WO2025020009A1 - 一种模型初始化方法及其装置 - Google Patents
一种模型初始化方法及其装置 Download PDFInfo
- Publication number
- WO2025020009A1 WO2025020009A1 PCT/CN2023/108752 CN2023108752W WO2025020009A1 WO 2025020009 A1 WO2025020009 A1 WO 2025020009A1 CN 2023108752 W CN2023108752 W CN 2023108752W WO 2025020009 A1 WO2025020009 A1 WO 2025020009A1
- Authority
- WO
- WIPO (PCT)
- Prior art keywords
- model
- communication device
- information
- csi
- csi feedback
- 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
Links
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04W—WIRELESS COMMUNICATION NETWORKS
- H04W24/00—Supervisory, monitoring or testing arrangements
- H04W24/02—Arrangements for optimising operational condition
Definitions
- the present disclosure relates to the field of communication technology, and in particular to a model initialization method and a device thereof.
- Massive Multiple-Input Multiple-Output is a key enabling technology for the 5th Generation (5G) mobile communication system. It helps to achieve ultra-large bandwidth, ultra-high rate, and low-latency communication, and will play a key role in the development of future wireless communication systems and provide technical support for various future wireless communication scenarios.
- 5G 5th Generation
- the base station In order to achieve the performance gains of large-scale MIMO technology in terms of system capacity, power efficiency, spectrum efficiency, etc., the base station (BS) needs to obtain accurate downlink channel state information (CSI).
- TDD time division duplex
- FDD frequency division duplex
- the CSI feedback method is usually adopted, that is, the user equipment (UE) first obtains the downlink CSI through channel estimation technology, and then feeds back the downlink CSI data to the base station through the uplink.
- UE user equipment
- Implementing downlink CSI feedback in FDD mode based on deep learning technology can effectively reduce feedback overhead and computational complexity, and can also significantly improve the accuracy of CSI feedback. It has become an important method to solve the CSI feedback problem in large-scale MIMO systems.
- the embodiment of the present disclosure proposes a model initialization method and a device thereof.
- a model initialization method comprising:
- the first communication device receives first information sent by the second communication device, where the first information includes model information of a first model configured by the second communication device, where the first model is a model for channel state information CSI feedback;
- the second information is sent to the second communication device, so that the second communication device performs model deployment based on the second information.
- a model initialization method comprising:
- the second communication device sends first information to the first communication device, where the first information includes model information of a first model configured by the second communication device, where the first model is a model for channel state information CSI feedback;
- a first communication device including:
- a transceiver module configured to receive first information sent by a second communication device, wherein the first information includes model information of a first model configured by the second communication device, and the first model is a model for channel state information CSI feedback;
- a processing module configured to determine second information based on the first information
- the transceiver module is further used to send second information to the second communication device so that the second communication device performs model deployment based on the second information.
- a second communication device including:
- a transceiver module configured to send first information to a first communication device, wherein the first information includes model information of a first model configured by the second communication device, wherein the first model is a model for channel state information CSI feedback;
- the transceiver module is further used to receive second information sent by the first communication device, where the second information is determined by the first communication device based on the first information, and the second information is model deployment information of the first model to be deployed in the second communication device;
- a processing module is used to deploy a model based on the second information.
- a communication system including:
- a first communication device configured to perform an optional implementation of the aforementioned first aspect
- the second communication device is configured to execute an optional implementation of the aforementioned second aspect.
- a communication device including: one or more processors;
- the processor is used to call instructions to enable the communication device to execute the optional implementation of the first and second aspects mentioned above.
- a storage medium stores instructions, and when the instructions are executed on a communication device, the communication device executes optional implementation methods of the aforementioned first and second aspects.
- model initialization between communication devices can be achieved, thereby ensuring the normal operation of the model used for CSI feedback and ensuring the accuracy of the CSI feedback solution based on artificial intelligence.
- FIG1 is a schematic diagram of the architecture of a communication system provided by an embodiment of the present disclosure.
- FIG2a is a diagram showing an example structure of a CSI feedback network according to an embodiment of the present disclosure
- FIG2b is an example diagram showing a switching scenario according to an embodiment of the present disclosure.
- FIG3 is an interactive schematic diagram of a model initialization method according to an embodiment of the present disclosure
- FIG4A is a schematic diagram of a flow chart of a model initialization method according to an embodiment of the present disclosure
- FIG4B is a flow chart of a model initialization method according to an embodiment of the present disclosure.
- FIG4C is an interactive schematic diagram of a model initialization method according to an embodiment of the present disclosure.
- FIG5A is a schematic diagram of a flow chart of a model initialization method according to an embodiment of the present disclosure
- FIG5B is a flow chart of a model initialization method according to an embodiment of the present disclosure.
- FIG5C is a flow chart of a model initialization method according to an embodiment of the present disclosure.
- FIG6 is an interactive schematic diagram of a model initialization method according to an embodiment of the present disclosure.
- FIG7 is a schematic diagram of a model initialization method according to an embodiment of the present disclosure.
- FIG8 is a schematic diagram of a model initialization method according to an embodiment of the present disclosure.
- FIG9A is a schematic diagram of the structure of a first communication device proposed in an embodiment of the present disclosure.
- FIG9B is a schematic diagram of the structure of a second communication device proposed in an embodiment of the present disclosure.
- FIG10A is a schematic diagram of the structure of a communication device 1100 provided in an embodiment of the present disclosure.
- FIG. 10B is a schematic diagram of the structure of the chip 1200 proposed in an embodiment of the present disclosure.
- the embodiment of the present disclosure proposes a model initialization method and a device thereof.
- an embodiment of the present disclosure provides a model initialization method, comprising:
- the first communication device receives first information sent by the second communication device, where the first information includes model information of a first model configured by the second communication device, where the first model is a model for channel state information CSI feedback;
- the second information is sent to the second communication device, so that the second communication device performs model deployment based on the second information.
- the model initialization between the communication devices can be realized, so as to effectively ensure that the CSI feedback networks between the communication devices can match each other during the actual system operation, thereby effectively ensuring the performance stability of the CSI feedback network and improving the practicality of the artificial intelligence solution.
- determining the second information based on the first information includes:
- the second information is determined based on the CSI real-time data and the first information.
- the CSI feedback network models between communication devices can match each other and be applicable to the current channel environment, thereby further effectively ensuring the performance stability of the CSI feedback network model and further improving the practicality of the artificial intelligence solution.
- the first communication device is a network device
- the second communication device is a terminal
- the first model is an encoder model in a CSI feedback network.
- the encoder model in the CSI feedback network is deployed on the terminal side, so that the terminal can complete feature extraction and dimensionality compression of the original CSI data based on the encoder model, compress the original CSI data into codewords with smaller data volume, and then transmit the codeword information to the network device through the uplink feedback link to fully reduce the feedback overhead.
- determining the CSI real-time data includes:
- pilot signal is used for the terminal to perform channel estimation at a current moment to obtain CSI real-time data
- data request indication information is used to instruct the network device to request the CSI real-time data from the terminal
- the received CSI real-time data that has been quantized is dequantized to obtain the CSI real-time data.
- the CSI real-time data obtained by the terminal after channel estimation can be obtained, so that the CSI real-time data that can accurately reflect the current channel environment can be obtained, which is convenient for the network device to select a model based on the CSI real-time data, thereby effectively ensuring that the CSI feedback network model between the network device and the terminal matches each other and is suitable for the current channel environment.
- the first communication device is a terminal
- the second communication device is a network device
- the first model is a decoder model in a CSI feedback network.
- determining the CSI real-time data includes:
- Channel estimation is performed based on the pilot signal at a current moment to obtain the CSI real-time data.
- the terminal can perform channel estimation based on the pilot signal sent by the network device to obtain CSI real-time data, thereby obtaining CSI real-time data that can accurately reflect the current channel environment, which is convenient for the terminal to select a model based on the CSI real-time data, thereby effectively ensuring that the CSI feedback network model between the network device and the terminal matches each other and is suitable for the current channel environment.
- determining the second information based on the CSI real-time data and the first information includes:
- the CSI feedback network includes the first model and the second model, and the CSI feedback network compresses and reconstructs the CSI real-time data through the first model and the second model;
- the second information is determined based on the model information of the first model in the first CSI feedback network and the model information of the first model configured by the second communication device.
- a CSI feedback network whose feedback accuracy meets certain conditions can be selected from the CSI feedback network deployed by the first communication device, and based on the model parameters of the first model in the selected CSI feedback network and the model information of the first model configured by the second communication device, the deployment of the first model in the second communication device is completed, which can ensure that the first model in the selected CSI feedback network is more suitable for the current channel environment.
- the first communication device deploys multiple CSI feedback networks; and selecting a first CSI feedback network from the CSI feedback networks deployed by the first communication device according to the feedback accuracy includes:
- the CSI feedback network with the highest feedback accuracy is selected as the first CSI feedback network.
- the performance stability of the first model in the feedback network can be further effectively guaranteed.
- determining the second information based on the model information of the first model in the first CSI feedback network and the model information of the first model configured by the second communication device includes any one of the following:
- model information of a first model in the first CSI feedback network as the second information, wherein the model information of the first model configured by the second communication device includes the model information of the first model in the first CSI feedback network;
- the model parameters of the first model in the first CSI feedback network are quantized, and the quantized model parameters are determined as the second information, wherein the model information of the first model configured by the second communication device does not include the model information of the first model in the first CSI feedback network.
- the model parameters of the model are determined as the information of the first model to be deployed in the second communication device, thereby completing the model initialization; if the model configured by the second communication device does not include the model selected by the first communication device, the model parameters of the model selected by the first communication device are provided to the second communication device for model deployment, thereby completing the model initialization. In this way, the normal operation of the model used for CSI feedback can be ensured, and the accuracy of the CSI feedback scheme based on artificial intelligence can be guaranteed.
- determining the quantized model parameters as the second information includes:
- the method also includes: receiving first indication information sent by the second communication device, the first indication information being used to indicate that the second communication device is configured with the first model and has the ability to use the first model to complete CSI feedback work.
- the method also includes: sending second indication information to the second communication device, the second indication information is used to instruct the first communication device to notify the second communication device to enter the model initialization process; receiving third indication information sent by the second communication device, the third indication information is used to indicate that the second communication device has accepted the model initialization instruction and entered the model initialization process.
- the method further includes: receiving confirmation information sent by the second communication device, the confirmation information including confirmation indication information for indicating that the second communication device has completed model initialization.
- the model information includes a model number and/or a model structure.
- the second communication device sends first information to the first communication device, where the first information includes model information of a first model configured by the second communication device, where the first model is a model for channel state information CSI feedback;
- the first communication device is a network device
- the second communication device is a terminal
- the first model is an encoder model in a CSI feedback network.
- the method further includes:
- pilot signal is used by the terminal to perform channel estimation at a current moment to obtain CSI real-time data
- data request indication information is used to instruct the network device to request the CSI real-time data from the terminal
- the quantized CSI real-time data is sent to the network device, wherein the CSI real-time data is used by the network device to determine second information in combination with the first information.
- the first communication device is a terminal
- the second communication device is a network device
- the first model is a decoder model in a CSI feedback network.
- the method further includes: sending a pilot signal to the terminal, wherein the pilot signal is used by the terminal to perform channel estimation at a current moment to obtain CSI real-time data, and the CSI real-time data is used by the terminal to determine the second information in combination with the first information.
- the performing model deployment based on the second information includes any of the following:
- the second information is model information of a first model in a first CSI feedback network, and activate the first model corresponding to the model information on the second communication device, wherein the first CSI feedback network is selected by the first communication device from CSI feedback networks deployed by the first communication device based on CSI real-time data;
- the second information includes the quantized model parameters, dequantize the model parameters, deploy the dequantized model parameters on the second communication device, and activate the deployed model.
- the method also includes: sending first indication information to the first communication device, the first indication information being used to indicate that the second communication device is configured with the first model and has the ability to use the first model to complete CSI feedback work.
- the method also includes: receiving second indication information sent by the first communication device, the second indication information is used to instruct the first communication device to notify the second communication device to enter the model initialization process; sending third indication information to the first communication device, the third indication information is used to indicate that the second communication device has accepted the model initialization instruction and entered the model initialization process.
- the method further includes: sending confirmation information to the first communication device, the confirmation information including confirmation indication information for indicating that the second communication device has completed model initialization.
- the model information includes a model number and/or a model structure.
- an embodiment of the present disclosure proposes a first communication device, comprising at least one of a transceiver module and a processing module; wherein the first communication device is used to execute an optional implementation method of the first aspect.
- an embodiment of the present disclosure proposes a second communication device, comprising at least one of a transceiver module and a processing module; wherein the second communication device is used to execute an optional implementation method of the second aspect.
- an embodiment of the present disclosure provides a communication system, including:
- a first communication device configured as an optional implementation of the first aspect
- the second communication device is configured to execute an optional implementation of the aforementioned second aspect.
- an embodiment of the present disclosure proposes a communication device, comprising: one or more processors; wherein the processor is used to call instructions so that the communication device executes the optional implementation method of the aforementioned first aspect.
- an embodiment of the present disclosure proposes a communication device, comprising: one or more processors; wherein the processor is used to call instructions so that the communication device executes the optional implementation method of the aforementioned second aspect.
- an embodiment of the present disclosure proposes a storage medium, wherein the storage medium stores instructions, and when the instructions are executed on a communication device, the communication device executes optional implementation methods of the aforementioned first and second aspects.
- an embodiment of the present disclosure proposes a program product.
- the communication device executes the method described in the optional implementation manner of the first aspect and the second aspect.
- an embodiment of the present disclosure proposes a computer program, which, when executed on a computer, enables the computer to execute the method described in the optional implementation of the first and second aspects.
- an embodiment of the present disclosure provides a chip or a chip system, wherein the chip or the chip system includes a processing circuit configured to execute the method described in the optional implementation of the first and second aspects above.
- the first communication device, the second communication device, the communication system, the storage medium, the program product, the computer program, the chip or the chip system are all used to execute the method proposed in the embodiment of the present disclosure. Therefore, the beneficial effects that can be achieved can refer to the beneficial effects in the corresponding method, which will not be repeated here.
- the embodiments of the present disclosure propose a model initialization method and a device thereof.
- the terms such as model initialization method, information processing method, communication method, etc. can be replaced with each other, the terms such as model initialization device, information processing device, communication device, etc. can be replaced with each other, and the terms such as model initialization system, information processing system, communication system, etc. can be replaced with each other.
- each step in a certain embodiment can be implemented as an independent embodiment, and the steps can be arbitrarily combined.
- a solution after removing some steps in a certain embodiment can also be implemented as an independent embodiment, and the order of the steps in a certain embodiment can be arbitrarily exchanged.
- the optional implementation methods in a certain embodiment can be arbitrarily combined; in addition, the embodiments can be arbitrarily combined, for example, some or all of the steps of different embodiments can be arbitrarily combined, and a certain embodiment can be arbitrarily combined with the optional implementation methods of other embodiments.
- elements expressed in the singular form such as “a”, “an”, “the”, “above”, “said”, “aforementioned”, “this”, etc., may mean “one and only one", or “one or more”, “at least one”, etc.
- the noun after the article may be understood as a singular expression or a plural expression.
- plurality refers to two or more.
- the terms "at least one of”, “one or more”, “a plurality of”, “multiple”, etc. can be used interchangeably.
- "at least one of A and B", “A and/or B", “A in one case, B in another case”, “in response to one case A, in response to another case B”, etc. may include the following technical solutions according to the situation: in some embodiments, A (A is executed independently of B); in some embodiments, B (B is executed independently of A); in some embodiments, execution is selected from A and B (A and B are selectively executed); in some embodiments, A and B (both A and B are executed). When there are more branches such as A, B, C, etc., the above is also similar.
- the recording method of "A or B” may include the following technical solutions according to the situation: in some embodiments, A (A is executed independently of B); in some embodiments, B (B is executed independently of A); in some embodiments, execution is selected from A and B (A and B are selectively executed).
- A A is executed independently of B
- B B is executed independently of A
- execution is selected from A and B (A and B are selectively executed).
- prefixes such as “first” and “second” in the embodiments of the present disclosure are only used to distinguish different description objects, and do not constitute restrictions on the position, order, priority, quantity or content of the description objects.
- the statement of the description object refers to the description in the context of the claims or embodiments, and should not constitute unnecessary restrictions due to the use of prefixes.
- the description object is a "field”
- the ordinal number before the "field” in the "first field” and the "second field” does not limit the position or order between the "fields”
- the "first” and “second” do not limit whether the "fields” they modify are in the same message, nor do they limit the order of the "first field” and the "second field”.
- the description object is a "level”
- the ordinal number before the "level” in the “first level” and the “second level” does not limit the priority between the "levels”.
- the number of description objects is not limited by the ordinal number, and can be one or more. Taking the "first device” as an example, the number of "devices” can be one or more.
- the objects modified by different prefixes may be the same or different. For example, if the description object is "device”, then the “first device” and the “second device” may be the same device or different devices, and their types may be the same or different. For another example, if the description object is "information”, then the "first information” and the “second information” may be the same information or different information, and their contents may be the same or different.
- “including A”, “comprising A”, “used to indicate A”, and “carrying A” can be interpreted as directly carrying A or indirectly indicating A.
- terms such as “greater than”, “greater than or equal to”, “not less than”, “more than”, “more than or equal to”, “not less than”, “higher than”, “higher than or equal to”, “not lower than”, and “above” can be replaced with each other, and terms such as “less than”, “less than or equal to”, “not greater than”, “less than”, “less than or equal to”, “no more than”, “lower than”, “lower than or equal to”, “not higher than”, and “below” can be replaced with each other.
- devices and equipment may be interpreted as physical or virtual, and their names are not limited to the names recorded in the embodiments. In some cases, they may also be understood as “equipment”, “device”, “circuit”, “network element”, “node”, “function”, “unit”, “section”, “system”, “network”, “chip”, “chip system”, “entity”, “subject”, etc.
- network can be interpreted as devices included in the network, such as access network equipment, core network equipment, etc.
- access network device may also be referred to as “radio access network device (RAN device)", “base station (BS)”, “radio base station (radio base station)”, “fixed station” and in some embodiments may also be understood as “node”, “access point (access point)”, “transmission point (TP)”, “reception point (RP)”, “transmission and/or reception point (transmission/reception point, TRP)", “panel”, “antenna panel”, “antenna array”, “cell”, “macro cell”, “small cell”, “femto cell”, “pico cell”, “sector”, “cell group”, “serving cell”, “carrier”, “component carrier”, “bandwidth part (bandwidth part, BWP)", etc.
- RAN device radio access network device
- base station base station
- RP radio base station
- TRP transmission and/or reception point
- terminal or “terminal device” may be referred to as "user equipment (UE)", “user terminal (user terminal)”, “mobile station (MS)”, “mobile terminal (MT)", subscriber station, mobile unit, subscriber unit, wireless unit, remote unit, mobile device, wireless device, wireless communication device, remote device, mobile subscriber station, access terminal, mobile terminal, wireless terminal, remote terminal, handset, user agent, mobile client, client, etc.
- UE user equipment
- MS mobile station
- MT mobile terminal
- acquisition of data, information, etc. may comply with the laws and regulations of the country where the data is obtained.
- data, information, etc. may be obtained with the user's consent.
- FIG1 is a schematic diagram of the architecture of a communication system according to an embodiment of the present disclosure.
- the communication system may include, but is not limited to, a first communication device and a second communication device.
- the number and form of devices shown in FIG1 are only used as examples and do not constitute a limitation on the embodiment of the present disclosure. In actual applications, two or more first communication devices and two or more second communication devices may be included.
- the communication system 100 shown in FIG1 takes a first communication device 101 and a second communication device 102 as an example.
- the first communication device 101 may be a network device, and the second communication device 102 may be a terminal.
- the first communication device 101 is configured with a second model in the CSI feedback network
- the second communication device 102 is configured with a first model in the CSI feedback network.
- the second model is a decoder model in the CSI feedback network
- the first model is an encoder model in the CSI feedback network.
- the first communication device 101 may be a terminal, and the second communication device 102 may be a network device.
- the first communication device 101 is configured with a second model in the CSI feedback network
- the second communication device 102 is configured with a first model in the CSI feedback network.
- the second model is an encoder model in the CSI feedback network
- the first model is a decoder model in the CSI feedback network.
- the terminal in this article can be an entity on the user side for receiving or transmitting signals, such as a mobile phone. It can also be called a terminal, a user equipment (UE), a mobile station (MS), a mobile terminal (MT), etc.
- UE user equipment
- MS mobile station
- MT mobile terminal
- the terminal can be a car with communication function, a smart car, a mobile phone, a wearable device, a tablet computer (Pad), a computer with wireless transceiver function, a virtual reality (VR) terminal, an augmented reality (AR) terminal, a wireless terminal in industrial control (industrial control), a wireless terminal in self-driving, a wireless terminal in remote medical surgery, a wireless terminal in smart grid (smart grid), a wireless terminal in transportation safety (transportation safety), a wireless terminal in smart city (smart city), a wireless terminal in smart home (smart home), etc. At least one of the above.
- the embodiments of the present disclosure do not limit the specific technology and specific device form adopted by the terminal.
- the network device in this article may be an access network device.
- the access network device is, for example, a node or device that accesses a terminal device to a wireless network
- the access network device may include an evolved NodeB (eNB), a next generation evolved NodeB (ng-eNB), a next generation NodeB (gNB), a nodeB (NB), a home nodeB (HNB), a home evolved nodeB (HeNB), a wireless backhaul device, a radio network controller (RNC), a base station controller (BSC), a base transceiver station (BTS), a baseband unit (BBU), a mobile switching center, a base station in a 6G communication system, an open base station (Open RAN), a cloud base station (Cloud RAN), a base station in other communication systems, and at least one of an access node in a Wi-Fi system, but is not limited thereto.
- eNB evolved NodeB
- ng-eNB next generation evolved NodeB
- the technical solution of the present disclosure may be applicable to the Open RAN architecture.
- the interfaces between access network devices or within access network devices involved in the embodiments of the present disclosure may become internal interfaces of Open RAN, and the processes and information interactions between these internal interfaces may be implemented through software or programs.
- the access network device may be composed of a centralized unit (central unit, CU) and a distributed unit (distributed unit, DU), wherein the CU may also be called a control unit (control unit).
- the CU-DU structure may be used to split the protocol layer of the access network device, with some functions of the protocol layer being centrally controlled by the CU, and the remaining part or all of the functions of the protocol layer being distributed in the DU, and the DU being centrally controlled by the CU, but not limited to this.
- the communication system described in the embodiment of the present disclosure is for the purpose of more clearly illustrating the technical solution of the embodiment of the present disclosure, and does not constitute a limitation on the technical solution proposed in the embodiment of the present disclosure.
- a person of ordinary skill in the art can know that with the evolution of the system architecture and the emergence of new business scenarios, the technical solution proposed in the embodiment of the present disclosure is also applicable to similar technical problems.
- the following embodiments of the present disclosure may be applied to the communication system 100 shown in FIG1 , or part of the subject, but are not limited thereto.
- the subjects shown in FIG1 are examples, and the communication system may include all or part of the subjects in FIG1 , or may include other subjects other than FIG1 , and the number and form of the subjects are arbitrary, and the subjects may be physical or virtual, and the connection relationship between the subjects is an example, and the subjects may be connected or disconnected, and the connection may be in any manner, and may be a direct connection or an indirect connection, and may be a wired connection or a wireless connection.
- LTE Long Term Evolution
- LTE-A LTE-Advanced
- LTE-B LTE-Beyond
- SUPER 3G IMT-Advanced
- 4G the fourth generation mobile communication system
- 5G 5G new radio
- FAA Future Radio Access
- RAT New Radio
- NR New Radio
- NX New radio access
- the present invention relates to wireless communication systems such as LTE, Wi-Fi (X), Global System for Mobile communications (GSM (registered trademark)), CDMA2000, Ultra Mobile Broadband (UMB), IEEE 802.11 (Wi-Fi (registered trademark)), IEEE 802.16 (WiMAX (registered trademark)), IEEE 802.20, Ultra-WideBand (UWB), Bluetooth (registered trademark), Public Land Mobile Network (PLMN) network, Device to Device (D2D) system, Machine to Machine (M2M) system, Internet of Things (IoT) system, Vehicle to Everything (V2X), systems using other communication methods, and next-generation systems expanded based on them.
- PLMN Public Land Mobile Network
- D2D Device to Device
- M2M Machine to Machine
- IoT Internet of Things
- V2X Vehicle to Everything
- systems using other communication methods and next-generation systems expanded based on them.
