能等级信息、所述第一测量数据对应的置信度水平信息。
可选地，所述获取用于监控所述第一AI单元的参考数据包括：
接收第二测量数据；
其中，所述第二测量数据用于确定所述参考数据，且所述第二测量数据包括第二测量结果；或者，
所述第二测量数据包括所述参考数据。
可选地，所述第二测量数据还包括第二说明信息，所述第二说明信息包括如下至少一项：
所述第二测量数据的时间戳信息、所述第二测量数据的感知模式信息、用于所述第二测量结果用于监控的指示信息、所述第二测量数据关联的参考数据类型信息、所述第二测量数据的性能指标信息、用于指示所述第二测量数据包括所述参考数据的指示信息、所述第二测量数据对应的AI单元的信息、所述第二测量数据对应的精度信息、所述第二测量数据的坐标系信息、所述第二测量数据对应的传感器类型信息。
可选地，所述监控结果包括如下至少一项：
有效性信息、性能等级信息、感知精度信息；
其中，所述感知精度信息包括如下至少一项：
距离精度信息、速度精度信息、角度精度信息、定位精度信息、检测精度信息、识别精度信息。
可选地，网络接口1902还用于：
发送目标信息，所述目标信息包括如下至少一项：信号配置信息、测量配置信息、指示信息；其中，所述指示信息用于指示如下至少一项：
获取第一测量数据、获取第二测量数据、上报第一测量数据、上报第二测量数据；
其中，所述第一测量数据用于输入到所述第一AI单元进行推理得到所述第一AI单元输出的感知数据，且所述第一测量数据包括第一测量结果；
所述第二测量数据用于确定所述参考数据，且所述第二测量数据包括第二测量结果。
可选地，所述信号配置信息包括如下至少一项：
第一信号的配置信息、第二信号的配置信息，所述第一信号为用于获取所述第一AI单元的输入数据的信号，所述第二信号为用于获取所述参考数据的信号，所述第一信号与所述第二信号满足如下至少一项：
所述第二信号的频域资源长度大于所述第一信号的频域资源长度；
所述第二信号的频域资源间隔小于所述第一信号的频域资源间隔；
所述第二信号的时域资源长度大于所述第一信号的时域资源长度；
所述第二信号的时域资源单元小于所述第一信号的时域资源间隔；
所述第二信号的发射功率大于所述第一信号的发射功率；
所述第一信号为所述第二信号的子集。
可选地，所述测量配置信息包括如下至少一项：
测量的信号资源指示、所述第一测量数据对应的测量用途信息、所述第二测量数据对应的测量用途信息、所述第一测量数据的处理方式信息、所述第二测量数据的处理方式信息、所述第一测量数据对应的AI单元的信息、所述第一测量数据的测量量、所述第二测量数据的测量量、所述第一测量数据的格式信息、所述第二测量数据的格式信息、所述参考数据的类型信息、上报配置、所述第一测量数据是否基于传感器进行测量的指示、所述第二测量数据是否基于传感器进行测量的指示、所述第一测量数据关联的传感器的类型信息、所述第二测量数据关联的传感器的类型信息、所述第一测量数据的测量方式、所述第二测量数据的测量方式。
可选地，所述第一测量数据的测量量包括如下至少一项：
接收信号、信道信息、谱信息、基本测量量；
所述第二测量数据的测量量包括如下至少一项：
接收信号、信道信息、谱信息、基本测量量。
可选地，所述第一测量数据的格式信息用于指示如下至少一项：
信道信息的维度、谱信息的范围、基本测量量的类型、传感器数据的格式；
所述第二测量数据的格式信息用于指示如下至少一项：
信道信息的维度、谱信息的范围、基本测量量的类型。
可选地，所述发送目标信息包括如下至少一项：
向第二设备发送所述目标信息的全部或者部分；
向第三设备发送所述目标信息的全部或者部分；
其中，所述第二设备为用于提供第一测量数据和第二测量数据中的至少一项的设备，所述第三设备为用于提供第二测量数据的设备。
可选地，所述第二设备满足如下至少一项条件：
支持目标感知模式；
位置固定；