- next-generation systems expanded based on them.
- a combination of multiple systems for example, a combination of
- Massive Multiple-Input Multiple-Output is a key enabling technology for the 5th Generation (5G) mobile communication system. It helps to achieve ultra-large bandwidth, ultra-high rate, and low-latency communication, and will play a key role in the development of future wireless communication systems and provide technical support for various future wireless communication scenarios.
- 5G 5th Generation
- network equipment such as base stations, also called BS
- CSI downlink channel state information
- TDD time division duplex
- FDD frequency division duplex
- the CSI feedback method is usually adopted, that is, the user equipment (User Equipment, UE) first obtains the downlink CSI through channel estimation technology, and then feeds back the downlink CSI data to the network equipment through the uplink.
- the data dimension of CSI is proportional to the number of receiving and transmitting antennas, and in large-scale MIMO systems, the number of transmitting and receiving antennas is large, resulting in a significant increase in the amount of CSI data. If the terminal directly transmits the complete CSI data feedback to the network equipment, it will require a huge uplink communication overhead, which will affect the efficient operation of the communication system. Therefore, achieving high-precision and low-overhead downlink CSI feedback in FDD mode is an important research direction in the field of massive MIMO technology.
- a diagram showing an example structure of a CSI feedback network is provided.
- the encoder model on the terminal side extracts features and compresses dimensions of the original CSI data, compresses the original CSI data into codewords with a smaller data volume, and then transmits the codeword information to the network device through an uplink feedback link to fully reduce feedback overhead.
- the codewords received by the network device are reconstructed by the decoder model to output CSI data of the original dimension.
- a certain scale of CSI sample data set is used to train the feedback network model composed of the encoder and decoder, and the difference between the reconstructed CSI data output by the decoder and the original CSI data input by the encoder is minimized, so that the feedback network model can fully learn the CSI data feature distribution of the current training set, and then when facing CSI data with the same or similar feature distribution, it can achieve CSI compression, feedback and reconstruction with lower error.
- the CSI feedback scheme based on deep learning effectively reduces the feedback overhead, and at the same time has great advantages in improving feedback accuracy and reducing computing costs.
- the deep learning-based CSI feedback technology solution usually uses an offline model training method (Offline Training), that is, before the neural network model is deployed in the actual communication system, a large-scale training data set collected in advance is used to obtain the feedback network model parameters after more training rounds. After the neural network model obtained by offline training is deployed in the actual system, it can better process real-time data that follows the same or similar distribution as the training data.
- offline model training method Offline Training
- the encoder model and decoder model in the trained CSI feedback network will be deployed in the terminal and network equipment respectively, and the two need to work together to achieve compression and reconstruction of CSI data.
- the CSI data features in different channel scenarios usually have large differences. If a neural network model is used to complete the CSI compression and reconstruction work in different channel scenarios, it is difficult to meet the requirements of CSI feedback accuracy. Therefore, in some embodiments, multiple models are trained using sample data in different channel scenarios to complete CSI feedback in corresponding channel scenarios, so the terminal and network equipment also need to deploy feedback network models in multiple channel scenarios at the same time.
- the deep learning-based CSI feedback network technology solution focuses more on the two stages of model training and model reasoning.
- the model training stage a larger training data set is used, and more training rounds are performed to obtain the feedback network model parameters. Since the distribution of CSI data features in different channel scenarios is quite different, different network models are usually required to be trained for different channel scenarios.
- the model reasoning stage the terminal and network equipment each use the trained encoder model and decoder model to complete the compression and reconstruction of real-time CSI data, thereby completing the CSI feedback work in the large-scale MIMO system.
- the collaborative work of the terminal and the network device is an important guarantee for the application of the neural network model to complete the downlink CSI feedback.
- the network model parameters obtained by the training will be deployed to the terminal and the network device.
- the terminal and the network device need to first complete the model initialization, that is, to interact with each other for information such as whether the CSI feedback network model is currently applied and the specific model to be used, so as to ensure that the terminal and the network device use mutually matching encoder models and decoder models suitable for the current channel environment to complete CSI compression and reconstruction.
- the terminal since the terminal has strong mobility, as shown in Figure 2b, when the terminal enters the coverage range of another network device from the coverage range of one network device, the terminal needs to reconfirm the relevant parameter information of the applied CSI feedback network model with the network device working with it. At the same time, when the terminal enters a new channel scenario, the encoder model under the corresponding channel scenario may not be deployed. It is necessary to confirm with the network device and complete the deployment of the model parameters before using the CSI feedback network model to complete the CSI feedback task.
- the CSI feedback network model can be based on an AI (Artificial Intelligence)/ML (Machine Learning) model.
- AI Artificial Intelligence
- ML Machine Learning
- the application of deep neural network models in large-scale MIMO systems to complete downlink CSI feedback requires the collaboration of terminals and network devices.
- the original CSI samples complete data compression on the terminal side and data reconstruction on the network device side. Only when the models of the terminal and the network device match can the downlink CSI feedback be effectively completed with high accuracy.
- the terminal may collaborate with different network devices at different times and in different channel scenarios to complete the tasks of CSI compression, feedback, and reconstruction.
- the CSI feedback network model initialization needs to be completed, that is, the terminal and the network device first determine the current need to apply the AI/ML model to complete the CSI feedback and the specific model to be used, and then use the corresponding model to complete the compression and reconstruction of the CSI. Only by determining a model initialization scheme between the terminal and the network device can the normal operation of the AI/ML model be ensured and the accuracy of the CSI feedback scheme based on artificial intelligence be guaranteed.
- the model initialization method and device provided in the embodiments of the present disclosure can realize model initialization between communication devices based on information interaction between communication devices, thereby ensuring the normal operation of the model used for CSI feedback and ensuring the accuracy of the CSI feedback solution based on artificial intelligence.
- Fig. 3 is an interactive schematic diagram of a model initialization method according to an embodiment of the present disclosure. As shown in Fig. 3, the model initialization method according to an embodiment of the present disclosure can be applied to a communication system 100, and the method includes but is not limited to the following steps.
- the first indication information is used to indicate that the second communication device 102 is configured with the first model and has the ability to use the first model to complete the CSI feedback work.
- the first model is a model for CSI feedback.
- the first model can be an encoder model in a CSI feedback network, or can also be a decoder model in a CSI feedback network.
- the first communication device 101 is a network device
- the second communication device 102 is a terminal
- the first model is an encoder model in a CSI feedback network.
- the first communication device 101 is configured with a decoder model in a CSI feedback network
- the second communication device 102 is configured with an encoder model in a CSI feedback network.
- the first communication device 101 may be a terminal, and the second communication device 102 may be a network device.
- the first communication device 101 is configured with an encoder model in a CSI feedback network
- the second communication device 102 is configured with a decoder model in a CSI feedback network.
- the terms “first indication information”, “model application request indication information”, “model application request indication”, “application request indication information”, “application request indication” and the like can be interchangeable.
- the first indication information can be an AI/ML model application request indication.
- the AI/ML model application request indication can be represented by Indicator init_request .
- the AI/ML model application request indication means that the terminal is configured with the first model in the CSI feedback network, that is, the AI/ML model is configured, and currently has the ability to use the AI/ML model to complete the CSI feedback work.
- CSI feedback network CSI feedback network model
- AI/ML model CSI feedback network model
- Step S3102 the second communication device 102 sends first information.
- the second communication device 102 sends the first information to the first communication device 101. In some embodiments, the first communication device 101 receives the first information. Optionally, the first communication device 101 receives the first information sent by the second communication device 102.
- the above model information may include but is not limited to a model number and/or a model structure, wherein the "model structure” is also called a “network structure", indicating the structure used to construct the model.
- the above model information may include a model number.
- the first communication device 101 and the second communication device 102 are configured with CSI feedback network models in multiple channel scenarios at the same time, and the model numbers of the CSI feedback network models in the multiple channel scenarios are unique, that is, the CSI feedback network models in different channel scenarios can be distinguished by the model number.
- the first information sent by the second communication device 102 includes the model number of the first model configured by the second communication device 102.
- the first indication information and the first information may be sent simultaneously by the second communication device 102.
- the second communication device 102 sends a message including the first indication information and the first information.
- the first indication information and the first information may be sent separately by the second communication device 102.
- the second communication device 102 first sends the first indication information and then sends the first information, or the second communication device 102 first sends the first information and then sends the first indication information. This disclosure does not limit this.
- the above-mentioned first indication information and/or the above-mentioned first information may be included in RRC (Radio Resource Control) signaling, such as dedicated RRC signaling, or may also be included in DCI (Downlink Control Information), or may also be included in other signaling.
- RRC Radio Resource Control
- DCI Downlink Control Information
- Step S3103 the first communication device 101 sends second indication information.
- the first communication device 101 sends the second indication information to the second communication device 102.
- the second communication device 102 receives the second indication information.
- the second communication device 102 receives the second indication information sent by the first communication device 101.
- the second indication information is used to instruct the first communication device 101 to notify the second communication device 102 to enter the model initialization process.
- the second indication information may be a model initialization indication, for example, it may be represented by Indicator initialize , which is used to instruct the first communication device 101 to notify the second communication device 102 to enter the model initialization process.
- the second indication information may instruct the network device to notify the terminal to enter the model initialization process.
- the first communication device 101 is a terminal and the second communication device 102 is a network device
- the second indication information may instruct the terminal to notify the network device to enter the model initialization process.
- the terms "second indication information", "model initialization indication”, “model initialization indication information”, "AI/ML model initialization indication”, "AI/ML model initialization indication information” and the like may be interchangeable.
- the second indication information may be included in RRC signaling, such as dedicated RRC signaling, or may be included in DCI, or may be included in other signaling.
- RRC signaling such as dedicated RRC signaling
- DCI DCI
- the present disclosure does not limit this and will not elaborate on it.
- Step S3104 the second communication device 102 sends third indication information.
- the second communication device 102 sends the third indication information to the first communication device 101.
- the first communication device 101 receives the third indication information.
- the first communication device 101 receives the third indication information sent by the second communication device 102.
- the third indication information is used to indicate that the second communication device 102 has accepted the model initialization instruction and entered the model initialization process.
- the first communication device 101 receives the third indication information sent by the second communication device 102, it can be said that the second communication device 102 has received the second indication information sent by the first communication device 101 and accepted the initialization instruction for the CSI feedback network model on the second communication device, and entered the model initialization process.
- the third indication information may be included in RRC signaling, such as dedicated RRC signaling, or may be included in DCI, or may be included in other signaling.
- RRC signaling such as dedicated RRC signaling
- DCI DCI
- the present disclosure does not limit this and will not elaborate on it.
- Step S3105 the first communication device 101 determines CSI real-time data.
- the above-mentioned CSI real-time data refers to the downlink CSI obtained by the terminal through channel estimation at the current moment, which can represent the CSI data characteristics in the current channel environment to a certain extent.
- CSI real-time data is the downlink CSI obtained by the terminal through channel estimation
- the model initialization initiator can be either a terminal or a network device
- different model initialization initiators will determine the implementation method of CSI real-time data in different ways.
- a possible implementation method for the first communication device 101 to determine CSI real-time data is as follows: the network device sends a pilot signal and data request indication information, the pilot signal is used for the terminal to perform channel estimation at the current moment to obtain CSI real-time data, and the data request indication information is used to indicate that the network device requests CSI real-time data from the terminal.
- the network device receives the quantized CSI real-time data, and dequantizes the received quantized CSI real-time data to obtain CSI real-time data.
- the network device sends the pilot signal and the data request indication information to the terminal.
- the terminal receives the pilot signal and the data request indication information, for example, the terminal receives the pilot signal and the data request indication information sent by the network device, and performs channel estimation at the current moment based on the pilot signal to obtain the CSI real-time data.
- the terminal sends the CSI real-time data, such as the terminal sends the CSI real-time data to the network device.
- the network device receives the CSI real-time data sent by the terminal, thereby obtaining the CSI real-time data, so that the network device can perform the next step of model selection based on the CSI real-time data.
- the CSI real-time data sent by the terminal is quantized data.
- the network device receives the quantized CSI real-time data, it needs to dequantize the received quantized CSI real-time data, so as to obtain the CSI real-time data obtained when the terminal performs channel estimation on the current channel environment.
- the above quantization refers to: the process of transforming a signal, sequence or weight value with continuous amplitude values into a discrete amplitude value, which can generally be divided into uniform quantization and non-uniform quantization.
- the original data needs to be converted into a bit stream before the terminal or network device transmits data.
- the codeword data output after the CSI data is compressed by the terminal's encoder, or the CSI feedback network model weight value sequence, its data form is usually higher-precision floating-point data, so it needs to be converted into a bit stream through data quantization before transmission.
- CSI real-time data is usually a multi-dimensional complex matrix, whose dimension is related to the specific system configuration such as the number of transmitting and receiving antennas, and each position in the matrix is a complex number.
- CSI data quantization means: transforming the real and imaginary amplitudes of the complex values in the downlink CSI real-time data into discrete values.
- the model initialization initiator is a network device, that is, when the network device sends the model initialization to the terminal side, the network device can also send data request indication information to the terminal, which is used to indicate that the network device requests CSI real-time data from the terminal to complete the next model selection.
- terms such as “data request indication information”, “data request indication”, “CSI real-time data request indication”, “CSI real-time data request indication information” and the like may be interchangeable.
- CSI real-time data may be interchangeable.
- the first communication device 101 when the first communication device 101 is a terminal, the second communication device 102 is a network device, and the first model is a decoder model in a CSI feedback network, a possible implementation manner for the first communication device 101 to determine the CSI real-time data is as follows: the network device sends a pilot signal. The terminal receives the pilot signal sent by the network device, performs channel estimation based on the pilot signal at the current moment, and obtains the CSI real-time data.
- the terminal when the model initialization initiator is the terminal, that is, when the terminal sends the model initialization to the network device side, the terminal can perform channel estimation based on the pilot signal sent by the network device at the current moment, thereby obtaining the above-mentioned CSI real-time data, which is convenient for the terminal to perform the next model selection based on the above-mentioned CSI real-time data.
- Step S3106 The first communication device 101 inputs the above CSI real-time data into a CSI feedback network deployed by the first communication device 101 .
- the CSI feedback network includes the first model and the second model, and the CSI feedback network compresses and reconstructs the CSI real-time data through the first model and the second model.
- the first communication device 101 may be a network device, and the second communication device 102 may be a terminal.
- the second communication device 102 is configured with an encoder model in a CSI feedback network for compressing the above-mentioned CSI real-time data;
- the first communication device 101 is configured with a decoder model in a CSI feedback network for reconstructing the CSI real-time data of the terminal after being compressed by the encoder model.
- the first communication device 101 may be a terminal, and the second communication device 102 may be a network device.
- the first communication device 101 is configured with an encoder model in a CSI feedback network for compressing the above-mentioned CSI real-time data;
- the second communication device 102 is configured with a decoder model in a CSI feedback network for reconstructing the CSI real-time data of the terminal after being compressed by the encoder model.
- one or more CSI feedback networks may be deployed on the first communication device 101.
- the multiple CSI feedback networks may be CSI feedback network models constructed based on different model structures, or may be CSI feedback network models applicable to multiple different channel environments.
- the CSI feedback network may compress and reconstruct the input CSI real-time data.
- the first communication device 101 deploys multiple CSI feedback networks
- the first communication device 101 may input the above CSI real-time data to the above multiple CSI feedback networks respectively.
- Each CSI feedback network compresses and reconstructs the input CSI real-time data.
- the CSI feedback network deployed on the first communication device 101 may be a trained (ie, trained) CSI feedback network model.
- the first communication device 101 may be a network device, and the second communication device 102 may be a terminal; or, in some embodiments, the first communication device 101 may be a terminal, and the second communication device 102 may be a network device.
- Step S3107 The first communication device 101 obtains a CSI output value output by the CSI feedback network.
- the CSI feedback network can compress and reconstruct the input CSI real-time data, and output the reconstructed CSI data.
- the CSI output value output by the above-mentioned CSI feedback network is the CSI data obtained after reconstruction, so that the first communication device 101 can obtain the CSI output value output by the above-mentioned CSI feedback network.
- the first communication device 101 may be a network device, and the second communication device 102 may be a terminal; or, in some embodiments, the first communication device 101 may be a terminal, and the second communication device 102 may be a network device.
- Step S3108 The first communication device 101 determines the feedback accuracy of the CSI feedback network according to the CSI output value and the CSI real-time data.
- the error between the CSI output value and the CSI real-time data can be calculated, and the feedback accuracy of the CSI feedback network can be determined based on the error.
- the larger the error the lower the feedback accuracy of the CSI feedback network, and the smaller the error, the higher the feedback accuracy of the CSI feedback network.
- the mapping relationship between the error and the feedback accuracy can be determined based on the calculated error between the CSI output value and the CSI real-time data, and the corresponding feedback accuracy can be determined from the mapping relationship, that is, the feedback accuracy of the corresponding CSI feedback network.
- the first communication device 101 may be a network device, and the second communication device 102 may be a terminal; or, in some embodiments, the first communication device 101 may be a terminal, and the second communication device 102 may be a network device.
- Step S3109 The first communication device 101 selects a first CSI feedback network from the CSI feedback networks deployed by the first communication device 101 according to the feedback accuracy.
- the deployed CSI feedback network may be directly determined as the first CSI feedback network.
- the first communication device 101 may select a CSI feedback network with the highest feedback accuracy from the multiple CSI feedback networks to determine as the first CSI feedback network.
- the calculation method of the feedback accuracy can refer to the description in the above steps and will not be repeated here.
- the first communication device 101 may be a network device, and the second communication device 102 may be a terminal; or, in some embodiments, the first communication device 101 may be a terminal, and the second communication device 102 may be a network device.
- Step S3110 The first communication device 101 determines second information based on the model information of the first model in the first CSI feedback network and the model information of the first model configured by the second communication device 102 .
- the second information may be carried in the model deployment indication information and sent.
- the second information may be model deployment information of the first model to be deployed in the second communication device.
- the above-mentioned second information may include a model number.
- the number of the selected model is included in the model number reported by the second communication device.
- the first communication device then gives the model number to the second communication device to facilitate the second communication device to activate the model corresponding to the model number.
- the second information may include a model number and model parameters.
- the first communication device will send the model parameters and the corresponding model number of the selected model to the second communication device to facilitate the second communication device to deploy the selected model and activate the model for subsequent application.
- the first communication device 101 determines the model information of the first model in the first CSI feedback network as the second information, wherein the model information of the first model configured by the second communication device 102 includes the model information of the first model in the first CSI feedback network.
- the model information of the first model configured by the second communication device 102 includes the model information of the first model in the above-mentioned first CSI feedback network
- the first model in the above-mentioned first CSI feedback network is configured on the second communication device 102
- the first communication device 101 can determine the model information of the first model in the above-mentioned first CSI feedback network as the second information, so as to facilitate the second communication device 102 to deploy the model based on the second information.
- the first communication device 101 quantizes the model parameters of the first model in the above-mentioned first CSI feedback network, and determines the above-mentioned model parameters after quantization as the second information, wherein the model information of the first model configured by the second communication device 102 does not include the model information of the first model in the above-mentioned first CSI feedback network.
- the model information of the first model configured by the second communication device 102 does not include the model information of the first model in the first CSI feedback network, it means that the first model in the first CSI feedback network is not configured on the second communication device 102, and the first communication device 101 can quantize the model parameters of the first model in the first CSI feedback network, and determine the quantized model parameters as the second information.
- the first communication device 101 may activate the second model associated with the first model on the first communication device 101, so that the first communication device 101 can perform CSI corresponding processing through the activated second model. That is, when the first communication device 101 determines which first model needs to be deployed on the second communication device 102, the first communication device 101 needs to activate the second model associated with the first model to be deployed on the second communication device 102, so that CSI feedback can be achieved through the activated first model and the second model.
- the model parameter quantization refers to: converting the weight sequence of the first model into a bit form. It can be understood that the dequantization process is the inverse process of the quantization process, which will not be described in detail here.
- the first communication device 101 determines the model information of the first model in the above-mentioned first CSI feedback network as the second information, wherein the model information of the first model configured by the second communication device 102 includes the model information of the first model in the above-mentioned first CSI feedback network; the first communication device 101 quantizes the model parameters of the first model in the above-mentioned first CSI feedback network, and determines the quantized model parameters as the second information, wherein the model information of the first model configured by the second communication device 102 does not include the model information of the first model in the above-mentioned first CSI feedback network.
- the first communication device 101 may determine the model information of the first model in the first CSI feedback network and the quantized model parameters as the second information. That is, the first communication device 101 may send the model information (such as the model number and/or model structure) and the quantized model parameters of the first model in the first CSI feedback network to the second communication device 102, so that the second communication device 102 can distinguish the newly deployed first model by number.
- the model information such as the model number and/or model structure
- the first communication device 101 can activate the second model associated with the above-mentioned first model on the first communication device 101, so that the first communication device 101 can perform CSI corresponding processing through the activated second model.
- the first communication device 101 may be a network device, and the second communication device 102 may be a terminal; or, in some embodiments, the first communication device 101 may be a terminal, and the second communication device 102 may be a network device.
- Step S3111 the first communication device 101 sends second information.
- the second information may be model deployment information of the first model to be deployed in the second communication device.
- the second information may be carried in the model deployment indication information and sent.
- the first communication device 101 sends the model deployment indication information to the second communication device 102.
- the second communication device 102 receives the model deployment indication information.
- the second communication device 102 receives the model deployment indication information sent by the first communication device 101.
- the above-mentioned model deployment indication information includes the above-mentioned second information, and the above-mentioned model deployment indication information is used to indicate that the second communication device 102 can perform model deployment. Exemplarily, it can indicate that the second communication device 102 can perform model deployment based on the above-mentioned second information.
- Step S3112 The second communication device 102 receives the second information, and activates the first model corresponding to the model information.
- the second information may be model deployment information of the first model to be deployed in the second communication device.
- the second information may be model information of the first model in the first CSI feedback network.
- the second information may be carried in the model deployment indication information and sent.
- the first communication device 101 may be a network device, and the second communication device 102 may be a terminal.
- the second communication device 102 is configured with an encoder model in the CSI feedback network.
- the second communication device 102 receives model information of the first model in the above-mentioned first CSI feedback network, and the second communication device 102 activates the encoder model corresponding to the model information.
- the first communication device 101 may be a terminal, and the second communication device 102 may be a network device.
- the second communication device 102 is configured with a decoder model in the CSI feedback network.
- the second communication device 102 receives model information of the first model in the above-mentioned first CSI feedback network, and the second communication device 102 activates the decoder model corresponding to the model information.
- the second communication device 102 receives the model parameters of the first model in the first CSI feedback network, or receives the model parameters and model information of the first model in the first CSI feedback network.
- the second communication device 102 dequantizes the model parameters of the first model, sets the model number of the first model to the model number in the model information, and activates the first model.
- the first communication device 101 may be a network device, and the second communication device 102 may be a terminal.
- the second communication device 102 is configured with an encoder model in the CSI feedback network, and the second communication device 102 receives the model parameters of the first model in the above-mentioned first CSI feedback network, or receives the model parameters and model information (such as the model number Model opt ) of the first model in the above-mentioned first CSI feedback network, and the second communication device 102 activates the encoder model corresponding to the model number Model opt .
- the first communication device 101 may be a terminal, and the second communication device 102 may be a network device.
- the second communication device 102 is configured with a decoder model in the CSI feedback network.
- the second communication device 102 receives model information of the first model in the above-mentioned first CSI feedback network (such as model number Model opt ), and the second communication device 102 activates the decoder model corresponding to the model number Model opt .
- Step S3114 the second communication device 102 sends a confirmation message.
- the second communication device 102 sends the confirmation information to the first communication device 101.
- the first communication device 101 receives the confirmation information.
- the first communication device 101 receives the confirmation information sent by the second communication device 102.
- the confirmation information includes confirmation indication information for indicating that the second communication device 102 has completed the model initialization.
- the confirmation indication information can be represented by Indicator complete .
- the first communication device 101 may be a network device
- the second communication device 102 may be a terminal.
- the terminal may use a newly deployed encoder model to send CSI
- the network device may use a decoder model corresponding to the newly deployed encoder model of the terminal to receive CSI.
- the first communication device 101 may be a terminal
- the second communication device 102 may be a network device.
- the network device has newly deployed a decoder model
- the terminal may use the encoder model corresponding to the newly deployed decoder model of the network device to send CSI
- the network device may use the newly deployed decoder to receive CSI.