支持的发射功率超过预设门限；
支持的天线个数超过预设门限；
支持的阵列孔径超过预设门限；
可分配给用于感知的资源量超过预设门限；
与所述第三设备之间的时钟同步误差小于预设门限；
或，所述第三设备满足如下至少一项条件：
支持目标感知模式；
位置固定；
支持的发射功率超过预设门限；
支持的天线个数超过预设门限；
支持的阵列孔径超过预设门限；
可分配给用于感知的资源量超过预设门限；
与所述第二设备之间的时钟同步误差小于预设门限。
可选地，处理器1901还用于：
基于所述监控结果确定针对所述第一AI单元的目标操作，所述目标操作包括如下至少一项：
去激活、切换、微调、重训练。
上述设备有利于提升或者保证感知性能。
可以理解，本实施例中提及的各实现方式的实现过程可以参照AI单元监控方法实施例的相关描述，并达到相同或相应的技术效果，为避免重复，在此不再赘述。
需要说明的是，上述设备也可以实现图8所示的方法中的步骤，或者可以实现图15所示的各模块执行的方法。
本申请实施例还提供一种可读存储介质，所述可读存储介质上存储有程序或指令，该程序或指令被处理器执行时实现上述AI单元监控方法实施例的各个过程，且能达到相同的技术效果，为避免重复，这里不再赘述。
其中，所述处理器为上述实施例中所述的终端中的处理器。所述可读存储介质，包括计算机可读存储介质，如计算机只读存储器ROM、随机存取存储器RAM、磁碟或者光盘等。在一些示例中，可读存储介质可以是非瞬态的可读存储介质。
本申请实施例另提供了一种芯片，所述芯片包括处理器和通信接口，所述通信接口和所述处理器耦合，所述处理器用于运行程序或指令，实现上述AI单元监控方法实施例的各个过程，且能达到相同的技术效果，为避免重复，这里不再赘述。
应理解，本申请实施例提到的芯片还可以称为系统级芯片，系统芯片，芯片系统或片上系统芯片等。
本申请实施例另提供了一种计算机程序/程序产品，所述计算机程序/程序产品被存储在存储介质中，所述计算机程序/程序产品被至少一个处理器执行以实现上述AI单元监控方法实施例的各个过程，且能达到相同的技术效果，为避免重复，这里不再赘述。
本申请实施例还提供了一种无线通信系统，包括：第一设备及第二设备，所述第一设备可用于执行如本申请实施例提供的第一设备侧的AI单元监控方法的步骤，所述第二设备可用于执行如本申请实施例提供的第二设备侧的AI单元监控方法的步骤。
需要说明的是，在本文中，术语“包括”、“包含”或者其任何其他变体意在涵盖非排他性的包含，从而使得包括一系列要素的过程、方法、物品或者装置不仅包括那些要素，而且还包括没有明确列出的其他要素，或者是还包括为这种过程、方法、物品或者装置所固有的要素。在没有更多限制的情况下，由语句“包括一个……”限定的要素，并不排除在包括该要素的过程、方法、物品或者装置中还存在另外的相同要素。此外，需要指出的是，本申请实施方式中的方法和装置的范围不限按示出或讨论的顺序来执行功能，还可包括根据所涉及的功能按基本同时的方式或按相反的顺序来执行功能，例如，可以按不同于所描述的次序来执行所描述的方法，并且还可以添加、省去或组合各种步骤。另外，参照某些示例所描述的特征可在其他示例中被组合。
通过以上的实施方式的描述，本领域的技术人员可以清楚地了解到上述实施例方法可借助计算机软件产品加必需的通用硬件平台的方式来实现，当然也可以通过硬件。该计算机软件产品存储在存储介质(如ROM、RAM、磁碟、光盘等)中，包括若干指令，用以使得终端或者网络侧设备执行本申请各个实施例所述的方法。
上面结合附图对本申请的实施例进行了描述，但是本申请并不局限于上述的具体实施方式，上述的具体实施方式仅仅是示意性的，而不是限制性的，本领域的普通技术人员在本申请的启示下，在不脱离本申请宗旨和权利要求所保护的范围情况下，还可做出很多形式的实施方式，这些实施方式均属于本申请的保护之内。