- the names of information, etc. are not limited to the names recorded in the embodiments, and terms such as “information”, “message”, “signal”, “signaling”, “report”, “configuration”, “indication”, “instruction”, “command”, “channel”, “parameter”, “domain”, “field”, “symbol”, “symbol”, “code element”, “codebook”, “codeword”, “codepoint”, “bit”, “data”, “program”, and “chip” can be used interchangeably.
- terms such as “uplink”, “uplink”, “physical uplink” can be interchangeable, and terms such as “downlink”, “downlink”, “physical downlink” can be interchangeable, and terms such as “side”, “sidelink”, “side communication”, “sidelink communication”, “direct connection”, “direct link”, “direct communication”, “direct link communication” can be interchangeable.
- obtain can be interchangeable, and can be interpreted as receiving from other entities, obtaining from protocols, obtaining from high levels, obtaining by self-processing, autonomous implementation, etc.
- terms such as “certain”, “preset”, “preset”, “set”, “indicated”, “some”, “any”, and “first” can be interchangeable, and "specific A”, “preset A”, “preset A”, “set A”, “indicated A”, “some A”, “any A”, and “first A” can be interpreted as A pre-defined in a protocol, etc., or as A obtained through setting, configuration, or indication, etc., and can also be interpreted as specific A, some A, any A, or first A, etc., but is not limited to this.
- the determination or judgment can be performed by a value represented by 1 bit (0 or 1), by a true or false value (Boolean value) represented by true or false, or by comparison of numerical values (for example, comparison with a predetermined value), but is not limited to this.
- step S3102+step S3105+step S3106+step S3107+step S3108+step S3109+step S3110+step S3111 can be implemented as an independent embodiment
- step S3101+step S3102+step S3105+step S3106+step S3107+step S3108+step S3109+step S3110+step S3111 can be implemented as an independent embodiment
- step S3102+step S3103+step S3104+step S3105+step S3106+step S3107+step S3108+step S3109+step S3110+step S3111 can be implemented as an independent embodiment.
- step S3102+step S3105+step S3106+step S3107+step S3108+step S3109+step S3110+step S3111+step S3112 can be implemented as an independent embodiment
- step S3101+step S3102+step S3105+step S3106+step S3107+step S3108+step S3109+step S3110+step S3111+step S3112 can be implemented as an independent embodiment
- step S3101+step S3102+step S3105+step S3106+step S3107+step S3108+step S3109+step S3110+step S3111+step S3112 can be implemented as an independent embodiment
- Step S3106+step S3107+step S3108+step S3109+step S3110+step S3111+step S3112 can be implemented as an independent embodiment
- step S3101, step S3102, and step S3103 may be executed in an order swapped or simultaneously, and step S3112 and step S3113 may be executed in an order swapped or simultaneously.
- step S3101, step S3103, step S3104, step S3112, step S3113, and step S3114 are optional, and one or more of these steps may be omitted or replaced in different embodiments.
- step S3101, step S3112, step S3113, and step S3114 are optional, and one or more of these steps may be omitted or replaced in different embodiments.
- step S3101, step S3113, and step S3114 are optional, and one or more of these steps may be omitted or replaced in different embodiments.
- step S3101, step S3112, and step S3114 are optional, and one or more of these steps may be omitted or replaced in different embodiments.
- FIG4A is a flow chart of a model initialization method according to an embodiment of the present disclosure.
- the method involved in the embodiment of the present disclosure may be performed by a first communication device 101.
- the first communication device 101 is a network device
- the second communication device 102 in this embodiment may be a terminal. That is, the method may be described from the network device side.
- the above method may include but is not limited to the following steps.
- Step S4101 receiving first indication information.
- the second communication device 102 (such as a terminal) sends the first indication information to the first communication device 101.
- the first communication device 101 (such as a network device) receives the first indication information, and optionally, the network device receives the first indication information sent by the terminal.
- the first communication device 101 is a network device
- the second communication device 102 is a terminal
- the first model is an encoder model in a CSI feedback network.
- the first communication device 101 is configured with a decoder model in a CSI feedback network
- the second communication device 102 is configured with an encoder model in a CSI feedback network.
- Step S4102 receiving first information.
- Step S4103 sending second indication information.
- step S4103 can refer to the optional implementation of step S3103 in FIG. 3 and other related parts in the embodiment involved in FIG. 3 , which will not be described in detail here.
- Step S4104 receiving third indication information.
- the third indication information is used to indicate that the second communication device 102 (such as a terminal) has accepted the model initialization instruction and entered the model initialization process.
- step S4104 can refer to the optional implementation of step S3104 in FIG. 3 and other related parts in the embodiment involved in FIG. 3 , which will not be described in detail here.
- Step S4105 sending a pilot signal and data request indication information.
- the pilot signal is used for the terminal to perform channel estimation at the current moment to obtain CSI real-time data.
- the data request indication information is used to indicate that the network device requests CSI real-time data from the terminal.
- the terminal receives the pilot signal and the data request indication information.
- the terminal receives the pilot signal and the data request indication information sent by the network device, and performs channel estimation at the current moment based on the pilot signal to obtain the CSI real-time data.
- the terminal sends the CSI real-time data, such as the terminal sends the CSI real-time data to the network device.
- the network device receives the CSI real-time data sent by the terminal, thereby obtaining the CSI real-time data, which facilitates the network device to perform the next step of model selection based on the CSI real-time data. It should be noted that after completing the model selection, the network device needs to activate the second model associated with the selected first model, so that the network device can reconstruct CSI through the activated second model.
- step S4105 can refer to the optional implementation of step S3105 in FIG. 3 and other related parts in the embodiment involved in FIG. 3 , which will not be described in detail here.
- Step S4106 receiving the quantized CSI real-time data.
- the network device receives the quantized CSI real-time data. In some embodiments, the terminal sends the quantized CSI real-time data to the network device, and the network device receives the quantized CSI real-time data sent by the terminal device.
- Step S4107 dequantize the received quantized CSI real-time data to obtain CSI real-time data.
- the network device may dequantize the received CSI real-time data that has been quantized to obtain CSI real-time data after the terminal device estimates the channel of the current channel environment.
- Step S4108 input the above CSI real-time data into the CSI feedback network deployed on the network device.
- the CSI feedback network includes the first model and the second model, and the CSI feedback network compresses and reconstructs the CSI real-time data through the first model and the second model.
- step S4108 can refer to the optional implementation of step S3106 in FIG3 and other related parts in the embodiment involved in FIG3 , which will not be described in detail here.
- Step S4109 Acquire the CSI output value output by the CSI feedback network.
- step S4109 can refer to the optional implementation of step S3107 in FIG. 3 and other related parts in the embodiment involved in FIG. 3 , which will not be described in detail here.
- Step S4110 determining the feedback accuracy of the CSI feedback network according to the CSI output value and the CSI real-time data.
- step S4110 can refer to the optional implementation of step S3108 in FIG. 3 and other related parts in the embodiment involved in FIG. 3 , which will not be described in detail here.
- Step S4111 selecting a first CSI feedback network from the CSI feedback networks deployed on the network device according to the feedback accuracy.
- step S4111 can refer to the optional implementation of step S3109 in FIG. 3 and other related parts in the embodiment involved in FIG. 3 , which will not be described in detail here.
- Step S4112 Determine second information based on the model information of the first model in the first CSI feedback network and the model information of the first model configured on the terminal.
- step S4112 can refer to the optional implementation of step S3110 in FIG. 3 and other related parts in the embodiment involved in FIG. 3 , which will not be described in detail here.
- Step S4113 sending the second information.
- the second information may be model deployment information of the first model to be deployed in the second communication device.
- the second information may be carried in the model deployment indication information and sent.
- the above-mentioned model deployment indication information includes the above-mentioned second information, and the above-mentioned model deployment indication information is used to indicate that the terminal can perform model deployment. Exemplarily, it can instruct the terminal to perform model deployment based on the above-mentioned second information.
- step S4113 can refer to the optional implementation of step S3111 in FIG. 3 and other related parts in the embodiment involved in FIG. 3 , which will not be described in detail here.
- Step S4114 receiving confirmation information.
- the confirmation information includes confirmation indication information for indicating that the terminal has completed model initialization.
- the confirmation indication information can be represented by Indicator complete .
- step S4114 can refer to the optional implementation of step S3114 in Figure 3 and other related parts in the embodiment involved in Figure 3, which will not be repeated here.
- step S4102+step S4105+step S4106+step S4107+step S4108+step S4109+step S4110+step S4111+step S4112+step S4113 can be implemented as an independent embodiment
- step S4102+step S4105+step S4106+step S4107+step S4108+step S4109+step S4110+step S4111+step S4112+step S4113+step S4114 can be implemented as an independent embodiment
- step S4102+step S4103+step S4104 +Step S4105+Step S4106+Step S4107+Step S4108+Step S4109+Step S4110+Step S4111+Step S4112+Step S4113 can be implemented as an independent embodiment
- step S4101, step S4102, and step S4103 may be executed in an interchangeable order or simultaneously.
- step S4101, step S4103, step S4104, and step S4114 are optional, and one or more of these steps may be omitted or replaced in different embodiments.
- step S4101, step S4103, and step S4104 are optional, and one or more of these steps may be omitted or replaced in different embodiments.
- step S4101 and step S4114 are optional, and one or more of these steps may be omitted or replaced in different embodiments.
- step S4114 is optional, and one or more of these steps may be omitted or replaced in different embodiments.
- step S4101 is optional, and one or more of these steps may be omitted or replaced in different embodiments.
- step S4103, step S4104, and step S4114 are optional, and one or more of these steps may be omitted or replaced in different embodiments.
- step S4103 and step S4104 are optional, and one or more of these steps may be omitted or replaced in different embodiments.
- FIG4B is a flow chart of a model initialization method according to an embodiment of the present disclosure.
- the method involved in the embodiment of the present disclosure may be performed by a first communication device 101.
- the first communication device 101 is a terminal
- the second communication device 102 in this embodiment is a network device, that is, the method may be described from the terminal side.
- the above method may include but is not limited to the following steps.
- the second communication device 102 (such as a network device) sends the first indication information to the first communication device 101 (such as a terminal).
- the first communication device 101 (such as a terminal) receives the first indication information.
- the terminal receives the first indication information sent by the network device.
- the first indication information is used to indicate that the second communication device 102 is configured with the first model and has the ability to use the first model to complete the CSI feedback work.
- the first model is a model for CSI feedback.
- the first model can be an encoder model in a CSI feedback network, or can also be a decoder model in a CSI feedback network.
- step S4201 can refer to the optional implementation of step S3101 in FIG. 3 and other related parts in the embodiment involved in FIG. 3 , which will not be described in detail here.
- Step S4202 receiving first information.
- the first information includes model information of a first model configured by the second communication device 102 (eg, a network device).
- the second communication device 102 eg, a network device.
- the first model please refer to the relevant description of step S4101, which will not be repeated here.
- step S4202 can refer to the optional implementation of step S3102 in FIG. 3 and other related parts in the embodiment involved in FIG. 3 , which will not be described in detail here.
- Step S4203 sending the second indication information.
- the second indication information is used to instruct the first communication device 101 (such as a terminal) to notify the second communication device 102 (such as a network device) to enter a model initialization process.
- step S4203 can refer to the optional implementation of step S3103 in FIG. 3 and other related parts in the embodiment involved in FIG. 3 , which will not be described in detail here.
- Step S4204 receiving third indication information.
- the third indication information is used to indicate that the second communication device 102 (eg, a network device) has accepted the model initialization instruction and entered the model initialization process.
- step S4204 can refer to the optional implementation of step S3104 in FIG. 3 and other related parts in the embodiment involved in FIG. 3 , which will not be described in detail here.
- Step S4205 receiving a pilot signal.
- the terminal device receives the pilot signal. In some embodiments, the network device sends the pilot signal, and the terminal receives the pilot signal sent by the network device.
- the pilot signal is used by the terminal to perform channel estimation at the current moment to obtain CSI real-time data.
- step S4205 can refer to the optional implementation of step S3105 in FIG. 3 and other related parts in the embodiment involved in FIG. 3 , which will not be described in detail here.
- Step S4206 Perform channel estimation based on the pilot signal at the current moment to obtain CSI real-time data.
- step S4206 can refer to the optional implementation of step S3105 in FIG. 3 and other related parts in the embodiment involved in FIG. 3 , which will not be described in detail here.
- Step S4207 input the above CSI real-time data into the CSI feedback network deployed on the terminal.
- the CSI feedback network includes the first model and the second model, and the CSI feedback network compresses and reconstructs the CSI real-time data through the first model and the second model.
- step S4207 can refer to the optional implementation of step S3106 in FIG3 and other related parts in the embodiment involved in FIG3 , which will not be described in detail here.
- Step S4208 Acquire the CSI output value output by the CSI feedback network.
- step S4208 can refer to the optional implementation of step S3107 in FIG. 3 and other related parts in the embodiment involved in FIG. 3 , which will not be described in detail here.
- Step S4209 determining the feedback accuracy of the CSI feedback network according to the CSI output value and the CSI real-time data.
- step S4209 can refer to the optional implementation of step S3108 in FIG. 3 and other related parts in the embodiment involved in FIG. 3 , which will not be described in detail here.
- Step S4210 Select a first CSI feedback network from the CSI feedback networks deployed on the terminal according to the feedback accuracy.
- step S4210 can refer to the optional implementation of step S3109 in Figure 3 and other related parts in the embodiment involved in Figure 3, which will not be repeated here.
- step S4211 can refer to the optional implementation of step S3110 in FIG. 3 and other related parts in the embodiment involved in FIG. 3 , which will not be described in detail here.
- Step S4212 sending the second information.
- the second information may be model deployment information of the first model to be deployed in the second communication device.
- the second information may be carried in the model deployment indication information and sent.
- the above-mentioned model deployment indication information includes the above-mentioned second information, and the above-mentioned model deployment indication information is used to indicate that the network device can perform model deployment. Exemplarily, it can instruct the network device to perform model deployment based on the above-mentioned second information.
- step S4212 can refer to the optional implementation of step S3111 in FIG. 3 and other related parts in the embodiment involved in FIG. 3 , which will not be described in detail here.
- Step S4213 receiving confirmation information.
- the confirmation information includes confirmation indication information for indicating that the network device has completed model initialization.
- the confirmation indication information can be represented by Iddicator complete .
- step S4213 can refer to the optional implementation of step S3114 in FIG. 3 and other related parts in the embodiment involved in FIG. 3 , which will not be described in detail here.
- step S4202+step S4205+step S4206+step S4207+step S4208+step S4209+step S4210+step S4211+step S4212 can be implemented as an independent embodiment
- step S4202+step S4205+step S4206+step S4207+step S4208+step S4209+step S4210+step S4211+step S4212+step S4213 can be implemented as an independent embodiment
- step S4202+step S4203+step S4204 +Step S4205+Step S4206+Step S4207+Step S4208+Step S4209+Step S4210+Step S4211+Step S4212 can be implemented as an independent embodiment
- step S4202+Step S4203+Step S4204+Step S4205+Step S4206+Step S4207+Step S4208+Step S4209+Step S4210+Step S4211+Step S4212 can be implemented as an independent embodiment
- step S4201, step S4202, and step S4203 may be executed in an interchangeable order or simultaneously.
- step S4201, step S4203, step S4204, and step S4213 are optional, and one or more of these steps may be omitted or replaced in different embodiments.
- step S4201, step S4203, and step S4204 are optional, and one or more of these steps may be omitted or replaced in different embodiments.
- step S4201 and step S4213 are optional, and one or more of these steps may be omitted or replaced in different embodiments.
- step S4213 is optional, and one or more of these steps may be omitted or replaced in different embodiments.
- step S4201 is optional, and one or more of these steps may be omitted or replaced in different embodiments.
- step S4203, step S4204, and step S4213 are optional, and one or more of these steps may be omitted or replaced in different embodiments.
- step S4203 and step S4204 are optional, and one or more of these steps may be omitted or replaced in different embodiments.
- Fig. 4C is a flow chart of a model initialization method according to an embodiment of the present disclosure. As shown in Fig. 4C, the method involved in the embodiment of the present disclosure may be executed by the first communication device 101. The above method may include but is not limited to the following steps.
- Step S4301 receiving first indication information.
- step S4301 can refer to the optional implementation of step S3101 in FIG. 3 and other related parts in the embodiment involved in FIG. 3 , which will not be described in detail here.
- Step S4302 receiving first information.
- the first information includes model information of a first model configured by the second communication device 102.
- model information of a first model configured by the second communication device 102.
- step S4302 can refer to the optional implementation of step S3102 in FIG. 3 and other related parts in the embodiment involved in FIG. 3 , which will not be described in detail here.
- Step S4303 sending second indication information.
- the second indication information is used to instruct the first communication device 101 to notify the second communication device 102 to enter the model initialization process.
- step S4303 can refer to the optional implementation of step S3103 in FIG. 3 and other related parts in the embodiment involved in FIG. 3 , which will not be described in detail here.
- Step S4304 receiving third indication information.
- the third indication information is used to indicate that the second communication device 102 has accepted the model initialization instruction and entered the model initialization process.
- step S4304 can refer to the optional implementation of step S3104 in FIG. 3 and other related parts in the embodiment involved in FIG. 3 , which will not be described in detail here.
- Step S4305 determine CSI real-time data.
- the above-mentioned CSI real-time data refers to the downlink CSI obtained by the terminal through channel estimation at the current moment, which can represent the CSI data characteristics in the current channel environment to a certain extent.
- step S4305 can refer to the optional implementation of step S3105 in FIG. 3 and other related parts in the embodiment involved in FIG. 3 , which will not be described in detail here.
- Step S4306 input the above CSI real-time data into the CSI feedback network deployed by the first communication device 101 .
- step S4306 can refer to the optional implementation of step S3106 in FIG3 and other related parts in the embodiment involved in FIG3 , which will not be described in detail here.
- Step S4307 Acquire the CSI output value output by the CSI feedback network.
- step S4307 can refer to the optional implementation of step S3107 in Figure 3 and other related parts in the embodiment involved in Figure 3, which will not be repeated here.
- Step S4308 Determine the feedback accuracy of the CSI feedback network according to the CSI output value and the CSI real-time data.
- step S4308 can refer to the optional implementation of step S3108 in Figure 3 and other related parts in the embodiment involved in Figure 3, which will not be repeated here.
- Step S4309 selecting a first CSI feedback network from the CSI feedback networks deployed by the first communication device 101 according to the feedback accuracy.
- step S4309 can refer to the optional implementation of step S3109 in Figure 3 and other related parts in the embodiment involved in Figure 3, which will not be repeated here.
- Step S4310 Determine second information based on the model information of the first model in the first CSI feedback network and the model information of the first model configured by the second communication device 102.
- step S4310 can refer to the optional implementation of step S3110 in Figure 3 and other related parts in the embodiment involved in Figure 3, which will not be repeated here.
- Step S4311 sending the second information.
- the second information may be model deployment information of the first model to be deployed in the second communication device.
- the second information may be carried in the model deployment indication information and sent.
- the above-mentioned model deployment indication information includes the above-mentioned second information, and the above-mentioned model deployment indication information is used to indicate that the second communication device 102 can perform model deployment. Exemplarily, it can indicate that the second communication device 102 can perform model deployment based on the above-mentioned second information.
- step S4311 can refer to the optional implementation of step S3111 in FIG. 3 and other related parts in the embodiment involved in FIG. 3 , which will not be described in detail here.
- Step S4312 receiving confirmation information.
- the confirmation information includes confirmation indication information for indicating that the second communication device 102 has completed the model initialization.
- the confirmation indication information can be represented by Indicator complete .
- step S4312 can refer to the optional implementation of step S3114 in Figure 3 and other related parts in the embodiment involved in Figure 3, which will not be repeated here.
- step S4302+step S4305+step S4306+step S4307+step S4308+step S4309+step S4310+step S4311 can be implemented as an independent embodiment
- step S4301+step S4302+step S4305+step S4306+step S4307+step S4308+step S4309+step S4310+step S4311 can be implemented as an independent embodiment
- step S4302+step S4303+step S4304 +Step S4305+Step S4306+Step S4307+Step S4308+Step S4309+Step S4310+Step S4311 can be implemented as an independent embodiment
- step S4301+Step S4302+Step S4303+Step S4304+Step S4305+Step S4306+Step S4307+Step S4308+Step S4309+Step S4310+Step S4311 can be implemented as an independent embodiment
- step S4301, step S4303, step S4304, and step S4312 are optional, and one or more of these steps may be omitted or replaced in different embodiments.
- step S4301 and step S4312 are optional, and one or more of these steps may be omitted or replaced in different embodiments.
- step S4301, step S4303, and step S4304 are optional, and one or more of these steps may be omitted or replaced in different embodiments.
- step S4301 is optional, and one or more of these steps may be omitted or replaced in different embodiments.
- step S4303, step S4304, and step S4312 are optional, and one or more of these steps may be omitted or replaced in different embodiments.
- step S4303 and step S4304 are optional, and one or more of these steps may be omitted or replaced in different embodiments.
- step S4312 is optional, and one or more of these steps may be omitted or replaced in different embodiments.
- Fig. 5A is a flow chart of a model initialization method according to an embodiment of the present disclosure.
- the model initialization method according to an embodiment of the present disclosure may be executed by a second communication device 102, where exemplarily, the second communication device 102 is a terminal, and the first communication device 101 in this embodiment is a network device.
- the above method may include but is not limited to the following steps.
- Step S5101 sending first indication information.
- the first indication information is used to indicate that the second communication device 102 is configured with the first model and has the ability to use the first model to complete the CSI feedback work.
- the first model is a model for CSI feedback.
- the first model can be an encoder model in a CSI feedback network, or can also be a decoder model in a CSI feedback network.
- step S5101 can refer to the optional implementation of step S3101 in FIG. 3 and other related parts in the embodiment involved in FIG. 3 , which will not be described in detail here.
- Step S5102 sending the first information.
- the first information includes model information of a first model configured by the second communication device 102 .
- step S5102 can refer to the optional implementation of step S3102 in FIG. 3 and other related parts in the embodiment involved in FIG. 3 , which will not be described in detail here.
- Step S5103 receiving second indication information.
- the second indication information is used to instruct the first communication device 101 to notify the second communication device 102 to enter the model initialization process.
- step S5103 can refer to the optional implementation of step S3103 in FIG. 3 and other related parts in the embodiment involved in FIG. 3 , which will not be described in detail here.
- Step S5104 sending third indication information.
- the third indication information is used to indicate that the second communication device 102 has accepted the model initialization instruction and entered the model initialization process.
- the first communication device 101 receives the third indication information sent by the second communication device 102, it can be said that the second communication device 102 has received the second indication information sent by the first communication device 101 and accepted the initialization instruction for the CSI feedback network model on the second communication device, and entered the model initialization process.
- step S5104 can refer to the optional implementation of step S3104 in FIG. 3 and other related parts in the embodiment involved in FIG. 3 , which will not be described in detail here.
- Step S5105 receiving a pilot signal and data request indication information.
- the pilot signal is used by the terminal to perform channel estimation at the current moment to obtain CSI real-time data.
- the above-mentioned data request indication information is used to instruct the network device to request CSI real-time data from the terminal.
- step S5105 can refer to the optional implementation of step S3105 in FIG. 3 and other related parts in the embodiment involved in FIG. 3 , which will not be described in detail here.
- Step S5106 Perform channel estimation at the current moment based on the pilot signal to obtain CSI real-time data.
- step S5106 can refer to the optional implementation of step S3105 in FIG. 3 and other related parts in the embodiment involved in FIG. 3 , which will not be described in detail here.
- Step S5107 sending the above CSI real-time data.
- step S5107 can refer to the optional implementation of step S3105 in FIG. 3 and other related parts in the embodiment involved in FIG. 3 , which will not be described in detail here.
- the CSI real-time data is used by the network device to select a model for the next step based on the CSI real-time data.
- the optional implementation method thereof can refer to the optional implementation methods of step S3105, step S3106, step S3107, step S3108, step S3109 and step S3110 of FIG. 3, and other related parts in the embodiment involved in FIG. 3, which will not be repeated here.
- Step S5108 receiving the second information.
- the second information may be model deployment information of the first model to be deployed in the second communication device.
- the second information may be carried in the model deployment indication information and sent.
- the above-mentioned model deployment indication information includes the above-mentioned second information, and the above-mentioned model deployment indication information is used to indicate that the second communication device 102 can perform model deployment. Exemplarily, it can indicate that the second communication device 102 can perform model deployment based on the above-mentioned second information.
- step S5108 can refer to the optional implementation of step S3111 in FIG. 3 and other related parts in the embodiment involved in FIG. 3 , which will not be described in detail here.