Claims (38)

  1. 一种人工智能AI单元监控方法，包括：
    第一设备获取第一AI单元输出的感知数据；
    所述第一设备获取用于监控所述第一AI单元的参考数据；
    所述第一设备基于所述参考数据和所述第一AI单元输出的感知数据，确定所述第一AI单元的监控结果。
  2. 如权利要求1所述的方法，其中，所述参考数据包括如下至少一项：
    通过非AI的方式获取的感知数据；
    第二AI单元输出的感知数据，所述第二AI单元的处理能力强于所述第一AI单元的处理能力。
  3. 如权利要求2所述的方法，其中，所述通过非AI的方式获取的感知数据包括如下至少一项：
    通过估计算法获取的感知数据；
    通过传感器获取的感知数据；
    通过目标感知模式获取的感知数据；
    通过目标感知设备获取的感知数据。
  4. 如权利要求1至3中任一项所述的方法，其中，所述第一设备获取第一AI单元输出的感知数据包括：
    所述第一设备接收第一测量数据；
    其中，所述第一测量数据用于输入到所述第一AI单元进行推理得到所述第一AI单元输出的感知数据，且所述第一测量数据包括第一测量结果；或者
    所述第一测量数据包括所述第一AI单元输出的感知数据。
  5. 如权利要求4所述的方法，其中，所述第一测量数据还包括：第一说明信息，所述第一说明信息包括如下至少一项：
    所述第一测量数据的时间戳信息、所述第一测量数据的感知模式信息、用于指示所述第一测量结果用于AI单元推理的指示信息、所述第一测量数据对应的AI单元的信息、所述第一测量数据的性能指标信息、用于指示所述第一测量数据包括第一AI单元输出的感知数据的指示信息、所述第一测量数据关联的参考数据类型信息、所述第一测量数据对应的精度信息、所述第一测量数据对应的性能等级信息、所述第一测量数据对应的置信度水平信息。
  6. 如权利要求1至5中任一项所述的方法，其中，所述第一设备获取用于监控所述第一AI单元的参考数据包括：
    所述第一设备接收第二测量数据；
    其中，所述第二测量数据用于确定所述参考数据，且所述第二测量数据包括第二测量结果；或者，
    所述第二测量数据包括所述参考数据。
  7. 如权利要求6所述的方法，其中，所述第二测量数据还包括第二说明信息，所述第二说明信息包括如下至少一项：
    所述第二测量数据的时间戳信息、所述第二测量数据的感知模式信息、用于所述第二测量结果用于监控的指示信息、所述第二测量数据关联的参考数据类型信息、所述第二测量数据的性能指标信息、用于指示所述第二测量数据包括所述参考数据的指示信息、所述第二测量数据对应的AI单元的信息、所述第二测量数据对应的精度信息、所述第二测量数据的坐标系信息、所述第二测量数据对应的传感器类型信息。
  8. 如权利要求1至7中任一项所述的方法，其中，所述监控结果包括如下至少一项：
    有效性信息、性能等级信息、感知精度信息；
    其中，所述感知精度信息包括如下至少一项：
    距离精度信息、速度精度信息、角度精度信息、定位精度信息、检测精度信息、识别精度信息。
  9. 如权利要求1至8中任一项所述的方法，其中，所述方法还包括：
    所述第一设备发送目标信息，所述目标信息包括如下至少一项：信号配置信息、测量配置信息、指示信息；其中，所述指示信息用于指示如下至少一项：
    获取第一测量数据、获取第二测量数据、上报第一测量数据、上报第二测量数据；
    其中，所述第一测量数据用于输入到所述第一AI单元进行推理得到所述第一AI单元输出的感知数据，且所述第一测量数据包括第一测量结果；
    所述第二测量数据用于确定所述参考数据，且所述第二测量数据包括第二测量结果。
  10. 如权利要求9所述的方法，其中，所述信号配置信息包括如下至少一项：
    第一信号的配置信息、第二信号的配置信息，所述第一信号为用于获取所述第一AI单元的输入数据的信号，所述第二信号为用于获取所述参考数据的信号，所述第一信号与所述第二信号满足如下至少一项：
    所述第二信号的频域资源长度大于所述第一信号的频域资源长度；
    所述第二信号的频域资源间隔小于所述第一信号的频域资源间隔；
    所述第二信号的时域资源长度大于所述第一信号的时域资源长度；
    所述第二信号的时域资源单元小于所述第一信号的时域资源间隔；
    所述第二信号的发射功率大于所述第一信号的发射功率；
    所述第一信号为所述第二信号的子集。
  11. 如权利要求9或10所述的方法，其中，所述测量配置信息包括如下至少一项：
    测量的信号资源指示、所述第一测量数据对应的测量用途信息、所述第二测量数据对应的测量用途信息、所述第一测量数据的处理方式信息、所述第二测量数据的处理方式信息、所述第一测量数据对应的AI单元的信息、所述第一测量数据的测量量、所述第二测量数据的测量量、所述第一测量数据的格式信息、所述第二测量数据的格式信息、所述参考数据的类型信息、上报配置、所述第一测量数据是否基于传感器进行测量的指示、所述第二测量数据是否基于传感器进行测量的指示、所述第一测量数据关联的传感器的类型信息、所述第二测量数据关联的传感器的类型信息、所述第一测量数据的测量方式、所述第二测量数据的测量方式。
  12. 如权利要求11所述的方法，其中，所述第一测量数据的测量量包括如下至少一项：
    接收信号、信道信息、谱信息、基本测量量；
    所述第二测量数据的测量量包括如下至少一项：
    接收信号、信道信息、谱信息、基本测量量。
  13. 如权利要求11或12所述的方法，其中，所述第一测量数据的格式信息用于指示如下至少一项：
    信道信息的维度、谱信息的范围、基本测量量的类型、传感器数据的格式；
    所述第二测量数据的格式信息用于指示如下至少一项：
    信道信息的维度、谱信息的范围、基本测量量的类型。
  14. 如权利要求9至13中任一项所述的方法，其中，所述第一设备发送目标信息包括如下至少一项：
    所述第一设备向第二设备发送所述目标信息的全部或者部分；
    所述第一设备向第三设备发送所述目标信息的全部或者部分；
    其中，所述第二设备为用于提供第一测量数据和第二测量数据中的至少一项的设备，所述第三设备为用于提供第二测量数据的设备。
  15. 如权利要求14所述的方法，其中，所述第二设备满足如下至少一项条件：
    支持目标感知模式；
    位置固定；
    支持的发射功率超过预设门限；
    支持的天线个数超过预设门限；
    支持的阵列孔径超过预设门限；
    可分配给用于感知的资源量超过预设门限；
    与所述第三设备之间的时钟同步误差小于预设门限；
    或，所述第三设备满足如下至少一项条件：
    支持目标感知模式；
    位置固定；
    支持的发射功率超过预设门限；
    支持的天线个数超过预设门限；
    支持的阵列孔径超过预设门限；
    可分配给用于感知的资源量超过预设门限；
    与所述第二设备之间的时钟同步误差小于预设门限。
  16. 如权利要求1至15中任一项所述的方法，其中，所述方法还包括：
    所述第一设备基于所述监控结果确定针对所述第一AI单元的目标操作，所述目标操作包括如下至少一项：
    去激活、切换、微调、重训练。
  17. 一种人工智能AI单元监控方法，包括：
    第二设备发送数据，所述发送数据包括如下至少一项：
    发送第一测量数据、发送第二测量数据；
    其中，所述第一测量数据用于输入到第一AI单元进行推理得到所述第一AI单元输出的感知数据，且所述第一测量数据包括第一测量结果；或者，所述第一测量数据包括所述第一AI单元输出的感知数据；
    所述第二测量数据用于确定参考信号，且所述第二测量数据包括第二测量结果；或者，所述第二测量数据包括参考数据，所述参考数据用于监控所述第一AI单元。
  18. 如权利要求17所述的方法，其中，所述参考数据包括如下至少一项：
    通过非AI的方式获取的感知数据；
    第二AI单元输出的感知数据，所述第二AI单元的处理能力强于所述第一AI单元的处理能力。
  19. 如权利要求18所述的方法，其中，所述通过非AI的方式获取的感知数据包括如下至少一项：
    通过估计算法获取的感知数据；
    通过传感器获取的感知数据；
    通过目标感知模式获取的感知数据；
    通过目标感知设备获取的感知数据。
  20. 如权利要求17至19中任一项所述的方法，其中，所述第一测量数据还包括：第一说明信息，所述第一说明信息包括如下至少一项：