- Step S5109 upon receiving the model information of the first model in the first CSI feedback network, the second communication device 102 activates the first model corresponding to the model information.
- step S5109 can refer to the optional implementation of step S3112 in FIG. 3 and other related parts in the embodiment involved in FIG. 3 , which will not be described in detail here.
- Step S5110 receiving the model parameters of the first model in the above-mentioned first CSI feedback network, or receiving the model parameters and model information of the first model in the above-mentioned first CSI feedback network, the second communication device 102 dequantizes the model parameters of the first model, sets the model number of the first model to the model number in the model information, and activates the first model.
- step S5110 can refer to the optional implementation of step S3113 in FIG. 3 and other related parts in the embodiment involved in FIG. 3 , which will not be described in detail here.
- Step S5111 send confirmation information.
- the confirmation information includes confirmation indication information for indicating that the second communication device 102 has completed the model initialization.
- the confirmation indication information can be represented by Indicator complete .
- step S5111 can refer to the optional implementation of step S3114 in FIG. 3 and other related parts in the embodiment involved in FIG. 3 , which will not be described in detail here.
- step S5102+step S5105+step S5106+step S5107+step S5108+step S5109 can be implemented as an independent embodiment
- step S5102+step S5105+step S5106+step S5107+step S5108+step S5110 can be implemented as an independent embodiment
- step S5102+step S5105+step S5106+step S5107+step S5108+step S5109+step S5111 can be implemented as an independent embodiment
- step S5102+step S5105+step S5106+step S5107+step S5108+step S5109+step S5111 can be implemented as an independent embodiment.
- Step S5108+step S5110+step S5111 can be implemented as an independent embodiment
- step S5101+step S5102+step S5105+step S5106+step S5107+step S5108+step S5109 can be implemented as an independent embodiment
- step S5101+step S5102+step S5105+step S5106+step S5107+step S5108+step S5110 can be implemented as an independent embodiment
- Step S5111 can be implemented as an independent embodiment
- step S5101+step S5102+step S5105+step S5106+step S5107+step S5108+step S5110+step S5111 can be implemented as an independent embodiment
- step S5101+step S5102+step S5103+step S5104+step S5105+step S5106+step S5107+step S5108+step S5109 can be implemented
- step S5101, step S5102, and step S5103 may be executed in an order swapped or simultaneously, and step S5112 and step S5113 may be executed in an order swapped or simultaneously.
- step S5101, step S5103, step S5104, step S5110, and step S5111 are optional, and one or more of these steps may be omitted or replaced in different embodiments.
- step S5101, step S5103, step S5104, step S5109, and step S5111 are optional, and one or more of these steps may be omitted or replaced in different embodiments.
- step S5103, step S5104, step S5110, and step S5111 are optional, and one or more of these steps may be omitted or replaced in different embodiments.
- step S5103, step S5104, step S5109, and step S5111 are optional, and one or more of these steps may be omitted or replaced in different embodiments.
- step S5101, step S5103, step S5104, and step S5109 are optional, and one or more of these steps may be omitted or replaced in different embodiments.
- step S5101, step S5103, step S5104, and step S5110 are optional, and one or more of these steps may be omitted or replaced in different embodiments.
- FIG5B is a flow chart of a model initialization method according to an embodiment of the present disclosure.
- the model initialization method according to an embodiment of the present disclosure may be performed by a second communication device 102.
- the second communication device 102 is a network device
- the first communication device 101 in this embodiment is a terminal.
- the above method may include but is not limited to the following steps.
- Step S5201 sending first indication information.
- the first indication information is used to indicate that the second communication device 102 is configured with the first model and has the ability to use the first model to complete the CSI feedback work.
- the first model is a model for CSI feedback.
- the first model can be an encoder model in a CSI feedback network, or can also be a decoder model in a CSI feedback network.
- step S5201 can refer to the optional implementation of step S3101 in FIG. 3 and other related parts in the embodiment involved in FIG. 3 , which will not be described in detail here.
- Step S5202 sending the first information.
- the first information includes model information of a first model configured by the second communication device 102 .
- step S5202 can refer to the optional implementation of step S3102 in FIG. 3 and other related parts in the embodiment involved in FIG. 3 , which will not be described in detail here.
- Step S5203 receiving second indication information.
- the second indication information is used to instruct the first communication device 101 to notify the second communication device 102 to enter the model initialization process.
- step S5203 can refer to the optional implementation of step S3103 in FIG. 3 and other related parts in the embodiment involved in FIG. 3 , which will not be described in detail here.
- Step S5204 sending third indication information.
- the third indication information is used to indicate that the second communication device 102 has accepted the model initialization instruction and entered the model initialization process.
- the first communication device 101 receives the third indication information sent by the second communication device 102, it can be said that the second communication device 102 has received the second indication information sent by the first communication device 101 and accepted the initialization instruction for the CSI feedback network model on the second communication device, and entered the model initialization process.
- step S5204 can refer to the optional implementation of step S3104 in FIG. 3 and other related parts in the embodiment involved in FIG. 3 , which will not be described in detail here.
- Step S5205 sending a pilot signal.
- the network device sends a pilot signal.
- the network device sends the pilot signal to the terminal.
- the terminal receives the pilot signal sent by the network device.
- the pilot signal is used by the terminal to perform channel estimation at the current moment to obtain CSI real-time data.
- step S5205 can refer to the optional implementation of step S3105 in FIG. 3 and other related parts in the embodiment involved in FIG. 3 , which will not be described in detail here.
- the above CSI real-time data is used by the terminal to select the next model.
- the next model selection can be performed based on the CSI real-time data.
- Its optional implementation method can refer to the optional implementation methods of step S3106, step S3107, step S3108, step S3109, and step S3110 of Figure 3, and other related parts in the embodiment involved in Figure 3, which will not be repeated here.
- Step S5206 receiving the second information.
- the second information may be model deployment information of the first model to be deployed in the second communication device.
- the second information may be carried in the model deployment indication information and sent.
- the above-mentioned model deployment indication information includes the above-mentioned second information, and the above-mentioned model deployment indication information is used to indicate that the second communication device 102 can perform model deployment. Exemplarily, it can indicate that the second communication device 102 can perform model deployment based on the above-mentioned second information.
- step S5206 can refer to the optional implementation of step S3111 in FIG. 3 and other related parts in the embodiment involved in FIG. 3 , which will not be described in detail here.
- Step S5207 upon receiving the model information of the first model in the first CSI feedback network, the second communication device 102 activates the first model corresponding to the model information.
- step S5207 can refer to the optional implementation of step S3112 in FIG. 3 and other related parts in the embodiment involved in FIG. 3 , which will not be described in detail here.
- Step S5208 receiving the model parameters of the first model in the above-mentioned first CSI feedback network, or receiving the model parameters and model information of the first model in the above-mentioned first CSI feedback network, the second communication device 102 dequantizes the model parameters of the first model, sets the model number of the first model to the model number in the model information, and activates the first model.
- step S5208 can refer to the optional implementation of step S3113 in FIG. 3 and other related parts in the embodiment involved in FIG. 3 , which will not be described in detail here.
- Step S5209 send confirmation information.
- the confirmation information includes confirmation indication information for indicating that the second communication device 102 has completed the model initialization.
- the confirmation indication information can be represented by Indicator complete .
- step S5209 can refer to the optional implementation of step S3114 in FIG. 3 and other related parts in the embodiment involved in FIG. 3 , which will not be described in detail here.
- step S5202+step S5205+step S5206+step S5207 can be implemented as an independent embodiment
- step S5202+step S5205+step S5206+step S5208 can be implemented as an independent embodiment
- step S5202+step S5205+step S5206+step S5207+step S5209 can be implemented as an independent embodiment
- step S5202+step S5205+step S5206+step S5208+step S5209 can be implemented as an independent embodiment
- step S5202+step S5205+step S5206+step S5208+step S5209 can be implemented as an independent embodiment
- step S5202+step S5205+step S5206+step S5208+step S5209 can be implemented as an independent embodiment.
- step S5201+step S5202+step S5205+step S5206+step S5207 can be implemented as an independent embodiment
- step S5201+step S5202+step S5205+step S5206+step S5207+step S5209 can be implemented as an independent embodiment
- step S5201+step S5202+step S5205+step S5206+step S5208 +Step S5209 can be implemented as an independent embodiment
- step S5201+step S5202+step S5203+step S5204+step S5205+step S5206+step S5207+step S5208+step S5209 can be implemented as an independent embodiment
- step S5201+step S5202+step S5203+step S5204+step S5205+step S5206+step S5207+step S5208+step S5209 can be implemented as an independent embodiment
- step S5201+step S5202+step S5203+step S5204+step S5205+step S5206+step S5207 can be implemented as
- step S5201, step S5202, and step S5203 may be executed in an exchanged order or simultaneously, and step S5207 and step S5208 may be executed in an exchanged order or simultaneously.
- step S5201, step S5203, step S5204, step S5208, and step S5209 are optional, and one or more of these steps may be omitted or replaced in different embodiments.
- step S5201, step S5203, step S5204, step S5207, and step S5209 are optional, and one or more of these steps may be omitted or replaced in different embodiments.
- step S5203, step S5204, step S5208, and step S5209 are optional, and one or more of these steps may be omitted or replaced in different embodiments.
- step S5203, step S5204, step S5207, and step S5209 are optional, and one or more of these steps may be omitted or replaced in different embodiments.
- step S5201, step S5203, step S5204, and step S5209 are optional, and one or more of these steps may be omitted or replaced in different embodiments.
- step S5201, step S5203, and step S5204 are optional, and one or more of these steps may be omitted or replaced in different embodiments.
- step S5201 is optional, and one or more of these steps may be omitted or replaced in different embodiments.
- step S5203 and step S5204 are optional, and one or more of these steps may be omitted or replaced in different embodiments.
- Fig. 5C is a flow chart of a model initialization method according to an embodiment of the present disclosure. As shown in Fig. 5C, the model initialization method according to an embodiment of the present disclosure may be executed by the second communication device 102, and the method may include but is not limited to the following steps.
- Step S5301 sending first indication information.
- the first indication information is used to indicate that the second communication device 102 is configured with the first model and has the ability to use the first model to complete the CSI feedback work.
- the first model is a model for CSI feedback.
- the first model can be an encoder model in a CSI feedback network, or can also be a decoder model in a CSI feedback network.
- step S5301 can refer to the optional implementation of step S3101 in Figure 3 and other related parts in the embodiment involved in Figure 3, which will not be repeated here.
- Step S5302 sending the first information.
- the first information includes model information of a first model configured by the second communication device 102 .
- step S5302 can refer to the optional implementation of step S3102 in FIG. 3 and other related parts in the embodiment involved in FIG. 3 , which will not be described in detail here.
- Step S5303 receiving second indication information.
- the second indication information is used to instruct the first communication device 101 to notify the second communication device 102 to enter the model initialization process.
- step S5303 can refer to the optional implementation of step S3103 in FIG. 3 and other related parts in the embodiment involved in FIG. 3 , which will not be described in detail here.
- Step S5304 sending third indication information.
- the third indication information is used to indicate that the second communication device 102 has accepted the model initialization instruction and entered the model initialization process.
- the first communication device 101 receives the third indication information sent by the second communication device 102, it can be said that the second communication device 102 has received the second indication information sent by the first communication device 101 and accepted the initialization instruction for the CSI feedback network model on the second communication device, and entered the model initialization process.
- step S5304 can refer to the optional implementation of step S3104 in FIG. 3 and other related parts in the embodiment involved in FIG. 3 , which will not be described in detail here.
- Step S5305 receiving the second information.
- the second information may be model deployment information of the first model to be deployed in the second communication device.
- the second information may be carried in the model deployment indication information and sent.
- the second communication device 102 receives the model deployment indication information.
- the second communication device 102 receives the model deployment indication information sent by the first communication device 101.
- the above-mentioned model deployment indication information includes second information, and the second information is model deployment information of the first model to be deployed in the second communication device.
- Its optional implementation method can refer to the optional implementation method of step S3111 of Figure 3 and other related parts of the embodiment involved in Figure 3, which will not be repeated here.
- Step S5306 upon receiving model information of a first model in the first CSI feedback network, the second communication device 102 activates the first model corresponding to the model information.
- the implementation of the first CSI feedback network may refer to the optional implementation of steps S3106 to S3109 in FIG. 3 , and other related parts in the embodiment involved in FIG. 3 , which will not be described in detail here.
- step S5306 can refer to the optional implementation of step S3112 in FIG. 3 and other related parts in the embodiment involved in FIG. 3 , which will not be described in detail here.
- Step S5307 receiving the model parameters of the first model in the first CSI feedback network, or receiving the model parameters and model information of the first model in the above-mentioned first CSI feedback network, the second communication device 102 dequantizes the model parameters of the first model, sets the model number of the first model to the model number in the model information, and activates the first model.
- step S5307 can refer to the optional implementation of step S3113 in FIG. 3 and other related parts in the embodiment involved in FIG. 3 , which will not be described in detail here.
- Step S5308 send confirmation information.
- the confirmation information includes confirmation indication information for indicating that the second communication device 102 has completed the model initialization.
- the confirmation indication information can be represented by Indicator complete .
- step S5308 can refer to the optional implementation of step S3114 in Figure 3 and other related parts in the embodiment involved in Figure 3, which will not be repeated here.
- step S5302+step S5305+step S5306 can be implemented as an independent embodiment
- step S5302+step S5305+step S5307 can be implemented as an independent embodiment
- step S5301+step S5302+step S5305+step S5306 can be implemented as an independent embodiment
- step S5301+step S5302+step S5305+step S5307 can be implemented as an independent embodiment
- step S5302+step S5305+step S5306 can be implemented as an independent embodiment
- step S5301+step S5302+step S5305+step S5307 can be implemented as an independent embodiment
- step S5302+step S5305+step Step S5306+step S5308 can be implemented as an independent embodiment
- step S5302+step S5305+step S5307+step S5308 can be implemented as an independent embodiment
- step S5301+step S5302+step S5305+step S5306+step S5308 can be implemented as an independent embodiment
- step S5301, step S5302, and step S5303 may be executed in an exchanged order or simultaneously, and step S5312 and step S5313 may be executed in an exchanged order or simultaneously.
- step S5301, step S5303, step S5304, step S5307, and step S5308 are optional, and one or more of these steps may be omitted or replaced in different embodiments.
- step S5301, step S5303, step S5304, step S5306, and step S5308 are optional, and one or more of these steps may be omitted or replaced in different embodiments.
- step S5303, step S5304, step S5307, and step S5308 are optional, and one or more of these steps may be omitted or replaced in different embodiments.
- step S5303, step S5304, step S5306, and step S5308 are optional, and one or more of these steps may be omitted or replaced in different embodiments.
- step S5301, step S5303, step S5304, and step S5306 are optional, and one or more of these steps may be omitted or replaced in different embodiments.
- step S5301, step S5303, step S5304, and step S5307 are optional, and one or more of these steps may be omitted or replaced in different embodiments.
- Fig. 6 is an interactive schematic diagram of a model initialization method according to an embodiment of the present disclosure. As shown in Fig. 6, the method involved in the embodiment of the present disclosure can be applied to a communication system 100, and the method includes but is not limited to the following steps.
- Step S6101 The second communication device 102 sends first information to the first communication device 101, where the first information includes model information of a first model configured by the second communication device, where the first model is a model for channel state information CSI feedback.
- step S6101 can refer to the optional implementation of step S3102 in FIG. 3 and other related parts in the embodiment involved in FIG. 3 , which will not be described in detail here.
- Step S6102 The first communication device 101 determines second information based on the first information.
- the second information may be model deployment information of the first model to be deployed in the second communication device 102 .
- the first communication device 101 determines CSI real-time data; based on the CSI real-time data and the first information, determines the second information.
- the optional implementation method thereof can refer to the optional implementation methods of step S3105, step S3106, step S3107, step S3108, step S3109, and step S3110 of FIG. 3, and other related parts in the embodiment involved in FIG. 3, which will not be repeated here.
- Step S6103 The first communication device 101 sends second information to the second communication device 102, so that the second communication device 102 performs model deployment based on the second information.
- the second information may be model deployment information of the first model to be deployed in the second communication device.
- the second information may be carried in the model deployment indication information and sent.
- step S6103 can refer to the optional implementation of step S3111 in FIG. 3 and other related parts in the embodiment involved in FIG. 3 , which will not be described in detail here.
- Step S6104 The second communication device 102 performs model deployment based on the second information.
- the second information is determined to be model information of a first model in a first CSI feedback network, and the first model corresponding to the model information is activated on the second communication device, wherein the first CSI feedback network is selected by the first communication device from the CSI feedback networks deployed by the first communication device based on CSI real-time data.
- the optional implementation method thereof can refer to the optional implementation method of step S3112 of FIG. 3 and other related parts of the embodiment involved in FIG. 3, which will not be described in detail here.
- the second information includes quantized model parameters
- the model parameters are dequantized
- the dequantized model parameters are deployed on the second communication device to activate the first model.
- the optional implementation method thereof can refer to the optional implementation method of step S3113 of FIG. 3 and other related parts of the embodiment involved in FIG. 3, which will not be repeated here.
- the second information is determined to be model information of a first model in a first CSI feedback network, and the first model corresponding to the model information is activated on the second communication device, wherein the first CSI feedback network is selected by the first communication device from CSI feedback networks deployed by the first communication device based on CSI real-time data.
- the second information is determined to include quantized model parameters, the model parameters are dequantized, and the dequantized model parameters are deployed on the second communication device, and the first model is activated.
- the above method may include the method described in the above embodiments with the first communication device, the second communication device, etc., which will not be repeated here.
- FIG7 is a schematic diagram of a model initialization method according to an embodiment of the present disclosure. As shown in FIG7 , the method of the embodiment of the present disclosure can be applied to a network device. The method includes:
- Step S7101 receiving the AI/ML model application request indication Indicator init_request reported by the terminal, the AI/ML model type and parameters for CSI feedback, etc.
- the indication means that the terminal is configured with the AI/ML model and currently has the ability to use the AI/ML model to complete the CSI feedback work; the information includes: the number Model 1 , Model 2 , ..., Model n of the CSI feedback network model configured by the terminal, the model structure, etc.
- Step S7102 sending an AI/ML model initialization indication Indicator initialize and a CSI real-time data request indication Indicator data to the terminal, wherein the AI/ML model initialization indication indicates that the network device notifies the terminal to enter the model initialization process, and the CSI real-time data request indication indicates that the network device requests real-time CSI data from the terminal to complete the next model selection, and the real-time data refers to: the downlink CSI obtained by the terminal through channel estimation at the current moment, which can represent the CSI data characteristics under the current channel environment to a certain extent.
- Step S7103 receiving the AI/ML model initialization acceptance indication Indicator init_accept and the real-time quantized CSI real-time data sent by the terminal, and dequantizing the received data.
- the model selection operation (the operation is a network device content behavior) can be implemented as follows: input CSI real-time data into the AI/ML model deployed by the network device, obtain the CSI output value after compression and reconstruction by the AI/ML model, calculate the error between the CSI output value and the original CSI real-time data, and then obtain the real-time feedback accuracy of the AI/ML model. Select the CSI feedback network model with the highest feedback accuracy and obtain its model number Model opt .
- Model n of the CSI feedback network model configured by the terminal reported by the terminal
- the network device sends a model deployment indication Indicator deploy to the terminal, the indication indicating that the terminal can perform model deployment, and the indication carries the model number Model opt .
- the network device quantizes the encoder model parameters corresponding to Model opt and sends a model deployment indication Indicator deploy , the quantized encoder model parameters and the model number Model opt to the terminal, wherein the indication indicates that the terminal can perform model deployment.
- Step S7104 receiving model initialization confirmation information sent by the terminal, wherein the confirmation information mainly includes: a confirmation indication Indicator complete indicating that the terminal has completed model initialization.
- Step S7105 use the decoder model corresponding to the model number Model opt selected by the network device to perform CSI reception.
- FIG8 is a schematic diagram of a model initialization method according to an embodiment of the present disclosure. As shown in FIG8 , the method of the embodiment of the present disclosure can be applied to a terminal. The method includes:
- Step S8101 reporting the AI/ML model application request indication Indicator init_request , the AI/ML model type and parameters used for CSI feedback to the network device.
- the indication means that the terminal is configured with the AI/ML model and currently has the ability to use the AI/ML model to complete the CSI feedback work; the information includes: the number Model 1 , Model 2 , ..., Model n of the CSI feedback network model configured by the terminal, the model structure and other information.
- Step S8102 Receive an AI/ML model initialization indication Indicator initialize and a CSI real-time data request indication Indicator data sent by a network device.
- Step S8103 obtain a certain amount of real-time CSI sample data through downlink channel estimation, and perform CSI data quantization processing (which is an internal behavior of the terminal).
- Step S8104 Report the AI/ML model initialization acceptance indication Indicator init_accept and real-time quantized CSI sample data to the network device.
- Step S8105 receiving a model deployment indication Indicator deploy , a model number Model opt or a quantized encoder model parameter sent by a network device.
- Model opt if a model number Model opt is received, activating an encoder model corresponding to Model opt ;
- the encoder model parameters and the model number Model opt are received, the encoder model parameters are dequantized and the model number is set to Model opt .
- the terminal receives the model number Model opt .
- the first model is activated.
- the second model such as the decoder model deployed on the network device and associated with the first model (such as the encoder model).
- the terminal receives the model number Model opt and the encoder model parameters, and dequantizes the encoder model parameters.
- the encoder model should also be activated.
- the network device completes the model selection, it is also necessary to activate the decoder model associated with the encoder model deployed on the network device.
- CSI feedback can be achieved through the activated encoder and decoder.
- Step S8106 Send model initialization confirmation information to the network device, where the confirmation information mainly includes: a confirmation indication Indicator complete that the terminal has completed model initialization.
- Step S8107 using the encoder model corresponding to the model number Model opt selected by the network device to send CSI.
- the embodiments of the present disclosure also propose a device for implementing any of the above methods, for example, a device is proposed, the above device includes a unit or module for implementing each step performed by the first communication device in any of the above methods.
- a device is also proposed, including a unit or module for implementing each step performed by the second communication device in any of the above methods.
- the division of the units or modules in the above device is only a division of logical functions, which can be fully or partially integrated into one physical entity or physically separated in actual implementation.
- the units or modules in the device can be implemented in the form of a processor calling software: for example, the device includes a processor, the processor is connected to a memory, and instructions are stored in the memory.
- the processor calls the instructions stored in the memory to implement any of the above methods or implement the functions of the units or modules of the above device, wherein the processor is, for example, a general-purpose processor, such as a central processing unit (CPU) or a microprocessor, and the memory is a memory inside the device or a memory outside the device.
- CPU central processing unit
- microprocessor a microprocessor
- the units or modules in the device may be implemented in the form of hardware circuits, and the functions of some or all of the units or modules may be implemented by designing the hardware circuits.
- the hardware circuits may be understood as one or more processors; for example, in one implementation, the hardware circuits are application-specific integrated circuits (ASICs), and the functions of some or all of the above units or modules may be implemented by designing the logical relationship of the components in the circuits; for another example, in another implementation, the hardware circuits may be implemented by programmable logic devices (PLDs), and Field Programmable Gate Arrays (FPGAs) may be used as an example, which may include a large number of logic gate circuits, and the connection relationship between the logic gate circuits may be configured by configuring the configuration files, thereby implementing the functions of some or all of the above units or modules. All units or modules of the above devices may be implemented in the form of software called by the processor, or in the form of hardware circuits, or in the form of software called by the processor, and the remaining part may be implemented in
- the processor is a circuit with information processing capability.
- the processor may be a circuit with instruction reading and running capability, such as a central processing unit (CPU), a microprocessor, a graphics processing unit (GPU) (which may be understood as a microprocessor), or a digital signal processor (DSP); in another implementation, the processor may implement certain functions through the logical relationship of a hardware circuit, and the logical relationship of the above hardware circuit may be fixed or reconfigurable, such as a hardware circuit implemented by an application-specific integrated circuit (ASIC) or a programmable logic device (PLD), such as an FPGA.
- ASIC application-specific integrated circuit
- PLD programmable logic device
- the process of the processor loading a configuration document to implement the hardware circuit configuration may be understood as the process of the processor loading instructions to implement the functions of some or all of the above units or modules.
- it can also be a hardware circuit designed for artificial intelligence, which can be understood as ASIC, such as Neural Network Processing Unit (NPU), Tensor Processing Unit (TPU), Deep Learning Processing Unit (DPU), etc.
- ASIC Neural Network Processing Unit
- NPU Neural Network Processing Unit
- TPU Tensor Processing Unit
- DPU Deep Learning Processing Unit
- FIG9A is a schematic diagram of the structure of the first terminal proposed in an embodiment of the present disclosure.
- the first communication device 9100 may include: at least one of a transceiver module 9101, a processing module 9102, etc.
- the transceiver module is used to receive the first information sent by the second communication device, the first information includes the model information of the first model configured by the second communication device, and the first model is a model for channel state information CSI feedback; the processing module is used to determine the second information based on the first information; the transceiver module is also used to send model deployment indication information to the second communication device, the model deployment indication information includes the second information, and the model deployment indication information is used to instruct the second communication device to perform model deployment based on the second information.
- the transceiver module is used to execute at least one of the communication steps such as sending and/or receiving performed by the first communication device 101 in any of the above methods (for example, step S3103, step S3111, but not limited to this), which will not be repeated here.
- the above processing module is used to execute at least one of the other steps (for example, step S3105, step S3106, step S3107, step S3108, step S3109, step S3110, step S3203, but not limited to these) performed by the second communication device 101 in any of the above methods, which are not repeated here.
- FIG9B is a schematic diagram of the structure of the second communication device proposed in an embodiment of the present disclosure.
- the second communication device 9200 may include: at least one of a transceiver module 9201, a processing module 9202, etc.
- the transceiver module is used to send first information to the first communication device, the first information includes model information of a first model configured by the second communication device, and the first model is a model for channel state information CSI feedback; the transceiver module is also used to receive second information sent by the first communication device, the second information is determined by the first communication device based on the first information, and the second information is model deployment information of the first model to be deployed in the second communication device; the processing module is used to perform model deployment based on the second information.
- the transceiver module is used to perform at least one of the communication steps such as sending and/or receiving performed by the second communication device 102 in any of the above methods (for example, step S3101, step S3102, step S3104, step S3114, but not limited to this), which will not be repeated here.
- the processing module is used to execute at least one of the other steps (such as step S3112 and step S3113, but not limited thereto) executed by the second communication device 102 in any of the above methods, which will not be described in detail here.
- the transceiver module may include a sending module and/or a receiving module, and the sending module and the receiving module may be separate or integrated.
- the transceiver module may be interchangeable with the transceiver.
- the processing module can be a module or include multiple submodules.
- the multiple submodules respectively execute all or part of the steps required to be executed by the processing module.
- the processing module can be replaced with the processor.
- FIG10A is a schematic diagram of the structure of a communication device 1100 proposed in an embodiment of the present disclosure.
- the communication device 1100 may be the first communication device described above, or the second communication device described above, or a chip, a chip system, or a processor that supports the first communication device to implement any of the above methods, or a chip, a chip system, or a processor that supports the second communication device to implement any of the above methods.
- the communication device 1100 may be used to implement the method described in the above method embodiment, and the details may refer to the description in the above method embodiment.
- the communication device 1100 includes one or more processors 1101.
- the processor 1101 may be a general-purpose processor or a dedicated processor, for example, a baseband processor or a central processing unit.
- the baseband processor may be used to process the communication protocol and the communication data
- the central processing unit may be used to control the communication device (such as a base station, a baseband chip, a terminal device, a terminal device chip, a DU or a CU, etc.), execute a program, and process the data of the program.
- the communication device 1100 is used to execute any of the above methods.
- the communication device 1100 further includes one or more memories 1102 for storing instructions.
- the memory 1102 may also be outside the communication device 1100.
- the communication device 1100 further includes one or more transceivers 1103.
- the transceiver 1103 performs at least one of the communication steps such as sending and/or receiving in the above method (for example, step S3103, step S3111, step S3101, step S3102, step S3104, step S3114, but not limited thereto), and the processor 1101 performs at least one of the other steps (for example, step S3105, step S3106, step S3107, step S3108, step S3109, step S3110, step S3203, step S3112, step S3113, but not limited thereto).
- the transceiver may include a receiver and/or a transmitter, and the receiver and the transmitter may be separate or integrated.
- the terms such as transceiver, transceiver unit, transceiver, transceiver circuit, etc. may be replaced with each other, the terms such as transmitter, transmission unit, transmitter, transmission circuit, etc. may be replaced with each other, and the terms such as receiver, receiving unit, receiver, receiving circuit, etc. may be replaced with each other.
- the communication device 1100 may include one or more interface circuits 1104.
- the interface circuit 1104 is connected to the memory 1102, and the interface circuit 1104 may be used to receive signals from the memory 1102 or other devices, and may be used to send signals to the memory 1102 or other devices.
- the interface circuit 1104 may read instructions stored in the memory 1102 and send the instructions to the processor 1101.
- the communication device 1100 described in the above embodiments may be the above-mentioned first communication device or the above-mentioned second communication device, but the scope of the communication device 1100 described in the present disclosure is not limited thereto, and the structure of the communication device 1100 may not be limited by FIG. 10A.
- the communication device may be an independent device or may be part of a larger device.
- the communication device may be: 1) an independent integrated circuit IC, or a chip, or a chip system or subsystem; (2) a collection of one or more ICs, optionally, the above-mentioned IC collection may also include a storage component for storing data and programs; (3) an ASIC, such as a modem; (4) a module that can be embedded in other devices; (5) a receiver, a terminal device, an intelligent terminal device, a cellular phone, a wireless device, a handheld device, a mobile unit, a vehicle-mounted device, a network device, a cloud device, an artificial intelligence device, etc.; (6) others, etc.
- Fig. 10B is a schematic diagram of the structure of the chip 1200 proposed in the embodiment of the present disclosure. If the communication device 1100 can be a chip or a chip system, please refer to the schematic diagram of the structure of the chip 1200 shown in Fig. 10B, but it is not limited thereto.
- the chip 1200 includes one or more processors 1201 , and the chip 1200 is configured to execute any of the above methods.
- the chip 1200 further includes one or more interface circuits 1202.
- the interface circuit 1202 is connected to the memory 1203.
- the interface circuit 1202 can be used to receive signals from the memory 1203 or other devices, and the interface circuit 1202 can be used to send signals to the memory 1203 or other devices.
- the interface circuit 1202 can read instructions stored in the memory 1203 and send the instructions to the processor 1201.
- the interface circuit 1202 executes at least one of the communication steps such as sending and/or receiving in the above method (for example, step S3103, step S3111, step S3101, step S3102, step S3104, step S3114, but not limited to this), and the processor 1201 executes at least one of the other steps (for example, step S3105, step S3106, step S3107, step S3108, step S3109, step S3110, step S3203, step S3112, step S3113, but not limited to this).
- the communication steps such as sending and/or receiving in the above method
- step S3103, step S3111, step S3101, step S3102, step S3104, step S3114 but not limited to this
- the processor 1201 executes at least one of the other steps (for example, step S3105, step S3106, step S3107, step S3108, step S3109, step S3110, step S3203, step S3112, step S3113, but not limited to this).
- interface circuit interface circuit
- transceiver pin transceiver
- the chip 1200 further includes one or more memories 1203 for storing instructions. Alternatively, all or part of the memory 1203 may be outside the chip 1200.
- the present disclosure also proposes a storage medium, on which instructions are stored, and when the instructions are executed on the communication device 1100, the communication device 1100 executes any of the above methods.
- the storage medium is an electronic storage medium.
- the storage medium is a computer-readable storage medium, but is not limited to this, and it can also be a storage medium readable by other devices.
- the storage medium can be a non-transitory storage medium, but is not limited to this, and it can also be a temporary storage medium.
- the present disclosure also proposes a program product, which, when executed by the communication device 1100, enables the communication device 1100 to execute any of the above methods.
- the program product is a computer program product.
- the present disclosure also proposes a computer program, which, when executed on a computer, causes the computer to execute any one of the above methods.
- the computer program product includes one or more computer programs.
- the computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device.
- the computer program can be stored in a computer-readable storage medium, or transmitted from one computer-readable storage medium to another computer-readable storage medium.
- the computer program can be transmitted from a website site, computer, server or data center by wired (e.g., coaxial cable, optical fiber, digital subscriber line (digital subscriber line, DSL)) or wireless (e.g., infrared, wireless, microwave, etc.) mode to another website site, computer, server or data center.
- the computer-readable storage medium can be any available medium that can be accessed by a computer or a data storage device such as a server or data center that includes one or more available media integrated.
- the available medium may be a magnetic medium (e.g., a floppy disk, a hard disk, a magnetic tape), an optical medium (e.g., a high-density digital video disc (DVD)), or a semiconductor medium (e.g., a solid state disk (SSD)), etc.
- a magnetic medium e.g., a floppy disk, a hard disk, a magnetic tape
- an optical medium e.g., a high-density digital video disc (DVD)
- DVD high-density digital video disc
- SSD solid state disk
Landscapes
- Engineering & Computer Science (AREA)
- Computer Networks & Wireless Communication (AREA)
- Signal Processing (AREA)
- Mobile Radio Communication Systems (AREA)
Abstract
本公开实施例公开了一种模型初始化方法及其装置。其中方法包括:第一通信设备接收第二通信设备发送的第一信息,第一信息包括由第二通信设备配置的第一模型的模型信息,第一模型为用于信道状态信息CSI反馈的模型;基于第一信息,确定第二信息,第二信息为待部署在第二通信设备中的第一模型的模型部署信息;向第二通信设备发送第二信息,以使第二通信设备基于第二信息进行模型部署。通过实施本公开实施例,可以实现通信设备之间的模型初始化,从而可以确保用于CSI反馈的模型的正常工作,保证基于人工智能的CSI反馈方案的精确度。
Description
本公开涉及通信技术领域,尤其涉及一种模型初始化方法及其装置。
大规模多输入多输出(Multiple-Input Multiple-Output,MIMO)是第五代移动通信(The 5th Generation,5G)系统的关键使能技术,有助于实现超大带宽、超高速率、低时延通信,并将在未来无线通信系统发展过程中发挥关键作用,为各种未来无线通信场景提供技术支持。