    所述第一测量数据的时间戳信息、所述第一测量数据的感知模式信息、用于指示所述第一测量结果用于AI单元推理的指示信息、所述第一测量数据对应的AI单元的信息、所述第一测量数据的性能指标信息、用于指示所述第一测量数据包括第一AI单元输出的感知数据的指示信息、所述第一测量数据关联的参考数据类型信息、所述第一测量数据对应的精度信息、所述第一测量数据对应的性能等级信息、所述第一测量数据对应的置信度水平信息。
  21. 如权利要求17至20中任一项所述的方法，其中，所述第二测量数据还包括第二说明信息，所述第二说明信息包括如下至少一项：
    所述第二测量数据的时间戳信息、所述第二测量数据的感知模式信息、用于所述第二测量结果用于监控的指示信息、所述第二测量数据关联的参考数据类型信息、所述第二测量数据的性能指标信息、用于指示所述第二测量数据包括所述参考数据的指示信息、所述第二测量数据对应的AI单元的信息、所述第二测量数据对应的精度信息、所述第二测量数据的坐标系信息、所述第二测量数据对应的传感器类型信息。
  22. 如权利要求17至21中任一项所述的方法，其中，所述方法还包括：
    所述第二设备接收目标信息，所述目标信息包括如下至少一项：信号配置信息、测量配置信息、指示信息；其中，所述指示信息用于指示如下至少一项：
    获取所述第一测量数据、获取所述第二测量数据、上报所述第一测量数据、上报所述第二测量数据。
  23. 如权利要求22所述的方法，其中，所述信号配置信息包括如下至少一项：
    第一信号的配置信息、第二信号的配置信息，所述第一信号为用于获取所述第一AI单元的输入数据的信号，所述第二信号为用于获取所述参考数据的信号，所述第一信号与所述第二信号满足如下至少一项：
    所述第二信号的频域资源长度大于所述第一信号的频域资源长度；
    所述第二信号的频域资源间隔小于所述第一信号的频域资源间隔；
    所述第二信号的时域资源长度大于所述第一信号的时域资源长度；
    所述第二信号的时域资源单元小于所述第一信号的时域资源间隔；
    所述第二信号的发射功率大于所述第一信号的发射功率；
    所述第一信号为所述第二信号的子集。
  24. 如权利要求22或23所述的方法，其中，所述测量配置信息包括如下至少一项：
    测量的信号资源指示、第一测量数据对应的测量用途信息、第二测量数据对应的测量用途信息、第一测量数据的处理方式信息、第二测量数据的处理方式信息、所述第一测量数据对应的AI单元的信息、第一测量数据的测量量、第二测量数据的测量量、第一测量数据的格式信息、第二测量数据的格式信息、所述参考数据的类型信息、上报配置、所述第一测量数据是否基于传感器进行测量的指示、所述第二测量数据是否基于传感器进行测量的指示、所述第一测量数据关联的传感器的类型信息、所述第二测量数据关联的传感器的类型信息、所述第一测量数据的测量方式、所述第二测量数据的测量方式。
  25. 如权利要求24所述的方法，其中，所述第一测量数据的测量量包括如下至少一项：
    接收信号、信道信息、谱信息、基本测量量；
    所述第二测量数据的测量量包括如下至少一项：
    接收信号、信道信息、谱信息、基本测量量。
  26. 如权利要求24或25所述的方法，其中，所述第一测量数据的格式信息用于指示如下至少一项：
    信道信息的维度、谱信息的范围、基本测量量的类型、传感器数据的格式；
    所述第二测量数据的格式信息用于指示如下至少一项：
    信道信息的维度、谱信息的范围、基本测量量的类型。
  27. 如权利要求22至26中任一项所述的方法，其中，所述方法还包括：
    所述第二设备基于所述目标信息执行目标操作，所述目标操作包括如下至少一项：
    执行第一测量，得到所述第一测量结果；
    执行第二测量，得到第二测量结果；
    执行第三测量，并将所述第三测量的测量结果输入到所述第一AI单元进行推理，得到所述第一AI单元输出的感知数据；
    执行第四测量，并基于所述第四测量的测量结果确定所述参考数据。
  28. 一种人工智能AI单元监控装置，包括：
    处理模块，用于获取第一AI单元输出的感知数据；
    所述处理模块，还用于获取用于监控所述第一AI单元的参考数据；
    所述处理模块，还用于基于所述参考数据和所述第一AI单元输出的感知数据，确定所述第一AI单元的监控结果。