为实现大规模MIMO技术在系统容量、功率效率、频谱效率等方面的性能增益,基站(Base Station,BS)需要获取精确的下行信道状态信息(Channel State Information,CSI)。时分双工(Time Division Duplex,TDD)模式下,基站可以首先获取上行CSI,再利用信道互易性获取下行CSI。频分双工(Frequency Division Duplex,FDD)模式下,由于上行与下行信道工作在不同频段,信道互易性较弱,因此通常采用CSI反馈的方式,即用户设备(User Equipment,UE)首先通过信道估计技术获取下行CSI,再将下行CSI数据通过上行链路反馈至基站。基于深度学习(Deep Learning)技术实现FDD模式下的下行CSI反馈能够以有效降低反馈开销和计算复杂度,同时也能够显著提升CSI反馈精确度,已经成为大规模MIMO系统中解决CSI反馈问题的重要方法。
但是,相关技术中基于深度学习的CSI反馈网络技术方案通常关注模型训练与模型推理两个阶段,目前尚缺乏用于通信设备之间模型初始化的有效手段。
发明内容
本公开实施例提出了一种模型初始化方法及其装置。
根据本公开实施例的第一方面,提出了一种模型初始化方法,包括:
第一通信设备接收第二通信设备发送的第一信息,所述第一信息包括由所述第二通信设备配置的第一模型的模型信息,所述第一模型为用于信道状态信息CSI反馈的模型;
基于所述第一信息,确定第二信息,所述第二信息为待部署在所述第二通信设备中的所述第一模型的第二信息;
向所述第二通信设备发送所述第二信息,以使所述第二通信设备基于所述第二信息进行模型部署。
根据本公开实施例的第二方面,提出了一种模型初始化方法,包括:
第二通信设备向第一通信设备发送第一信息,所述第一信息包括所述第二通信设备所配置的第一模型的模型信息,所述第一模型为用于信道状态信息CSI反馈的模型;
接收所述第一通信设备发送的第二信息,所述第二信息为所述第一通信设备基于所述第一信息确定的,所述第二信息为待部署在所述第二通信设备中的所述第一模型的模型部署信息;
基于所述第二信息进行模型部署。
根据本公开实施例的第三方面,提出了一种第一通信装置,包括:
收发模块,用于接收第二通信设备发送的第一信息,所述第一信息包括所述第二通信设备所配置的第一模型的模型信息,所述第一模型为用于信道状态信息CSI反馈的模型;
处理模块,用于基于所述第一信息,确定第二信息;
所述收发模块,还用于向所述第二通信设备发送第二信息,以使所述第二通信设备基于所述第二信息进行模型部署。
根据本公开实施例的第四方面,提出了一种第二通信装置,包括:
收发模块,用于向第一通信设备发送第一信息,所述第一信息包括所述第二通信设备所配置的第一模型的模型信息,所述第一模型为用于信道状态信息CSI反馈的模型;
所述收发模块,还用于接收所述第一通信设备发送的第二信息,所述第二信息为所述第一通信设备基于所述第一信息确定的,所述第二信息为待部署在所述第二通信设备中的所述第一模型的模型部署信息;
处理模块,用于基于所述第二信息进行模型部署。
根据本公开实施例的第五方面,提出了一种通信系统,包括:
第一通信设备,被配置为执行前述第一方面的可选实现方式;
第二通信设备,被配置为执行前述第二方面的可选实现方式。
根据本公开实施例的第六方面,提出了一种通信设备,包括:一个或多个处理器;
其中,所述处理器用于调用指令以使得所述通信设备执行前述第一方面和第二方面的可选实现方式。
根据本公开实施例的第七方面,提出了一种存储介质,所述存储介质存储有指令,当所述指令在通信设备上运行时,使得所述通信设备执行前述第一方面和第二方面的可选实现方式。
根据本公开技术方案,基于通信设备之间的信息交互,可以实现通信设备之间的模型初始化,从而可以确保用于CSI反馈的模型的正常工作,保证基于人工智能的CSI反馈方案的精确度。
为了更清楚地说明本公开实施例中的技术方案,以下对实施例描述所需的附图进行介绍,以下附图仅仅是本公开的一些实施例,不对本公开的保护范围造成具体限制。
图1是本公开实施例提供的一种通信系统的架构示意图;
图2a是根据本公开实施例示出的CSI反馈网络的结构示例图;
图2b是根据本公开实施例示出的切换场景的示例图;
图3是根据本公开实施例示出的模型初始化方法的交互示意图;
图4A是根据本公开实施例示出的一种模型初始化方法的流程示意图;
图4B是根据本公开实施例示出的一种模型初始化方法的流程示意图;
图4C是根据本公开实施例示出的模型初始化方法的交互示意图;
图5A是根据本公开实施例示出的一种模型初始化方法的流程示意图;
图5B是根据本公开实施例示出的一种模型初始化方法的流程示意图;
图5C是根据本公开实施例示出的一种模型初始化方法的流程示意图;
图6是根据本公开实施例示出的模型初始化方法的交互示意图;
图7是根据本公开实施例示出的模型初始化方法的示意图;
图8是根据本公开实施例示出的模型初始化方法的示意图;
图9A是本公开实施例提出的第一通信设备的结构示意图;
图9B是本公开实施例提出的第二通信设备的结构示意图;
图10A是本公开实施例提出的通信设备1100的结构示意图;
图10B是本公开实施例提出的芯片1200的结构示意图。
本公开实施例提出了一种模型初始化方法及其装置。
第一方面,本公开实施例提出一种模型初始化方法,包括:
第一通信设备接收第二通信设备发送的第一信息,所述第一信息包括所述第二通信设备所配置的第一模型的模型信息,所述第一模型为用于信道状态信息CSI反馈的模型;
基于所述第一信息,确定第二信息;
向所述第二通信设备发送所述第二信息,以使所述第二通信设备基于所述第二信息进行模型部署。
在上述实施例中,基于通信设备之间的信息交互,可以实现通信设备之间的模型初始化,从而可以有效确保实际系统工作过程中,通信设备间的CSI反馈网络能够相互匹配,从而有效保障CSI反馈网络的性能稳定性,提高人工智能方案的实用性。
结合第一方面的一些实施例,在一些实施例中,所述基于所述第一信息,确定第二信息,包括:
确定CSI实时数据;
基于所述CSI实时数据和所述第一信息,确定所述第二信息。
在上述实施例中,能够有效确保实际系统工作过程中,通信设备之间的CSI反馈网络模型能够相互匹配且适用于当前的信道环境,从而可以进一步有效保障CSI反馈网络模型的性能稳定性,进一步提高人工智能方案的实用性。
结合第一方面的一些实施例,在一些实施例中,所述第一通信设备为网络设备,所述第二通信设备为终端,所述第一模型为CSI反馈网络中的编码器模型。
在上述实施例中,CSI反馈网络中的编码器模型部署在终端侧,便于终端基于编码器模型完成对原始CSI数据的特征提取与维度压缩,将原始CSI数据压缩为数据量更小的码字,再通过上行反馈链路将码字信息传输至网络设备,以充分降低反馈开销。
结合第一方面的一些实施例,在一些实施例中,所述确定CSI实时数据,包括:
向所述终端发送导频信号和数据请求指示信息,所述导频信号用于所述终端在当前时刻进行信道估计以得到CSI实时数据,所述数据请求指示信息用于指示所述网络设备向所述终端请求所述CSI实时数据;
接收所述终端发送的经量化处理的CSI实时数据;
将接收到的所述经量化处理的CSI实时数据进行去量化处理,得到所述CSI实时数据。
在上述实施例中,通过网络设备向终端发送导频信号和数据请求指示信息,可以获得终端经过信道估计得到的CSI实时数据,从而可以得到能够准确反映出当前信道环境的CSI实时数据,便于网络设备基于CSI实时数据进行模型选择,从而可以有效确保网络设备与终端间的CSI反馈网络模型相互匹配且适用于当前的信道环境。
结合第一方面的一些实施例,在一些实施例中,所述第一通信设备为终端,所述第二通信设备为网络设备,所述第一模型为CSI反馈网络中的译码器模型。
结合第一方面的一些实施例,在一些实施例中,所述确定CSI实时数据,包括:
接收所述网络设备发送的导频信号;
在当前时刻基于所述导频信号进行信道估计,得到所述CSI实时数据。
在上述实施例中,终端可以基于网络设备发送的导频信号进行信道估计,得到CSI实时数据,从而可以得到能够准确反映出当前信道环境的CSI实时数据,便于终端基于CSI实时数据进行模型选择,从而可以有效确保网络设备与终端间的CSI反馈网络模型相互匹配且适用于当前的信道环境。
结合第一方面的一些实施例,在一些实施例中,所述基于所述CSI实时数据和所述第一信息,确定所述第二信息,包括:
将所述CSI实时数据输入所述第一通信设备所部署的CSI反馈网络,其中,所述CSI反馈网络包括所述第一模型和第二模型,所述CSI反馈网络通过所述第一模型和所述第二模型对所述CSI实时数据进行压缩和重建处理;
获取所述CSI反馈网络输出的CSI输出值;
根据所述CSI输出值与所述CSI实时数据,确定所述CSI反馈网络的反馈精确度;
根据所述反馈精确度,从所述第一通信设备所部署的CSI反馈网络中选择第一CSI反馈网络;
基于所述第一CSI反馈网络中第一模型的模型信息和所述第二通信设备配置的第一模型的模型信息,确定所述第二信息。
在上述实施例中,基于CSI实时数据和上述第一信息,可以从第一通信设备所部署的CSI反馈网络选择出反馈精确度满足一定条件的CSI反馈网络,基于所选择的CSI反馈网络中的第一模型的模型参数和第二通信设备配置的第一模型的模型信息,完成第二通信设备中第一模型的部署,可以保证所选择的CSI反馈网络中的第一模型更加适用于当前的信道环境。
结合第一方面的一些实施例,在一些实施例中,所述第一通信设备所部署的CSI反馈网络为多个;所述根据所述反馈精确度,从所述第一通信设备所部署的CSI反馈网络中选择第一CSI反馈网络,包括:
从多个所述CSI反馈网络中,选择所述反馈精确度最高的CSI反馈网络确定为所述第一CSI反馈网络。
在上述实施例中,可以进一步有效保障反馈网络中的第一模型的性能稳定性。
结合第一方面的一些实施例,在一些实施例中,所述基于所述第一CSI反馈网络中第一模型的模型信息和所述第二通信设备配置的第一模型的模型信息,确定所述第二信息,包括以下任一项:
将所述第一CSI反馈网络中第一模型的模型信息确定为所述第二信息,其中,所述第二通信设备配置的第一模型的模型信息中包含所述第一CSI反馈网络中第一模型的模型信息;
将所述第一CSI反馈网络中第一模型的模型参数进行量化处理,并将经量化处理的所述模型参数确定为所述第二信息,其中,所述第二通信设备配置的第一模型的模型信息中不包含所述第一CSI反馈网络中第一模型的模型信息。
在上述实施例中,如果第二通信设备配置的模型与第一通信设备所选择的模型相互匹配,则将该模型的模型参数确定为待部署在第二通信设备中第一模型的信息,从而完成模型初始化;如果第二通信设备配置的模型不包含第一通信设备所选择的模型,则将第一通信设备所选择的模型的模型参数提供给第二通信设备进行模型部署,从而完成模型初始化,这样,可以确保用于CSI反馈的模型的正常工作,保证基于人工智能的CSI反馈方案的精确度。
结合第一方面的一些实施例,在一些实施例中,所述将经量化处理的所述模型参数确定为所述第二信息,包括:
将所述第一CSI反馈网络中第一模型的模型信息和经量化处理的所述模型参数,确定为所述第二信息。
结合第一方面的一些实施例,在一些实施例中,所述方法还包括:接收所述第二通信设备发送的第一指示信息,所述第一指示信息用于指示所述第二通信设备配置了所述第一模型,且具备使用所述第一模型完成CSI反馈工作的能力。
结合第一方面的一些实施例,在一些实施例中,所述方法还包括:向所述第二通信设备发送第二指示信息,所述第二指示信息用于指示所述第一通信设备通知所述第二通信设备进入模型初始化流程;接收所述第二通信设备发送的第三指示信息,所述第三指示信息用于指示所述第二通信设备接受了模型初始化指令,进入模型初始化流程。
结合第一方面的一些实施例,在一些实施例中,所述方法还包括:接收所述第二通信设备发送的确认信息,所述确认信息包括用于指示所述第二通信设备完成模型初始化的确认指示信息。
结合第一方面的一些实施例,在一些实施例中,所述模型信息包括模型编号和/或模型结构。
第二方面,本公开实施例提出一种模型初始化方法,包括:
第二通信设备向第一通信设备发送第一信息,所述第一信息包括由所述第二通信设备配置的第一模型的模型信息,所述第一模型为用于信道状态信息CSI反馈的模型;
接收所述第一通信设备发送的第二信息,所述第二信息为所述第一通信设备基于所述第一信息确定的,所述第二信息为待部署在所述第二通信设备中的所述第一模型的模型部署信息;
基于所述第二信息进行模型部署。
结合第二方面的一些实施例,在一些实施例中,所述第一通信设备为网络设备,所述第二通信设备为终端,所述第一模型为CSI反馈网络中的编码器模型。
结合第二方面的一些实施例,在一些实施例中,所述方法还包括:
接收所述网络设备发送的导频信号和数据请求指示信息,所述导频信号用于所述终端在当前时刻进行信道估计以得到CSI实时数据,所述数据请求指示信息用于指示所述网络设备向所述终端请求所述CSI实时数据;
基于所述导频信号进行信道估计,得到所述CSI实时数据;
将所述CSI实时数据进行量化处理;
将经量化处理的所述CSI实时数据发送给所述网络设备,其中,所述CSI实时数据用于所述网络设备结合所述第一信息确定第二信息。
结合第二方面的一些实施例,在一些实施例中,所述第一通信设备为终端,所述第二通信设备为网络设备,所述第一模型为CSI反馈网络中的译码器模型。
结合第二方面的一些实施例,在一些实施例中,所述方法还包括:向所述终端发送导频信号,其中,所述导频信号用于所述终端在当前时刻进行信道估计以得到CSI实时数据,所述CSI实时数据用于所述终端结合所述第一信息确定第二信息。
结合第二方面的一些实施例,在一些实施例中,所述基于所述第二信息进行模型部署,包括以下任一项:
确定所述第二信息为第一CSI反馈网络中第一模型的模型信息,在所述第二通信设备上激活所述模型信息所对应的第一模型,其中,所述第一CSI反馈网络是由所述第一通信设备基于CSI实时数据从所述第一通信设备所部署的CSI反馈网络中选择的;
确定所述第二信息包括经量化处理的所述模型参数,将所述模型参数进行去量化处理,并将去量化处理的所述模型参数部署在所述第二通信设备上,激活所述部署的模型。
结合第二方面的一些实施例,在一些实施例中,所述方法还包括:向所述第一通信设备发送第一指示信息,所述第一指示信息用于指示所述第二通信设备配置了所述第一模型,且具备使用所述第一模型完成CSI反馈工作的能力。
结合第二方面的一些实施例,在一些实施例中,所述方法还包括:接收所述第一通信设备发送的第二指示信息,所述第二指示信息用于指示所述第一通信设备通知所述第二通信设备进入模型初始化流程;向所述第一通信设备发送第三指示信息,所述第三指示信息用于指示所述第二通信设备接受了模型初始化指令,进入模型初始化流程。
结合第二方面的一些实施例,在一些实施例中,所述方法还包括:向所述第一通信设备发送确认信息,所述确认信息包括用于指示所述第二通信设备完成模型初始化的确认指示信息。
结合第二方面的一些实施例,在一些实施例中,所述模型信息包括模型编号和/或模型结构。
第三方面,本公开实施例提出一种第一通信装置,包括收发模块、处理模块中的至少一者;其中,上述第一通信装置用于执行第一方面的可选实现方式。
第四方面,本公开实施例提出一种第二通信装置,包括收发模块、处理模块中的至少一者;其中,上述第二通信装置用于执行第二方面的可选实现方式。
第五方面,本公开实施例提出一种通信系统,包括:
第一通信装置,被配置为前述第一方面的可选实现方式;
第二通信装置,被配置为执行前述第二方面的可选实现方式。
第六方面,本公开实施例提出一种通信设备,包括:一个或多个处理器;其中,所述处理器用于调用指令以使得所述通信设备执行前述第一方面的可选实现方式。
第七方面,本公开实施例提出一种通信设备,包括:一个或多个处理器;其中,所述处理器用于调用指令以使得所述通信设备执行前述第二方面的可选实现方式。
第八方面,本公开实施例提出一种存储介质,所述存储介质存储有指令,当所述指令在通信设备上运行时,使得所述通信设备执行前述第一方面和第二方面的可选实现方式。
第九方面,本公开实施例提出程序产品,上述程序产品被通信设备执行时,使得上述通信设备执行如第一方面和第二面的可选实现方式所描述的方法。
第十方面,本公开实施例提出了计算机程序,当其在计算机上运行时,使得计算机执行如第一方面和第二方面的可选实现方式所描述的方法。
第十一方面,本公开实施例提供了一种芯片或芯片系统。该芯片或芯片系统包括处理电路,被配置为执行根据上述第一方面和第二方面的可选实现方式所描述的方法。
可以理解地,上述第一通信装置、第二通信装置、通信系统、存储介质、程序产品、计算机程序、芯片或芯片系统均用于执行本公开实施例所提出的方法。因此,其所能达到的有益效果可以参考对应方法中的有益效果,此处不再赘述。
本公开实施例提出了模型初始化方法及其装置。在一些实施例中,模型初始化方法、信息处理方法、通信方法等术语可以相互替换,模型初始化的装置、信息处理的装置、通信装置等术语可以相互替换,模型初始化系统、信息处理系统、通信系统等术语可以相互替换。
本公开实施例并非穷举,仅为部分实施例的示意,不作为对本公开保护范围的具体限制。在不矛盾的情况下,某一实施例中的每个步骤均可以作为独立实施例来实施,且各步骤之间可以任意组合,例如,在某一实施例中去除部分步骤后的方案也可以作为独立实施例来实施,且在某一实施例中各步骤的顺序可以任意交换,另外,某一实施例中的可选实现方式可以任意组合;此外,各实施例之间可以任意组合,例如,不同实施例的部分或全部步骤可以任意组合,某一实施例可以与其他实施例的可选实现方式任意组合。
在各本公开实施例中,如果没有特殊说明以及逻辑冲突,各实施例之间的术语和/或描述具有一致性,且可以互相引用,不同实施例中的技术特征根据其内在的逻辑关系可以组合形成新的实施例。
本公开实施例中所使用的术语只是为了描述特定实施例的目的,而并非作为对本公开的限制。
在本公开实施例中,除非另有说明,以单数形式表示的元素,如“一个”、“一种”、“该”、“上述”、“所述”、“前述”、“这一”等,可以表示“一个且只有一个”,也可以表示“一个或多个”、“至少一个”等。例如,在翻译中使用如英语中的“a”、“an”、“the”等冠词(article)的情况下,冠词之后的名词可以理解为单数表达形式,也可以理解为复数表达形式。
在本公开实施例中,“多个”是指两个或两个以上。
在一些实施例中,“至少一者(至少一项、至少一个)(at least one of)”、“一个或多个(one or more)”、“多个(a plurality of)”、“多个(multiple)等术语可以相互替换。
在一些实施例中,“A、B中的至少一者”、“A和/或B”、“在一情况下A,在另一情况下B”、“响应于一情况A,响应于另一情况B”等记载方式,根据情况可以包括以下技术方案:在一些实施例中A(与B无关地执行A);在一些实施例中B(与A无关地执行B);在一些实施例中从A和B中选择执行(A和B被选择性执行);在一些实施例中A和B(A和B都被执行)。当有A、B、C等更多分支时也类似上述。
在一些实施例中,“A或B”等记载方式,根据情况可以包括以下技术方案:在一些实施例中A(与B无关地执行A);在一些实施例中B(与A无关地执行B);在一些实施例中从A和B中选择执行(A和B被选择性执行)。当有A、B、C等更多分支时也类似上述。
本公开实施例中的“第一”、“第二”等前缀词,仅仅为了区分不同的描述对象,不对描述对象的位置、顺序、优先级、数量或内容等构成限制,对描述对象的陈述参见权利要求或实施例中上下文的描述,不应因为使用前缀词而构成多余的限制。例如,描述对象为“字段”,则“第一字段”和“第二字段”中“字段”之前的序数词并不限制“字段”之间的位置或顺序,“第一”和“第二”并不限制其修饰的“字段”是否在同一个消息中,也不限制“第一字段”和“第二字段”的先后顺序。再如,描述对象为“等级”,则“第一等级”和“第二等级”中“等级”之前的序数词并不限制“等级”之间的优先级。再如,描述对象的数量并不受序数词的限制,可以是一个或者多个,以“第一装置”为例,其中“装置”的数量可以是一个或者多个。此外,不同前缀词修饰的对象可以相同或不同,例如,描述对象为“装置”,则“第一装置”和“第二装置”可以是相同的装置或者不同的装置,其类型可以相同或不同;再如,描述对象为“信息”,则“第一信息”和“第二信息”可以是相同的信息或者不同的信息,其内容可以相同或不同。
在一些实施例中,“包括A”、“包含A”、“用于指示A”、“携带A”,可以解释为直接携带A,也可以解释为间接指示A。
在一些实施例中,“响应于……”、“响应于确定……”、“在……的情况下”、“在……时”、“当……时”、“若……”、“如果……”等术语可以相互替换。
在一些实施例中,“大于”、“大于或等于”、“不小于”、“多于”、“多于或等于”、“不少于”、“高于”、“高于或等于”、“不低于”、“以上”等术语可以相互替换,“小于”、“小于或等于”、“不大于”、“少于”、“少于或等于”、“不多于”、“低于”、“低于或等于”、“不高于”、“以下”等术语可以相互替换。
在一些实施例中,装置和设备可以解释为实体的、也可以解释为虚拟的,其名称不限定于实施例中所记载的名称,在一些情况下也可以被理解为“设备(equipment)”、“设备(device)”、“电路”、“网元”、“节点”、“功能”、“单元”、“部件(section)”、“系统”、“网络”、“芯片”、“芯片系统”、“实体”、“主体”等。
在一些实施例中,“网络”可以解释为网络中包含的装置,例如,接入网设备、核心网设备等。
在一些实施例中,“接入网设备(access network device,AN device)”也可以被称为“无线接入网设备(radio access network device,RAN device)”、“基站(base station,BS)”、“无线基站(radio base station)”、“固定台(fixed station)”,在一些实施例中也可以被理解为“节点(node)”、“接入点(access point)”、“发送点(transmission point,TP)”、“接收点(reception point,RP)”、“发送和/或接收点(transmission/reception point,TRP)”、“面板(panel)”、“天线面板(antenna panel)”、“天线阵列(antenna array)”、“小区(cell)”、“宏小区(macro cell)”、“小型小区(small cell)”、“毫微微小区(femto cell)”、“微微小区(pico cell)”、“扇区(sector)”、“小区组(cell group)”、“服务小区”、“载波(carrier)”、“分量载波(component carrier)”、“带宽部分(bandwidth part,BWP)”等。
在一些实施例中,“终端(terminal)”或“终端设备(terminal device)”可以被称为“用户设备(user equipment,UE)”、“用户终端(user terminal)”、“移动台(mobile station,MS)”、“移动终端(mobile terminal,MT)”、订户站(subscriber station)、移动单元(mobile unit)、订户单元(subscriber unit)、无线单元(wireless unit)、远程单元(remote unit)、移动设备(mobile device)、无线设备(wireless device)、无线通信设备(wireless communication device)、 远程设备(remote device)、移动订户站(mobile subscriber station)、接入终端(access terminal)、移动终端(mobile terminal)、无线终端(wireless terminal)、远程终端(remote terminal)、手持设备(handset)、用户代理(user agent)、移动客户端(mobile client)、客户端(client)等。
在一些实施例中,获取数据、信息等可以遵照所在地国家的法律法规。
在一些实施例中,可以在得到用户同意后获取数据、信息等。
图1是根据本公开实施例示出的通信系统的架构示意图。该通信系统可包括但不限于一个第一通信设备和一个第二通信设备,图1所示的设备数量和形态仅用于举例并不构成对本公开实施例的限定,实际应用中可以包括两个或两个以上的第一通信设备,两个或两个以上的第二通信设备。图1所示的通信系统100以包括一个第一通信设备101和一个第二通信设备102为例。
在一些实施例中,第一通信设备101可以为网络设备,第二通信设备102可以为终端。其中,在一些实施例中,第一通信设备101上配置有CSI反馈网络中的第二模型,第二通信设备102上配置有CSI反馈网络中的第一模型。当第一通信设备101可以为网络设备,第二通信设备102可以为终端,则第二模型为CSI反馈网络中的译码器模型,第一模型为CSI反馈网络中的编码器模型。
在一些实施例中,第一通信设备101可以为终端,第二通信设备102可以为网络设备。其中,在一些实施例中,第一通信设备101上配置有CSI反馈网络中的第二模型,第二通信设备102上配置有CSI反馈网络中的第一模型。当第一通信设备101可以为终端,第二通信设备102可以为网络设备,则第二模型为CSI反馈网络中的编码器模型,第一模型为CSI反馈网络中的译码器模型。
在一些实施例中,本文中的终端可以是用户侧的一种用于接收或发射信号的实体,如手机。也可以称为终端(terminal)、用户设备(user equipment,UE)、移动台(mobile station,MS)、移动终端(mobile terminal,MT)等。终端可以是具备通信功能的汽车、智能汽车、手机(mobile phone)、穿戴式设备、平板电脑(Pad)、带无线收发功能的电脑、虚拟现实(virtual reality,VR)终端、增强现实(augmented reality,AR)终端、工业控制(industrial control)中的无线终端、无人驾驶(self-driving)中的无线终端、远程手术(remote medical surgery)中的无线终端、智能电网(smart grid)中的无线终端、运输安全(transportation safety)中的无线终端、智慧城市(smart city)中的无线终端、智慧家庭(smart home)中的无线终端等等中的至少一者。本公开的实施例对终端所采用的具体技术和具体设备形态不做限定。
在一些实施例中,本文中的网络设备可以是接入网设备。在一些实施例中,接入网设备例如是将终端设备接入到无线网络的节点或设备,接入网设备可以包括5G通信系统中的演进节点B(evolved NodeB,eNB)、下一代演进节点B(next generation eNB,ng-eNB)、下一代节点B(next generation NodeB,gNB)、节点B(node B,NB)、家庭节点B(home node B,HNB)、家庭演进节点B(home evolved nodeB,HeNB)、无线回传设备、无线网络控制器(radio network controller,RNC)、基站控制器(base station controller,BSC)、基站收发台(base transceiver station,BTS)、基带单元(base band unit,BBU)、移动交换中心、6G通信系统中的基站、开放型基站(Open RAN)、云基站(Cloud RAN)、其他通信系统中的基站、Wi-Fi系统中的接入节点中的至少一者,但不限于此。
在一些实施例中,本公开的技术方案可适用于Open RAN架构,此时,本公开实施例所涉及的接入网设备间或者接入网设备内的接口可变为Open RAN的内部接口,这些内部接口之间的流程和信息交互可以通过软件或者程序实现。
在一些实施例中,接入网设备可以由集中单元(central unit,CU)与分布式单元(distributed unit,DU)组成的,其中,CU也可以称为控制单元(control unit),采用CU-DU的结构可以将接入网设备的协议层拆分开,部分协议层的功能放在CU集中控制,剩下部分或全部协议层的功能分布在DU中,由CU集中控制DU,但不限于此。
可以理解的是,本公开实施例描述的通信系统是为了更加清楚的说明本公开实施例的技术方案,并不构成对于本公开实施例提出的技术方案的限定,本领域普通技术人员可知,随着系统架构的演变和新业务场景的出现,本公开实施例提出的技术方案对于类似的技术问题同样适用。
下述本公开实施例可以应用于图1所示的通信系统100、或部分主体,但不限于此。图1所示的各主体是例示,通信系统可以包括图1中的全部或部分主体,也可以包括图1以外的其他主体,各主体数量和形态为任意,各主体可以是实体的也可以是虚拟的,各主体之间的连接关系是例示,各主体之间可以不连接也可以连接,其连接可以是任意方式,可以是直接连接也可以是间接连接,可以是有线连接也可以是无线连接。
本公开各实施例可以应用于长期演进(Long Term Evolution,LTE)、LTE-Advanced(LTE-A)、LTE-Beyond(LTE-B)、SUPER 3G、IMT-Advanced、第四代移动通信系统(4th generation mobile communication system,4G)、)、第五代移动通信系统(5th generation mobile communication system,5G)、5G新空口(new radio,NR)、未来无线接入(Future Radio Access,FRA)、新无线接入技术(New-Radio Access Technology,RAT)、新无线(New Radio,NR)、新无线接入(New radio access,NX)、未来一代无线接入(Future generation radio access,FX)、Global System for Mobile communications(GSM(注册商标))、CDMA2000、超移动宽带(Ultra Mobile Broadband,UMB)、IEEE 802.11(Wi-Fi(注册商标))、IEEE 802.16(WiMAX(注册商标))、IEEE 802.20、超宽带(Ultra-WideBand,UWB)、蓝牙(Bluetooth(注册商标))、陆上公用移动通信网(Public Land Mobile Network,PLMN)网络、设备到设备(Device-to-Device,D2D)系统、机器到机器(Machine to Machine,M2M)系统、物联网(Internet of Things,IoT)系统、车联网(Vehicle-to-Everything,V2X)、利用其他通信方法的系统、基于它们而扩展的下一代系统等。此外,也可以将多个系统组合(例如,LTE或者LTE-A与5G的组合等)应用。
大规模多输入多输出(Multiple-Input Multiple-Output,MIMO)是第五代移动通信(The 5th Generation,5G)系统的关键使能技术,有助于实现超大带宽、超高速率、低时延通信,并将在未来无线通信系统发展过程中发挥关键作用,为各种未来无线通信场景提供技术支持。
为实现大规模MIMO技术在系统容量、功率效率、频谱效率等方面的性能增益,网络设备(如基站,Base Station,也叫BS)需要获取精确的下行信道状态信息(Channel State Information,CSI)。时分双工(Time Division Duplex,TDD)模式下,网络设备可以首先获取上行CSI,再利用信道互易性获取下行CSI。频分双工(Frequency Division Duplex,FDD)模式下,由于上行与下行信道工作在不同频段,信道互易性较弱,因此通常采用CSI反馈的方式,即用户设备(User Equipment,UE)首先通过信道估计技术获取下行CSI,再将下行CSI数据通过上行链路反馈至网络设备。然而,CSI的数据维度与接收、发送天线数量成正比,而大规模MIMO系统中,收发天线数量较多,导致CSI数据量显著增加。若终端直接将完整的CSI数据反馈传输至网络设备,需要极大的上行链路通信开销,对通信系统的高效工作造成影响。因此,实现FDD模式下高精度、低开销的下行CSI反馈是大规模MIMO技术领域中的重要研究方向。
基于深度学习(Deep Learning)技术实现FDD模式下的下行CSI反馈能够以有效降低反馈开销和计算复杂度,同时也能够显著提升CSI反馈精确度,已经成为大规模MIMO系统中解决CSI反馈问题的重要方法。在一些实施例中,基于自编码器(Autoencoder)网络结构,提出了多种CSI反馈网络模型,致力于充分利用CSI数据特征信息完成信道状态信息在终端的压缩和在网络设备的重建。
在一些实施例中,如图2a所示,为CSI反馈网络的结构示例图。其中,终端侧的编码器模型完成对原始CSI数据的特征提取与维度压缩,将原始CSI数据压缩为数据量更小的码字,再通过上行反馈链路将码字信息传输至网络设备以充分降低反馈开销。网络设备接收到的码字经过译码器模型完成重建以输出原始维度的CSI数据。
在一些实施例中,CSI反馈网络模型训练过程中,使用一定规模的CSI样本数据集训练编码器(encoder)和译码器(decoder)组成的反馈网络模型,最小化译码器输出的重建CSI数据与编码器输入的原始CSI数据之间的差异,以使反馈网络模型充分学习当前训练集的CSI数据特征分布,进而在面对服从相同或相近特征分布的CSI数据时,能够以较低的误差实现CSI的压缩、反馈和重建工作。基于深度学习的CSI反馈方案有效降低了反馈开销,同时在提升反馈精度和降低计算成本等方面具有较大优势。
在一些实施例中,基于深度学习的CSI反馈技术方案通常使用离线模型训练方法(Offline Training),即在神经网络模型部署在实际通信系统前,使用较大规模提前收集的训练数据集,经过较多的训练轮次以获取反馈网络模型参数。离线训练所获取的神经网络模型被部署于实际系统中后,能够较好地处理与训练数据服从相同或相近分布的实时数据。
在一些实施例中,训练好的CSI反馈网络中编码器模型和译码器模型将被分别被部署于终端与网络设备,二者需要协同工作以实现CSI数据的压缩、与重建。
由于实际通信系统中通常存在多个不同的信道场景,如室内与室外场景,不同信道场景下的CSI数据特征通常具有较大差异,若使用一个神经网络模型完成不同信道场景下的CSI压缩和重建工作,难以达到CSI反馈精确度的要求。因此,在一些实施例中,使用不同信道场景下的样本数据分别训练多个模型来完成相应信道场景下的CSI反馈,因此终端与网络设备也需要同时部署多个信道场景下的反馈网络模型。
在一些实施例中,基于深度学习的CSI反馈网络技术方案较多关注模型训练与模型推理两个阶段。在模型训练阶段,使用较大规模的训练数据集,经过较多的训练轮次以获取反馈网络模型参数,由于不同信道场景下的CSI数据特征分布差异较大,通常情况下需要为不同的信道场景训练不同的网络模型。在模型推理阶段,终端与网络设备各自使用训练好的编码器模型和译码器模型完成实时CSI数据的压缩和重建,从而完成大规模MIMO系统中的CSI反馈工作。
在一些实施例中,终端与网络设备的协同工作是应用神经网络模型完成下行CSI反馈的重要保证。CSI反馈网络模型训练完成后,训练获取的网络模型参数将会被部署至终端与网络设备。在模型应用即模型推理前,终端与网络设备需要首先完成模型初始化,即针对当前是否应用CSI反馈网络模型、需要使用的具体模型等信息进行交互,才能保证终端与网络设备使用互相匹配的并且适合当前信道环境的编码器模型、译码器模型完成CSI压缩与重建。此外,由于终端具有较强的移动性,如图2b所示,当终端从一个网络设备的覆盖范围进入另一个网络设备的覆盖范围时,终端需要与其协同工作的网络设备重新确认所应用的CSI反馈网络模型的相关参数信息。同时,终端在进入一个新的信道场景时,可能并未部署相应信道场景下的编码器模型,需要与网络设备进行确认并完成模型参数的部署后,再使用CSI反馈网络模型完成CSI反馈任务。
在一些实施例中,CSI反馈网络模型可以为基于AI(Artificial Intelligence,人工智能)/ML(Machine Learning,机器学习)模型。
在一些实施例中,大规模MIMO系统中应用深度神经网络模型完成下行CSI反馈,需要终端与网络设备的协同工作。原始CSI样本在终端侧完成数据压缩,在网络设备侧完成数据重建,在终端与网络设备的模型匹配的情况下才能以较高的精确度有效地完成下行CSI反馈。由于终端较强的移动性,终端可能在不同的时间、不同的信道场景下与不同的网络设备协作完成CSI压缩、反馈、重建的任务,在进行模型推理前,需要完成CSI反馈网络模型初始化,即终端与网络设备首先确定当前需要应用AI/ML模型完成CSI反馈以及需要使用的具体模型,其次使用相应的模型完成CSI的压缩与重建工作。只有确定一种终端与网络设备之间的模型初始化方案,才能确保AI/ML模型的正常工作,保证基于人工智能的CSI反馈方案的精确度。
本公开实施例提供的模型初始化方法及其装置,基于通信设备之间的信息交互,可以实现通信设备之间的模型初始化,从而可以确保用于CSI反馈的模型的正常工作,保证基于人工智能的CSI反馈方案的精确度。
图3是根据本公开实施例示出的模型初始化方法的交互示意图。如图3所示,本公开实施例涉及模型初始化方法可应用于通信系统100,上述方法包括但不限于如下步骤。
步骤S3101,第二通信设备102发送第一指示信息。
在一些实施例中,第二通信设备102向第一通信设备101发送上述第一指示信息。在一些实施例中,第一通信设备101接收上述第一指示信息,可选的,第一通信设备101接收第二通信设备102发送的上述第一指示信息。
在一些实施例中,上述第一指示信息用于指示第二通信设备102配置了第一模型,且具备使用第一模型完成CSI反馈工作的能力。在一些实施例中,上述第一模型为用于CSI反馈的模型,作为一种示例,第一模型可以是CSI反馈网络中的编码器模型,或者也可以是CSI反馈网络中的译码器模型。
在一些实施例中,第一通信设备101为网络设备,第二通信设备102为终端,第一模型为CSI反馈网络中的编码器模型。可选的,第一通信设备101上配置有CSI反馈网络中的译码器模型,第二通信设备102上配置有CSI反馈网络中的编码器模型。
在一些实施例中,第一通信设备101可以为终端,第二通信设备102可以为网络设备。可选的,第一通信设备101上配置有CSI反馈网络中的编码器模型,第二通信设备102上配置有CSI反馈网络中的译码器模型。
在一些实施例中,“第一指示信息”、“模型应用请求指示信息”、“模型应用请求指示”、“应用请求指示信息”、“应用请求指示”等术语可以相互替换。作为一种示例,上述第一指示信息可以为AI/ML模型应用请求指示,可选的,该AI/ML模型应用请求指示可以用Indicatorinit_request表示,该AI/ML模型应用请求指示是指终端配置了CSI反馈网络中的第一模型,即配置了AI/ML模型,且当前具备使用AI/ML模型完成CSI反馈工作的能力。
在一些实施例中,“CSI反馈网络”、“CSI反馈网络模型”、“AI/ML模型”等术语可以相互替换。
步骤S3102,第二通信设备102发送第一信息。
在一些实施例中,第二通信设备102向第一通信设备101发送上述第一信息。在一些实施例中,第一通信设备101接收上述第一信息。可选的,第一通信设备101接收第二通信设备102发送的上述第一信息。
在一些实施例中,上述第一信息包括由第二通信设备102配置的第一模型的模型信息。其中,针对第一模型的描述可参见上述步骤S3101的相关描述,在此不再赘述。
在一些实施例中,上述模型信息可以包括但不限于模型编号和/或模型结构。其中,“模型结构”也叫“网络结构”,指示构建模型所使用的结构。
在一些实施例中,上述模型信息可以包括模型编号。示例性的,第一通信设备101与第二通信设备102上同时配置了多个信道场景下的CSI反馈网络模型,该多个信道场景下的CSI反馈网络模型的模型编号具有唯一性,也就是说可以利用模型编号即可区分不同信道场景下的CSI反馈网络模型。在这种场景下,第二通信设备102发送的第一信息中包括由第二通信设备102配置的第一模型的模型编号。
在一些实施例中,上述模型信息可以包括模型编号和模型结构。示例性的,第一通信设备101与第二通信设备102上同时配置了多个信道场景下的CSI反馈网络模型,当多个信道场景下的CSI反馈网络模型对应多个模型结构,不同模型结构下可能采用相同的模型编号区分模型,因此,基于模型编号和模型结构可以唯一标识不同信道场景下的CSI反馈网络模型。在这种场景下,第二通信设备102发送的第一信息中包括由第二通信设备102配置的第一模型的模型编号和模型结构。
在一些实施例中,上述第一指示信息和上述第一信息可以由第二通信设备102同时发送,例如,第二通信设备102发送一信息,该信息中包括上述第一指示信息和上述第一信息。可选的,在一些实施例中,上述第一指示信息和上述第一信息可以由第二通信设备102分开发送。例如,第二通信设备102先发送上述第一指示信息,再发送上述第一信息,或者,第二通信设备102先发送上述第一信息再发送上述第一指示信息。在此本公开并不对此作出限定。
在一些实施例中,上述第一指示信息和/或上述第一信息可以包含于RRC(Radio Resource Control,无线资源控制)信令,例如专用RRC信令,或者,也可以包含于DCI(Downlink Control Information,下行控制信息),或者也可以包含于其他信令中,在此本公开并不此做出限定,也不再赘述。
步骤S3103,第一通信设备101发送第二指示信息。
在一些实施例中,第一通信设备101向第二通信设备102发送第二指示信息。在一些实施例中,第二通信设备102接收上述第二指示信息,可选的,第二通信设备102接收第一通信设备101发送的上述第二指示信息。
在一些实施例中,上述第二指示信息用于指示第一通信设备101通知第二通信设备102进入模型初始化流程。
在一些实施例中,上述第二指示信息可以是模型初始化指示,例如,可以用Indicatorinitialize表示,用于指示第一通信设备101通知第二通信设备102进入模型初始化流程。示例性的,当第一通信设备101为网络设备,第二通信设备102为终端,则上述第二指示信息可以指示网络设备通知终端进入模型初始化流程。示例性的,当第一通信设备101为终端,第二通信设备102为网络设备,则上述第二指示信息可以指示终端通知网络设备进入模型初始化流程。可选的,在一些实施例中,“第二指示信息”、“模型初始化指示”、“模型初始化指示信息”、“AI/ML模型初始化指示”、“AI/ML模型初始化指示信息”等术语可以相互替换。
在一些实施例中,上述第二指示信息可以包含于RRC信令,例如专用RRC信令,或者,也可以包含于DCI,或者也可以包含于其他信令中,在此本公开并不此做出限定,也不再赘述。
步骤S3104,第二通信设备102发送第三指示信息。
在一些实施例中,第二通信设备102向第一通信设备101发送上述第三指示信息。在一些实施例中,第一通信设备101接收上述第三指示信息,可选的,第一通信设备101接收第二通信设备102发送的上述第三指示信息。
在一些实施例中,上述第三指示信息用于指示第二通信设备102接受了模型初始化指令,进入模型初始化流程。可选的,第一通信设备101接收到第二通信设备102发送的上述第三指示信息,则可以说明第二通信设备102收到了第一通信设备101发送的上述第二指示信息,且接受了对第二通信设备上CSI反馈网络模型的初始化指令,进入模型初始化流程。
在一些实施例中,上述第三指示信息可以包含于RRC信令,例如专用RRC信令,或者,也可以包含于DCI,或者也可以包含于其他信令中,在此本公开并不此做出限定,也不再赘述。
步骤S3105,第一通信设备101确定CSI实时数据。
在一些实施例中,上述CSI实时数据是指终端在当前时刻通过信道估计所获取的下行CSI,能够在一定程度上代表当前信道环境下的CSI数据特征。
需要说明的是,由于上述CSI实时数据是由终端经过信道估计获得的下行CSI,而模型初始化发起端既可以是终端,也可以是网络设备,不同的模型初始化发起端,则确定CSI实时数据的实现方式会所有不同。
在一些实施例中,当第一通信设备101为网络设备,上述第二通信设备102为终端,上述第一模型为CSI反馈网络中的编码器模型时,第一通信设备101确定CSI实时数据的一种可能实现方式如下:网络设备发送导频信号和数据请求指示信息,上述导频信号用于终端在当前时刻进行信道估计以得到CSI实时数据,上述数据请求指示信息用于指示网络设备向终端请求CSI实时数据。网络设备接收经量化处理的CSI实时数据,并将接收到的经量化处理的CSI实时数据进行去量化处理,得到CSI实时数据。
可选的,在一些实施例中,网络设备向终端发送上述导频信号和上述数据请求指示信息。终端接收上述导频信号和上述数据请求指示信息,例如,终端收到网络设备发送的上述导频信号和上述数据请求指示信息,基于上述导频信号在当前时刻进行信道估计,以得到上述CSI实时数据。终端发送上述CSI实时数据,如终端向网络设备发送上述CSI实时数据。网络设备接收终端发送的上述CSI实时数据,从而得到上述CSI实时数据,便于网络设备基于上述CSI实时数据进行下一步的模型选择。
可选的,终端与网络设备之间的数据交互均需要经过量化处理。也就是说,终端发送的CSI实时数据是经过量化处理后的数据。网络设备收到经量化处理的CSI实时数据时,需要将收到的经量化处理的CSI实时数据进行去量化处理,从而可以得到终端对当前信道环境进行信道估计时得到的CSI实时数据。
在一些实施例中,上述量化是指:将幅度连续取值的信号、序列或权重值变换为幅度离散取值的过程,通常可分为均匀量化和非均匀量化。实际通信系统中,终端或网络设备进行数据传输之前,需要将原始数据转换为比特流形式。无论是原始CSI数据、CSI数据经过终端的编码器压缩之后输出的码字数据,还是CSI反馈网络模型权重值序列,其数据形式通常为较高精度的浮点数据,因此需要通过数据量化将其转换为比特流,再进行传输。
在一些实施例中,“CSI实时数据”通常为多维的复数矩阵,其维度和发送、接收天线数量等具体的系统配置相关,矩阵中的每个位置为一个复数。“CSI数据量化”是指:将下行CSI实时数据中的复数值的实部和虚部幅度变换为离散值。
可选的,模型初始化发起端为网络设备,也即由网络设备向终端侧发送模型初始化时,网络设备还可以向终端发送数据请求指示信息,用于表示网络设备向终端请求CSI实时数据以完成下一步的模型选择。
在一些实施例中,“数据请求指示信息”、“数据请求指示”、“CSI实时数据请求指示”、“CSI实时数据请求指示信息”等术语可以相互替换。
在一些实施例中,“CSI实时数据”、“CSI数据”、“实时CSI数据”、“下行CSI数据”、“下行CSI实时数据”、“实时下行CSI数据”等术语可以相互替换。
在一些实施例中,当第一通信设备101为终端,上述第二通信设备102为网络设备,上述第一模型为CSI反馈网络中的译码器模型时,第一通信设备101确定CSI实时数据的一种可能实现方式如下:网络设备发送导频信号。终端接收网络设备发送的导频信号,在当前时刻基于该导频信号进行信道估计,得到上述CSI实时数据。
也就是说,模型初始化发起端为终端,也即由终端向网络设备侧发送模型初始化时,终端可以在当前时刻基于网络设备发送的导频信号进行信道估计,从而得到上述CSI实时数据,便于终端基于上述CSI实时数据进行下一步的模型选择。
步骤S3106,第一通信设备101将上述CSI实时数据输入第一通信设备101所部署的CSI反馈网络。
其中,在一些实施例中,上述CSI反馈网络包括上述第一模型和第二模型,上述CSI反馈网络通过上述第一模型和上述第二模型对上述CSI实时数据进行压缩和重建处理。
示例性的,第一通信设备101可以为网络设备,第二通信设备102可以为终端,则第二通信设备102上配置有CSI反馈网络中的编码器模型,用于对上述CSI实时数据的压缩;第一通信设备101上配置有CSI反馈网络中的译码器模型,用于对终端经过编码器模型压缩后的CSI实时数据进行重建。
示例性的,第一通信设备101可以为终端,第二通信设备102可以为网络设备,则第一通信设备101上配置有CSI反馈网络中的编码器模型,用于对上述CSI实时数据的压缩;第二通信设备102上配置有CSI反馈网络中的译码器模型,用于对终端经过编码器模型压缩后的CSI实时数据进行重建。
在一些实施例中,第一通信设备101上可以部署一个或多个CSI反馈网络。上述多个CSI反馈网络可以是基于不同模型结构构建的CSI反馈网络模型,或者也可以是适用于多个不同信道环境的CSI反馈网络模型。
在一些实施例中,CSI反馈网络可以对输入的CSI实时数据进行压缩和重建处理。当第一通信设备101部署多个CSI反馈网络时,第一通信设备101可以将上述CSI实时数据分别输入至上述多个CSI反馈网络。每个CSI反馈网络均对输入的CSI实时数据进行压缩和重建处理。
可选的,上述第一通信设备101上部署的CSI反馈网络可以是经过训练后(即训练好)的CSI反馈网络模型。
在一些实施例中,第一通信设备101可以为网络设备,第二通信设备102可以为终端;或者,在一些实施例中,第一通信设备101可以为终端,第二通信设备102可以为网络设备。
步骤S3107,第一通信设备101获取上述CSI反馈网络输出的CSI输出值。
在一些实施例中,CSI反馈网络可以对输入的CSI实时数据进行压缩和重建处理,并输出重建处理后的CSI数据,上述CSI反馈网络输出的CSI输出值即为经过重建后得到的CSI数据,从而第一通信设备101可以得到上述CSI反馈网络输出的CSI输出值。
在一些实施例中,第一通信设备101可以为网络设备,第二通信设备102可以为终端;或者,在一些实施例中,第一通信设备101可以为终端,第二通信设备102可以为网络设备。
步骤S3108,第一通信设备101根据上述CSI输出值与上述CSI实时数据,确定上述CSI反馈网络的反馈精确度。
在一些实施例中,可以计算上述CSI输出值与上述CSI实时数据之间的误差,基于该误差即可确定上述CSI反馈网络的反馈精确度。示例性的,误差越大,则上述CSI反馈网络的反馈精确度越低,误差越小,则上述CSI反馈网络的反馈精确度越高。例如,确定上述误差与上述反馈精确度之间的映射关系,可以基于计算得到的上述CSI输出值与上述CSI实时数据之间的误差,从该映射关系中确定出对应的反馈精确度,即为对应CSI反馈网络的反馈精确度。
在一些实施例中,第一通信设备101可以为网络设备,第二通信设备102可以为终端;或者,在一些实施例中,第一通信设备101可以为终端,第二通信设备102可以为网络设备。
步骤S3109,第一通信设备101根据上述反馈精确度,从上述第一通信设备101所部署的CSI反馈网络中选择第一CSI反馈网络。
在一些实施例中,当第一通信设备101所部署的CSI反馈网络为一个,则可以直接将该所部署的CSI反馈网络确定为上述第一CSI反馈网络。
在一些实施例中,当第一通信设备101所部署的CSI反馈网络为多个;第一通信设备101可以从该多个CSI反馈网络中,选择反馈精确度最高的CSI反馈网络确定为第一CSI反馈网络。其中,该反馈精确度的计算方式可参见上述步骤中的描述,在此不再赘述。
在一些实施例中,第一通信设备101可以为网络设备,第二通信设备102可以为终端;或者,在一些实施例中,第一通信设备101可以为终端,第二通信设备102可以为网络设备。
步骤S3110,第一通信设备101基于上述第一CSI反馈网络中第一模型的模型信息和第二通信设备102配置的第一模型的模型信息,确定第二信息。
在一些实施例中,上述第二信息可以承载在模型部署指示信息中发送。
在一些实施例中,上述第二信息可以为待部署在所述第二通信设备中的所述第一模型的模型部署信息。
作为一种示例,上述第二信息可以包括模型编号,示例性的,在第一通信设备上进行模型选择时,所选中的模型的编号包含于第二通信设备上报的模型编号中,则第一通信设备将该模型编号给第二通信设备,便于第二通信设备激活该模型编号所对应的模型。
作为另一种示例,上述第二信息可以包括模型编号和模型参数,在第一通信设备上进行模型选择时,所选中的模型的编号未包含于第二通信设备上报的模型编号中,则第一通信设备将所选中的模型的模型参数和对应的模型编号发给第二通信设备,便于第二通信设备部署该所选中的模型,并激活该模型,便于后续应用。
在一些实施例中,第一通信设备101将上述第一CSI反馈网络中第一模型的模型信息确定为第二信息,其中,第二通信设备102配置的第一模型的模型信息中包含上述第一CSI反馈网络中第一模型的模型信息。
示例性的,当第二通信设备102配置的第一模型的模型信息中包含上述第一CSI反馈网络中第一模型的模型信息时,说明第二通信设备102上配置了上述第一CSI反馈网络中第一模型,第一通信设备101可以将上述第一CSI反馈网络中第一模型的模型信息确定为第二信息,便于第二通信设备102基于该第二信息进行模型部署。
在一些实施例中,第一通信设备101将上述第一CSI反馈网络中第一模型的模型参数进行量化处理,并将经量化处理的上述模型参数确定为第二信息,其中,第二通信设备102配置的第一模型的模型信息中不包含上述第一CSI反馈网络中第一模型的模型信息。
示例性的,当第二通信设备102配置的第一模型的模型信息中不包含上述第一CSI反馈网络中第一模型的模型信息时,说明第二通信设备102上未配置上述第一CSI反馈网络中第一模型,第一通信设备101可以将上述第一CSI反馈网络中第一模型的模型参数进行量化处理,并将经量化处理的上述模型参数确定为第二信息。
在一些实施例中,第一通信设备101在将经量化处理的上述模型参数确定为第二信息之后,可以激活该第一通信设备101上与上述第一模型关联的第二模型,便于第一通信设备101可以通过该激活的第二模型进行CSI相应处理。也就是说,第一通信设备101确定需要在第二通信设备102上部署哪个第一模型时,第一通信设备101需要激活与待部署在第二通信设备102的第一模型关联的第二模型,这样可以通过激活的第一模型和第二模型实现CSI反馈。
可选的,模型参数量化是指:将上述第一模型的权重序列转换为比特形式。可以理解去量化处理是量化处理的逆过程,在此不再赘述。
在一些实施例中,第一通信设备101将上述第一CSI反馈网络中第一模型的模型信息确定为第二信息,其中,第二通信设备102配置的第一模型的模型信息中包含上述第一CSI反馈网络中第一模型的模型信息;第一通信设备101将上述第一CSI反馈网络中第一模型的模型参数进行量化处理,并将经量化处理的上述模型参数确定为第二信息,其中,第二通信设备102配置的第一模型的模型信息中不包含上述第一CSI反馈网络中第一模型的模型信息。
为了方便第二通信设备102将新部署的第一模型进行编号,在一些实施例中,第一通信设备101可以将上述第一CSI反馈网络中第一模型的模型信息和经量化处理的上述模型参数,确定为第二信息。也就是说,第一通信设备101可以将上述第一CSI反馈网络中第一模型的模型信息(如模型编号和/或模型结构)和经量化处理的上述模型参数,均发送给第二通信设备102,便于第二通信设备102对新部署的第一模型进行编号区分。
在一些实施例中,第一通信设备101在将上述第一CSI反馈网络中第一模型的模型信息确定为第二信息之后,可以激活该第一通信设备101上与上述第一模型关联的第二模型,便于第一通信设备101可以通过该激活的第二模型进行CSI相应处理。
在一些实施例中,第一通信设备101可以为网络设备,第二通信设备102可以为终端;或者,在一些实施例中,第一通信设备101可以为终端,第二通信设备102可以为网络设备。
步骤S3111,第一通信设备101发送第二信息。
在一些实施例中,上述第二信息可以为待部署在所述第二通信设备中的所述第一模型的模型部署信息。
在一些实施例中,上述第二信息可以承载在模型部署指示信息中发送。
在一些实施例中,第一通信设备101向第二通信设备102发送上述模型部署指示信息。在一些实施例中,第二通信设备102接收上述模型部署指示信息,可选的,第二通信设备102接收第一通信设备101发送的上述模型部署指示信息。
在一些实施例中,上述模型部署指示信息包括上述第二信息,上述模型部署指示信息用于指示第二通信设备102可以进行模型部署,示例性的,可以指示第二通信设备102基于上述第二信息进行模型部署。
步骤S3112,第二通信设备102收到第二信息,第二通信设备102激活该模型信息对应的第一模型。
在一些实施例中,上述第二信息可以为待部署在所述第二通信设备中的所述第一模型的模型部署信息。示例性的,该第二信息可以为上述第一CSI反馈网络中第一模型的模型信息。
在一些实施例中,上述第二信息可以承载在模型部署指示信息中发送。
在一些实施例中,第一通信设备101可以为网络设备,第二通信设备102可以为终端,则第二通信设备102上配置有CSI反馈网络中的编码器模型,第二通信设备102收到上述第一CSI反馈网络中第一模型的模型信息,第二通信设备102激活该模型信息对应的编码器模型。
在一些实施例中,第一通信设备101可以为终端,第二通信设备102可以为网络设备,则第二通信设备102上配置有CSI反馈网络中的译码器模型,第二通信设备102收到上述第一CSI反馈网络中第一模型的模型信息,第二通信设备102激活该模型信息对应的译码器模型。
步骤S3113,第二通信设备102收到上述第一CSI反馈网络中第一模型的模型参数,或者收到上述第一CSI反馈网络中第一模型的模型参数和模型信息,第二通信设备102将该第一模型的模型参数进行去量化处理,并设定该第一模型的模型编号为该模型信息中的模型编号,激活该第一模型。
在一些实施例中,第一通信设备101可以为网络设备,第二通信设备102可以为终端,则第二通信设备102上配置有CSI反馈网络中的编码器模型,第二通信设备102收到上述第一CSI反馈网络中第一模型的模型参数,或者收到上述第一CSI反馈网络中第一模型的模型参数和模型信息(如模型编号Modelopt),第二通信设备102激活该模型编号Modelopt对应的编码器模型。
在一些实施例中,第一通信设备101可以为终端,第二通信设备102可以为网络设备,则第二通信设备102上配置有CSI反馈网络中的译码器模型,第二通信设备102收到上述第一CSI反馈网络中第一模型的模型信息(如模型编号Modelopt),第二通信设备102激活该模型编号Modelopt对应的译码器模型。
步骤S3114,第二通信设备102发送确认信息。
在一些实施例中,第二通信设备102向第一通信设备101发送上述确认信息。在一些实施例中,第一通信设备101接收上述确认信息,可选的,第一通信设备101接收第二通信设备102发送的上述确认信息。
在一些实施例中,上述确认信息包括用于指示第二通信设备102完成模型初始化的确认指示信息。示例性的,上述确认指示信息可以用Indicatorcomplete表示。
在一些实施例中,第一通信设备101可以为网络设备,第二通信设备102可以为终端,可选的,终端可以使用新部署的编码器模型进行CSI发送,网络设备可以使用该终端新部署的编码器模型所对应的译码器模型进行CSI接收。
在一些实施例中,第一通信设备101可以为终端,第二通信设备102可以为网络设备,可选的,网络设备新部署了译码器模型,终端可以使用该网络设备新部署的译码器模型所对应的编码器模型进行CSI发送,网络设备可以使用该新部署的译码器进行CSI接收。