  29. 如权利要求28所述的装置，其中，所述获取第一AI单元输出的感知数据包括：
    接收第一测量数据；
    其中，所述第一测量数据用于输入到所述第一AI单元进行推理得到所述第一AI单元输出的感知数据，且所述第一测量数据包括第一测量结果；或者
    所述第一测量数据包括所述第一AI单元输出的感知数据。
  30. 如权利要求28或29所述的装置，其中，所述获取用于监控所述第一AI单元的参考数据包括：
    接收第二测量数据；
    其中，所述第二测量数据用于确定所述参考数据，且所述第二测量数据包括第二测量结果；或者，
    所述第二测量数据包括所述参考数据。
  31. 如权利要求28至30中任一项所述的装置，其中，所述装置还包括：
    发送模块，用于发送目标信息，所述目标信息包括如下至少一项：信号配置信息、测量配置信息、指示信息；其中，所述指示信息用于指示如下至少一项：
    获取第一测量数据、获取第二测量数据、上报第一测量数据、上报第二测量数据；
    其中，所述第一测量数据用于输入到所述第一AI单元进行推理得到所述第一AI单元输出的感知数据，且所述第一测量数据包括第一测量结果；
    所述第二测量数据用于确定所述参考数据，且所述第二测量数据包括第二测量结果。
  32. 如权利要求28至31中任一项所述的装置，其中，所述处理模块还用于基于所述监控结果确定针对所述第一AI单元的目标操作，所述目标操作包括如下至少一项：
    去激活、切换、微调、重训练。
  33. 一种人工智能AI单元监控装置，包括：
    发送模块，用于发送数据，所述发送数据包括如下至少一项：
    发送第一测量数据、发送第二测量数据；
    其中，所述第一测量数据用于输入到第一AI单元进行推理得到所述第一AI单元输出的感知数据，且所述第一测量数据包括第一测量结果；或者，所述第一测量数据包括所述第一AI单元输出的感知数据；
    所述第二测量数据用于确定参考信号，且所述第二测量数据包括第二测量结果；或者，所述第二测量数据包括参考数据，所述参考数据用于监控所述第一AI单元。
  34. 如权利要求33所述的装置，其中，所述装置还包括：
    接收模块，用于接收目标信息，所述目标信息包括如下至少一项：信号配置信息、测量配置信息、指示信息；其中，所述指示信息用于指示如下至少一项：
    获取所述第一测量数据、获取所述第二测量数据、上报所述第一测量数据、上报所述第二测量数据。
  35. 如权利要求34所述的装置，其中，所述装置还包括：
    处理模块，用于基于所述目标信息执行目标操作，所述目标操作包括如下至少一项：
    执行第一测量，得到所述第一测量结果；
    执行第二测量，得到第二测量结果；
    执行第三测量，并将所述第三测量的测量结果输入到所述第一AI单元进行推理，得到所述第一AI单元输出的感知数据；
    执行第四测量，并基于所述第四测量的测量结果确定所述参考数据。
  36. 一种设备，包括处理器和存储器，所述存储器存储可在所述处理器上运行的程序或指令，所述程序或指令被所述处理器执行时实现如权利要求1至16任一项所述的AI单元监控方法的步骤，或者所述程序或指令被所述处理器执行时实现如权利要求17至27任一项所述的AI单元监控方法的步骤。
  37. 一种可读存储介质，所述可读存储介质上存储程序或指令，所述程序或指令被处理器执行时实现如权利要求1至16任一项所述的AI单元监控方法的步骤，或者实现如权利要求17至27任一项所述的AI单元监控方法的步骤。
  38. 一种计算机程序产品，所述计算机程序产品被存储在存储介质中，所述计算机程序产品被至少一个处理器执行以实现如权利要求1至16任一项所述的AI单元监控方法的步骤，或者实现如权利要求17至27任一项所述的AI单元监控方法的步骤。
PCT/CN2025/093340 2024-05-14 2025-05-08 Ai单元监控方法、装置及设备 Pending WO2025237147A1 (zh)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN202410597731.2A CN120957107A (zh) 2024-05-14 2024-05-14 Ai单元监控方法、装置及设备
CN202410597731.2 2024-05-14

Publications (1)

Publication Number Publication Date
WO2025237147A1 true WO2025237147A1 (zh) 2025-11-20

Family

ID=97619077

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2025/093340 Pending WO2025237147A1 (zh) 2024-05-14 2025-05-08 Ai单元监控方法、装置及设备

Country Status (2)

Country Link
CN (1) CN120957107A (zh)
WO (1) WO2025237147A1 (zh)

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2023026413A1 (ja) * 2021-08-25 2023-03-02 株式会社Ｎｔｔドコモ 端末、無線通信方法及び基地局
CN116033456A (zh) * 2022-12-20 2023-04-28 京信网络系统股份有限公司 压缩模型更新方法、装置、系统和存储介质
WO2024031569A1 (zh) * 2022-08-11 2024-02-15 北京小米移动软件有限公司 一种通信方法、装置、设备及存储介质
WO2024093997A1 (zh) * 2022-11-04 2024-05-10 维沃移动通信有限公司 确定模型适用性的方法、装置及通信设备

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2023026413A1 (ja) * 2021-08-25 2023-03-02 株式会社Ｎｔｔドコモ 端末、無線通信方法及び基地局
WO2024031569A1 (zh) * 2022-08-11 2024-02-15 北京小米移动软件有限公司 一种通信方法、装置、设备及存储介质
WO2024093997A1 (zh) * 2022-11-04 2024-05-10 维沃移动通信有限公司 确定模型适用性的方法、装置及通信设备
CN116033456A (zh) * 2022-12-20 2023-04-28 京信网络系统股份有限公司 压缩模型更新方法、装置、系统和存储介质

Also Published As

Publication number Publication date
CN120957107A (zh) 2025-11-14

Similar Documents

Publication Publication Date Title
WO2025124302A1 (zh) 测量结果处理方法、发送方法、装置及设备
WO2024131761A1 (zh) 感知协作方法、装置及通信设备
WO2025237147A1 (zh) Ai单元监控方法、装置及设备
WO2025237143A1 (zh) 联合推理方法、装置及设备
WO2025237141A1 (zh) 能力信息交互方法、装置及设备
WO2025237195A9 (zh) 感知处理的方法、装置、设备及介质
CN120957109A (zh) Ai单元监控方法、装置、第一设备及第二设备
WO2025237146A1 (zh) 信息处理方法、装置及设备
WO2025185586A9 (zh) 测量方法、测量配置方法、装置及设备
WO2026021466A1 (zh) 位置信息获取方法、装置及设备
WO2025185587A9 (zh) 测量方法、测量指示方法、装置及设备
WO2026046135A1 (zh) 感知方法、配置方法、装置及相关设备
WO2026086684A1 (zh) 协作感知处理方法、装置、终端及网络侧设备
WO2026067305A1 (zh) 位置测量方法、装置及设备
WO2026092520A1 (zh) 无线通信方法、装置以及设备
WO2025140367A9 (zh) 感知测量方法、装置及设备
CN120224266A (zh) 感知测量结果反馈方法、接收方法、装置及设备
WO2025130792A1 (zh) 模型推理方法及装置
CN120238216A (zh) 感知设备选择方法、装置及相关设备
WO2025237117A1 (zh) 感知处理的方法、装置、设备及介质
CN120238215A (zh) 感知设备选择方法、装置及相关设备
WO2026067307A1 (zh) 信号测量方法、装置及设备
WO2025185588A1 (zh) 测量信息上报方法、分组指示方法、装置及设备
CN120201482A (zh) Ai单元推理性能的监控方法、装置及通信设备
CN120201483A (zh) Ai单元推理性能的监控方法、装置及通信设备

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 25802798

Country of ref document: EP

Kind code of ref document: A1