在一些实施例中,信息等的名称不限定于实施例中所记载的名称,“信息(information)”、“消息(message)”、“信号(signal)”、“信令(signaling)”、“报告(report)”、“配置(configuration)”、“指示(indication)”、“指令(instruction)”、“命令(command)”、“信道”、“参数(parameter)”、“域”、“字段”、“符号(symbol)”、“码元(symbol)”、“码本(codebook)”、“码字(codeword)”、“码点(codepoint)”、“比特(bit)”、“数据(data)”、“程序(program)”、“码片(chip)”等术语可以相互替换。
在一些实施例中,“上行”、“上行链路”、“物理上行链路”等术语可以相互替换,“下行”、“下行链路”、“物理下行链路”等术语可以相互替换,“侧行(side)”、“侧行链路(sidelink)”、“侧行通信”、“侧行链路通信”、“直连”、“直连链路”、“直连通信”、“直连链路通信”等术语可以相互替换。
在一些实施例中,“获取”、“获得”、“得到”、“接收”、“传输”、“双向传输”、“发送和/或接收”可以相互替换,其可以解释为从其他主体接收,从协议中获取,从高层获取,自身处理得到、自主实现等多种含义。
在一些实施例中,“发送”、“发射”、“上报”、“下发”、“传输”、“双向传输”、“发送和/或接收”等术语可以相互替换。
在一些实施例中,“特定(certain)”、“预定(preseted)”、“预设”、“设定”、“指示(indicated)”、“某一”、“任意”、“第一”等术语可以相互替换,“特定A”、“预定A”、“预设A”、“设定A”、“指示A”、“某一A”、“任意A”、“第一A”可以解释为在协议等中预先规定的A,也可以解释为通过设定、配置、或指示等得到的A,也可以解释为特定A、某一A、任意A、或第一A等,但不限于此。
在一些实施例中,判定或判断可以通过以1比特表示的值(0或1)来进行,也可以通过以真(true)或者假(false)表示的真假值(布尔值(boolean))来进行,也可以通过数值的比较(例如,与预定值的比较)来进行,但不限于此。
本公开实施例所涉及的方法可以包括步骤S3101~步骤S3114中的至少一者。例如,步骤S3102+步骤S3105+步骤S3106+步骤S3107+步骤S3108+步骤S3109+步骤S3110+步骤S3111可以作为独立实施例来实施,步骤S3101+步骤S3102+步骤S3105+步骤S3106+步骤S3107+步骤S3108+步骤S3109+步骤S3110+步骤S3111可以作为独立实施例来实施,步骤S3102+步骤S3103+步骤S3104+步骤S3105+步骤S3106+步骤S3107+步骤S3108+步骤S3109+步骤S3110+步骤S3111可以作为独立实施例来实施,步骤S3101+步骤S3102+步骤S3103+步骤S3104+步骤S3105+步骤S3106+步骤S3107+步骤S3108+步骤S3109+步骤S3110+步骤S3111可以作为独立实施例来实施,步骤S3102+步骤S3105+步骤S3106+步骤S3107+步骤S3108+步骤S3109+步骤S3110+步骤S3111+步骤S3112可以作为独立实施例来实施,步骤S3101+步骤S3102+步骤S3105+步骤S3106+步骤S3107+步骤S3108+步骤S3109+步骤S3110+步骤S3111+步骤S3112可以作为独立实施例来实施,步骤S3102+步骤S3103+步骤S3104+步骤S3105+步骤S3106+步骤S3107+步骤S3108+步骤S3109+步骤S3110+步骤S3111+步骤S3112可以作为独立实施例来实施,步骤S3101+步骤S3102+步骤S3103+步骤S3104+步骤S3105+步骤S3106+步骤S3107+步骤S3108+步骤S3109+步骤S3110+步骤S3111+步骤S3112可以作为独立实施例来实施,步骤S3102+步骤S3105+步骤S3106+步骤S3107+步骤S3108+步骤S3109+步骤S3110+步骤S3111+步骤S3113可以作为独立实施例来实施,步骤S3101+步骤S3102+步骤S3105+步骤S3106+步骤S3107+步骤S3108+步骤S3109+步骤S3110+步骤S3111+步骤S3113可以作为独立实施例来实施,步骤S3102+步骤S3103+步骤S3104+步骤S3105+步骤S3106+步骤S3107+步骤S3108+步骤S3109+步骤S3110+步骤S3111+步骤S3113可以作为独立实施例来实施,步骤S3101+步骤S3102+步骤S3103+步骤S3104+步骤S3105+步骤S3106+步骤S3107+步骤S3108+步骤S3109+步骤S3110+步骤S3111+步骤S3113可以作为独立实施例来实施,但不限于此。
在一些实施例中,步骤S3101、步骤S3102、步骤S3103可以交换顺序或同时执行,步骤S3112、步骤S3113可以交换顺序或同时执行。
在一些实施例中,步骤S3101、步骤S3103、步骤S3104、步骤S3112、步骤S3113、步骤S3114是可选地,在不同实施例中可以对这些步骤中的一个或多个步骤进行省略或替代。
在一些实施例中,步骤S3101、步骤S3112、步骤S3113、步骤S3114是可选地,在不同实施例中可以对这些步骤中的一个或多个步骤进行省略或替代。
在一些实施例中,步骤S3101、步骤S3113、步骤S3114是可选地,在不同实施例中可以对这些步骤中的一个或多个步骤进行省略或替代。
在一些实施例中,步骤S3101、步骤S3112、步骤S3114是可选地,在不同实施例中可以对这些步骤中的一个或多个步骤进行省略或替代。
在一些实施例中,可参见图3所对应的说明书之前或之后记载的其他可选实现方式。
图4A是根据本公开实施例示出的一种模型初始化方法的流程示意图。如图4A所示,本公开实施例涉及的方法可由第一通信设备101执行,示例性的,该第一通信设备101为网络设备,本实施例中的第二通信设备102可以为终端,也即是说,该方法可以从网络设备侧描述。上述方法可以包括但不限于如下步骤。
步骤S4101,接收第一指示信息。
在一些实施例中,第二通信设备102(如终端)向第一通信设备101发送上述第一指示信息。在一些实施例中,第一通信设备101(如网络设备)接收上述第一指示信息,可选的,网络设备接收终端发送的上述第一指示信息。
在一些实施例中,上述第一指示信息用于指示第二通信设备102配置了第一模型,且具备使用第一模型完成CSI反馈工作的能力。在一些实施例中,上述第一模型为用于CSI反馈的模型,作为一种示例,第一模型可以是CSI反馈网络中的编码器模型,或者也可以是CSI反馈网络中的译码器模型。
在一些实施例中,第一通信设备101为网络设备,第二通信设备102为终端,第一模型为CSI反馈网络中的编码器模型。可选的,第一通信设备101上配置有CSI反馈网络中的译码器模型,第二通信设备102上配置有CSI反馈网络中的编码器模型。
步骤S4102,接收第一信息。
在一些实施例中,上述第一信息包括由第二通信设备102(如终端)配置的第一模型的模型信息。其中,针对第一模型的描述可参见上述步骤S4101的相关描述,在此不再赘述。
步骤S4102的可选实现方式可以参见图3的步骤S3102的可选实现方式、及图3所涉及的实施例中其他关联部分,此处不再赘述。
步骤S4103,发送第二指示信息。
在一些实施例中,上述第二指示信息用于指示第一通信设备101(如网络设备)通知第二通信设备102(如终端)进入模型初始化流程。
步骤S4103的可选实现方式可以参见图3的步骤S3103的可选实现方式、及图3所涉及的实施例中其他关联部分,此处不再赘述。
步骤S4104,接收第三指示信息。
在一些实施例中,上述第三指示信息用于指示第二通信设备102(如终端)接受了模型初始化指令,进入模型初始化流程。
步骤S4104的可选实现方式可以参见图3的步骤S3104的可选实现方式、及图3所涉及的实施例中其他关联部分,此处不再赘述。
步骤S4105,发送导频信号和数据请求指示信息。
在一些实施例中,上述导频信号用于终端在当前时刻进行信道估计以得到CSI实时数据。在一些实施例中,上述数据请求指示信息用于指示网络设备向终端请求CSI实时数据。可选的,终端接收上述导频信号和上述数据请求指示信息,例如,终端收到网络设备发送的上述导频信号和上述数据请求指示信息,基于上述导频信号在当前时刻进行信道估计,以得到上述CSI实时数据。终端发送上述CSI实时数据,如终端向网络设备发送上述CSI实时数据。网络设备接收终端发送的上述CSI实时数据,从而得到上述CSI实时数据,便于网络设备基于上述CSI实时数据进行下一步的模型选择。需要说明的是,网络设备在完成模型选择后,需要将与选择的第一模型关联的第二模型进行激活,便于网络设备通过激活的第二模型进行CSI重建。
步骤S4105的可选实现方式可以参见图3的步骤S3105的可选实现方式、及图3所涉及的实施例中其他关联部分,此处不再赘述。
步骤S4106,接收经量化处理的CSI实时数据。
在一些实施例中,网络设备接收经量化处理的CSI实时数据。在一些实施例中,终端向向网络设备发送经过量化处理的CSI实时数据,网络设备接收终端设备发送的所述经量化处理的CSI实时数据。
步骤S4107,将接收到的经量化处理的CSI实时数据进行去量化处理,得到CSI实时数据。
在一些实施例中,网络设备可以将收到的经量化处理的CSI实时数据进行去量化处理,得到终端设备对当前信道环境的信道估计后的CSI实时数据。
步骤S4108,将上述CSI实时数据输入网络设备上所部署的CSI反馈网络。
其中,在一些实施例中,上述CSI反馈网络包括上述第一模型和第二模型,上述CSI反馈网络通过上述第一模型和上述第二模型对上述CSI实时数据进行压缩和重建处理。
步骤S4108的可选实现方式可以参见图3的步骤S3106的可选实现方式、及图3所涉及的实施例中其他关联部分,此处不再赘述。
步骤S4109,获取上述CSI反馈网络输出的CSI输出值。
步骤S4109的可选实现方式可以参见图3的步骤S3107的可选实现方式、及图3所涉及的实施例中其他关联部分,此处不再赘述。
步骤S4110,根据上述CSI输出值与上述CSI实时数据,确定上述CSI反馈网络的反馈精确度。
步骤S4110的可选实现方式可以参见图3的步骤S3108的可选实现方式、及图3所涉及的实施例中其他关联部分,此处不再赘述。
步骤S4111,根据上述反馈精确度,从上述网络设备上所部署的CSI反馈网络中选择第一CSI反馈网络。
步骤S4111的可选实现方式可以参见图3的步骤S3109的可选实现方式、及图3所涉及的实施例中其他关联部分,此处不再赘述。
步骤S4112,基于上述第一CSI反馈网络中第一模型的模型信息和终端上配置的第一模型的模型信息,确定第二信息。
步骤S4112的可选实现方式可以参见图3的步骤S3110的可选实现方式、及图3所涉及的实施例中其他关联部分,此处不再赘述。
步骤S4113,发送第二信息。
在一些实施例中,上述第二信息可以为待部署在所述第二通信设备中的所述第一模型的模型部署信息。
在一些实施例中,上述第二信息可以承载在模型部署指示信息中发送。
在一些实施例中,上述模型部署指示信息包括上述第二信息,上述模型部署指示信息用于指示终端可以进行模型部署,示例性的,可以指示终端基于上述第二信息进行模型部署。
步骤S4113的可选实现方式可以参见图3的步骤S3111的可选实现方式、及图3所涉及的实施例中其他关联部分,此处不再赘述。
步骤S4114,接收确认信息。
在一些实施例中,上述确认信息包括用于指示终端完成模型初始化的确认指示信息。示例性的,上述确认指示信息可以用Indicatorcomplete表示。
步骤S4114的可选实现方式可以参见图3的步骤S3114的可选实现方式、及图3所涉及的实施例中其他关联部分,此处不再赘述。
本公开实施例所涉及的方法可以包括步骤S4101~步骤S4114中的至少一者。例如,步骤S4102+步骤S4105+步骤S4106+步骤S4107+步骤S4108+步骤S4109+步骤S4110+步骤S4111+步骤S4112+步骤S4113可以作为独立实施例来实施,步骤S4102+步骤S4105+步骤S4106+步骤S4107+步骤S4108+步骤S4109+步骤S4110+步骤S4111+步骤S4112+步骤S4113+步骤S4114可以作为独立实施例来实施,步骤S4102+步骤S4103+步骤S4104+步骤S4105+步骤S4106+步骤S4107+步骤S4108+步骤S4109+步骤S4110+步骤S4111+步骤S4112+步骤S4113可以作为独立实施例来实施,步骤S4102+步骤S4103+步骤S4104+步骤S4105+步骤S4106+步骤S4107+步骤S4108+步骤S4109+步骤S4110+步骤S4111+步骤S4112+步骤S4113+步骤S4114可以作为独立实施例来实施,步骤S4101+步骤S4102+步骤S4105+步骤S4106+步骤S4107+步骤S4108+步骤S4109+步骤S4110+步骤S4111+步骤S4112+步骤S4113可以作为独立实施例来实施,步骤S4101+步骤S4102+步骤S4105+步骤S4106+步骤S4107+步骤S4108+步骤S4109+步骤S4110+步骤S4111+步骤S4112+步骤S4113+步骤S4114可以作为独立实施例来实施,步骤S4101+步骤S4102+步骤S4103+步骤S4104+步骤S4105+步骤S4106+步骤S4107+步骤S4108+步骤S4109+步骤S4110+步骤S4111+步骤S4112+步骤S4113可以作为独立实施例来实施,步骤S4101+步骤S4102+步骤S4103+步骤S4104+步骤S4105+步骤S4106+步骤S4107+步骤S4108+步骤S4109+步骤S4110+步骤S4111+步骤S4112+步骤S4113+步骤S4114可以作为独立实施例来实施,但不限于此。
在一些实施例中,步骤S4101、步骤S4102、步骤S4103可以交换顺序或同时执行。
在一些实施例中,步骤S4101、步骤S4103、步骤S4104、步骤S4114是可选地,在不同实施例中可以对这些步骤中的一个或多个步骤进行省略或替代。
在一些实施例中,步骤S4101、步骤S4103、步骤S4104、是可选地,在不同实施例中可以对这些步骤中的一个或多个步骤进行省略或替代。
在一些实施例中,步骤S4101、步骤S4114是可选地,在不同实施例中可以对这些步骤中的一个或多个步骤进行省略或替代。
在一些实施例中,步骤S4114是可选地,在不同实施例中可以对这些步骤中的一个或多个步骤进行省略或替代。
在一些实施例中,步骤S4101是可选地,在不同实施例中可以对这些步骤中的一个或多个步骤进行省略或替代。
在一些实施例中,步骤S4103、步骤S4104、步骤S4114是可选地,在不同实施例中可以对这些步骤中的一个或多个步骤进行省略或替代。
在一些实施例中,步骤S4103、步骤S4104是可选地,在不同实施例中可以对这些步骤中的一个或多个步骤进行省略或替代。
图4B是根据本公开实施例示出的一种模型初始化方法的流程示意图。如图4B所示,本公开实施例涉及的方法可由第一通信设备101执行,示例性的,该第一通信设备101为终端,本实施例中的第二通信设备102为网络设备,也就是说,该方法可以从终端侧描述。上述方法可以包括但不限于如下步骤。
步骤S4201,接收第一指示信息。
在一些实施例中,第二通信设备102(如网络设备)向第一通信设备101(如终端)发送上述第一指示信息。在一些实施例中,第一通信设备101(如终端)接收上述第一指示信息,可选的,终端接收网络设备发送的上述第一指示信息。
在一些实施例中,上述第一指示信息用于指示第二通信设备102配置了第一模型,且具备使用第一模型完成CSI反馈工作的能力。在一些实施例中,上述第一模型为用于CSI反馈的模型,作为一种示例,第一模型可以是CSI反馈网络中的编码器模型,或者也可以是CSI反馈网络中的译码器模型。
步骤S4201的可选实现方式可以参见图3的步骤S3101的可选实现方式、及图3所涉及的实施例中其他关联部分,此处不再赘述。
步骤S4202,接收第一信息。
在一些实施例中,上述第一信息包括由第二通信设备102(如网络设备)配置的第一模型的模型信息。其中,针对第一模型的描述可参见上述步骤S4101的相关描述,在此不再赘述。
步骤S4202的可选实现方式可以参见图3的步骤S3102的可选实现方式、及图3所涉及的实施例中其他关联部分,此处不再赘述。
步骤S4203,发送第二指示信息。
在一些实施例中,上述第二指示信息用于指示第一通信设备101(如终端)通知第二通信设备102(如网络设备)进入模型初始化流程。
步骤S4203的可选实现方式可以参见图3的步骤S3103的可选实现方式、及图3所涉及的实施例中其他关联部分,此处不再赘述。
步骤S4204,接收第三指示信息。
在一些实施例中,上述第三指示信息用于指示第二通信设备102(如网络设备)接受了模型初始化指令,进入模型初始化流程。
步骤S4204的可选实现方式可以参见图3的步骤S3104的可选实现方式、及图3所涉及的实施例中其他关联部分,此处不再赘述。
步骤S4205,接收导频信号。
在一些实施例中,终端设备接收上述导频信号。在一些实施例中,网络设备发送上述导频信号,终端接收网络设备发送的上述导频信号。
在一些实施例中,上述导频信号用于终端在当前时刻进行信道估计以得到CSI实时数据。
步骤S4205的可选实现方式可以参见图3的步骤S3105的可选实现方式、及图3所涉及的实施例中其他关联部分,此处不再赘述。
步骤S4206,在当前时刻基于上述导频信号进行信道估计,得到CSI实时数据。
步骤S4206的可选实现方式可以参见图3的步骤S3105的可选实现方式、及图3所涉及的实施例中其他关联部分,此处不再赘述。
步骤S4207,将上述CSI实时数据输入终端上所部署的CSI反馈网络。
其中,在一些实施例中,上述CSI反馈网络包括上述第一模型和第二模型,上述CSI反馈网络通过上述第一模型和上述第二模型对上述CSI实时数据进行压缩和重建处理。
步骤S4207的可选实现方式可以参见图3的步骤S3106的可选实现方式、及图3所涉及的实施例中其他关联部分,此处不再赘述。
步骤S4208,获取上述CSI反馈网络输出的CSI输出值。
步骤S4208的可选实现方式可以参见图3的步骤S3107的可选实现方式、及图3所涉及的实施例中其他关联部分,此处不再赘述。
步骤S4209,根据上述CSI输出值与上述CSI实时数据,确定上述CSI反馈网络的反馈精确度。
步骤S4209的可选实现方式可以参见图3的步骤S3108的可选实现方式、及图3所涉及的实施例中其他关联部分,此处不再赘述。
步骤S4210,根据上述反馈精确度,从上述终端上所部署的CSI反馈网络中选择第一CSI反馈网络。
步骤S4210的可选实现方式可以参见图3的步骤S3109的可选实现方式、及图3所涉及的实施例中其他关联部分,此处不再赘述。
步骤S4211,基于上述第一CSI反馈网络中第一模型的模型信息和网络设备上配置的第一模型的模型信息,确定第二信息。
步骤S4211的可选实现方式可以参见图3的步骤S3110的可选实现方式、及图3所涉及的实施例中其他关联部分,此处不再赘述。
步骤S4212,发送第二信息。
在一些实施例中,上述第二信息可以为待部署在所述第二通信设备中的所述第一模型的模型部署信息。
在一些实施例中,上述第二信息可以承载在模型部署指示信息中发送。
在一些实施例中,上述模型部署指示信息包括上述第二信息,上述模型部署指示信息用于指示网络设备可以进行模型部署,示例性的,可以指示网络设备基于上述第二信息进行模型部署。
步骤S4212的可选实现方式可以参见图3的步骤S3111的可选实现方式、及图3所涉及的实施例中其他关联部分,此处不再赘述。
步骤S4213,接收确认信息。
在一些实施例中,上述确认信息包括用于指示网络设备完成模型初始化的确认指示信息。示例性的,上述确认指示信息可以用Iddicatorcomplete表示。
步骤S4213的可选实现方式可以参见图3的步骤S3114的可选实现方式、及图3所涉及的实施例中其他关联部分,此处不再赘述。
本公开实施例所涉及的方法可以包括步骤S4201~步骤S4213中的至少一者。例如,步骤S4202+步骤S4205+步骤S4206+步骤S4207+步骤S4208+步骤S4209+步骤S4210+步骤S4211+步骤S4212可以作为独立实施例来实施,步骤S4202+步骤S4205+步骤S4206+步骤S4207+步骤S4208+步骤S4209+步骤S4210+步骤S4211+步骤S4212+步骤S4213可以作为独立实施例来实施,步骤S4202+步骤S4203+步骤S4204+步骤S4205+步骤S4206+步骤S4207+步骤S4208+步骤S4209+步骤S4210+步骤S4211+步骤S4212可以作为独立实施例来实施,步骤S4202+步骤S4203+步骤S4204+步骤S4205+步骤S4206+步骤S4207+步骤S4208+步骤S4209+步骤S4210+步骤S4211+步骤S4212+步骤S4213可以作为独立实施例来实施,步骤S4201+步骤S4202+步骤S4205+步骤S4206+步骤S4207+步骤S4208+步骤S4209+步骤S4210+步骤S4211+步骤S4212可以作为独立实施例来实施,步骤S4201+步骤S4202+步骤S4205+步骤S4206+步骤S4207+步骤S4208+步骤S4209+步骤S4210+步骤S4211+步骤S4212+步骤S4213可以作为独立实施例来实施,步骤S4201+步骤S4202+步骤S4203+步骤S4204+步骤S4205+步骤S4206+步骤S4207+步骤S4208+步骤S4209+步骤S4210+步骤S4211+步骤S4212可以作为独立实施例来实施,步骤S4201+步骤S4202+步骤S4203+步骤S4204+步骤S4205+步骤S4206+步骤S4207+步骤S4208+步骤S4209+步骤S4210+步骤S4211+步骤S4212+步骤S4213可以作为独立实施例来实施,但不限于此。
在一些实施例中,步骤S4201、步骤S4202、步骤S4203可以交换顺序或同时执行。
在一些实施例中,步骤S4201、步骤S4203、步骤S4204、步骤S4213是可选地,在不同实施例中可以对这些步骤中的一个或多个步骤进行省略或替代。
在一些实施例中,步骤S4201、步骤S4203、步骤S4204、是可选地,在不同实施例中可以对这些步骤中的一个或多个步骤进行省略或替代。
在一些实施例中,步骤S4201、步骤S4213是可选地,在不同实施例中可以对这些步骤中的一个或多个步骤进行省略或替代。
在一些实施例中,步骤S4213是可选地,在不同实施例中可以对这些步骤中的一个或多个步骤进行省略或替代。
在一些实施例中,步骤S4201是可选地,在不同实施例中可以对这些步骤中的一个或多个步骤进行省略或替代。
在一些实施例中,步骤S4203、步骤S4204、步骤S4213是可选地,在不同实施例中可以对这些步骤中的一个或多个步骤进行省略或替代。
在一些实施例中,步骤S4203、步骤S4204是可选地,在不同实施例中可以对这些步骤中的一个或多个步骤进行省略或替代。
图4C是根据本公开实施例示出的一种模型初始化方法的流程示意图。如图4C所示,本公开实施例涉及的方法可由第一通信设备101执行。上述方法可以包括但不限于如下步骤。
步骤S4301,接收第一指示信息。
步骤S4301的可选实现方式可以参见图3的步骤S3101的可选实现方式、及图3所涉及的实施例中其他关联部分,此处不再赘述。
步骤S4302,接收第一信息。
在一些实施例中,上述第一信息由包括第二通信设备102配置的第一模型的模型信息。其中,针对第一模型的描述可参见上述步骤S3101的相关描述,在此不再赘述。
步骤S4302的可选实现方式可以参见图3的步骤S3102的可选实现方式、及图3所涉及的实施例中其他关联部分,此处不再赘述。
步骤S4303,发送第二指示信息。
在一些实施例中,上述第二指示信息用于指示第一通信设备101通知第二通信设备102进入模型初始化流程。
步骤S4303的可选实现方式可以参见图3的步骤S3103的可选实现方式、及图3所涉及的实施例中其他关联部分,此处不再赘述。
步骤S4304,接收第三指示信息。
在一些实施例中,上述第三指示信息用于指示第二通信设备102接受了模型初始化指令,进入模型初始化流程。
步骤S4304的可选实现方式可以参见图3的步骤S3104的可选实现方式、及图3所涉及的实施例中其他关联部分,此处不再赘述。
步骤S4305,确定CSI实时数据。
在一些实施例中,上述CSI实时数据是指终端在当前时刻通过信道估计所获取的下行CSI,能够在一定程度上代表当前信道环境下的CSI数据特征。
步骤S4305的可选实现方式可以参见图3的步骤S3105的可选实现方式、及图3所涉及的实施例中其他关联部分,此处不再赘述。
步骤S4306,将上述CSI实时数据输入第一通信设备101所部署的CSI反馈网络。
步骤S4306的可选实现方式可以参见图3的步骤S3106的可选实现方式、及图3所涉及的实施例中其他关联部分,此处不再赘述。
步骤S4307,获取上述CSI反馈网络输出的CSI输出值。
步骤S4307的可选实现方式可以参见图3的步骤S3107的可选实现方式、及图3所涉及的实施例中其他关联部分,此处不再赘述。
步骤S4308,根据上述CSI输出值与上述CSI实时数据,确定上述CSI反馈网络的反馈精确度。
步骤S4308的可选实现方式可以参见图3的步骤S3108的可选实现方式、及图3所涉及的实施例中其他关联部分,此处不再赘述。
步骤S4309,根据上述反馈精确度,从上述第一通信设备101所部署的CSI反馈网络中选择第一CSI反馈网络。
步骤S4309的可选实现方式可以参见图3的步骤S3109的可选实现方式、及图3所涉及的实施例中其他关联部分,此处不再赘述。
步骤S4310,基于上述第一CSI反馈网络中第一模型的模型信息和第二通信设备102配置的第一模型的模型信息,确定第二信息。
步骤S4310的可选实现方式可以参见图3的步骤S3110的可选实现方式、及图3所涉及的实施例中其他关联部分,此处不再赘述。
步骤S4311,发送第二信息。
在一些实施例中,上述第二信息可以为待部署在所述第二通信设备中的所述第一模型的模型部署信息。
在一些实施例中,上述第二信息可以承载在模型部署指示信息中发送。
在一些实施例中,上述模型部署指示信息包括上述第二信息,上述模型部署指示信息用于指示第二通信设备102可以进行模型部署,示例性的,可以指示第二通信设备102基于上述第二信息进行模型部署。
步骤S4311的可选实现方式可以参见图3的步骤S3111的可选实现方式、及图3所涉及的实施例中其他关联部分,此处不再赘述。
步骤S4312,接收确认信息。
在一些实施例中,上述确认信息包括用于指示第二通信设备102完成模型初始化的确认指示信息。示例性的,上述确认指示信息可以用Indicatorcomplete表示。
步骤S4312的可选实现方式可以参见图3的步骤S3114的可选实现方式、及图3所涉及的实施例中其他关联部分,此处不再赘述。
本公开实施例所涉及的方法可以包括步骤S4301~步骤S4312中的至少一者。例如,步骤S4302+步骤S4305+步骤S4306+步骤S4307+步骤S4308+步骤S4309+步骤S4310+步骤S4311可以作为独立实施例来实施,步骤S4301+步骤S4302+步骤S4305+步骤S4306+步骤S4307+步骤S4308+步骤S4309+步骤S4310+步骤S4311可以作为独立实施例来实施,步骤S4302+步骤S4303+步骤S4304+步骤S4305+步骤S4306+步骤S4307+步骤S4308+步骤S4309+步骤S4310+步骤S4311可以作为独立实施例来实施,步骤S4301+步骤S4302+步骤S4303+步骤S4304+步骤S4305+步骤S4306+步骤S4307+步骤S4308+步骤S4309+步骤S4310+步骤S4311可以作为独立实施例来实施,步骤S4302+步骤S4305+步骤S4306+步骤S4307+步骤S4308+步骤S4309+步骤S4310+步骤S4311+步骤S4312可以作为独立实施例来实施,步骤S4301+步骤S4302+步骤S4305+步骤S4306+步骤S4307+步骤S4308+步骤S4309+步骤S4310+步骤S4311+步骤S4312可以作为独立实施例来实施,步骤S4302+步骤S4303+步骤S4304+步骤S4305+步骤S4306+步骤S4307+步骤S4308+步骤S4309+步骤S4310+步骤S4311+步骤S4312可以作为独立实施例来实施,步骤S4301+步骤S4302+步骤S4303+步骤S4304+步骤S4305+步骤S4306+步骤S4307+步骤S4308+步骤S4309+步骤S4310+步骤S4311+步骤S4312可以作为独立实施例来实施,但不限于此。
在一些实施例中,步骤S4301、步骤S4303、步骤S4304、步骤S4312是可选地,在不同实施例中可以对这些步骤中的一个或多个步骤进行省略或替代。
在一些实施例中,步骤S4301、步骤S4312是可选地,在不同实施例中可以对这些步骤中的一个或多个步骤进行省略或替代。
在一些实施例中,步骤S4301、步骤S4303、步骤S4304是可选地,在不同实施例中可以对这些步骤中的一个或多个步骤进行省略或替代。
在一些实施例中,步骤S4301是可选地,在不同实施例中可以对这些步骤中的一个或多个步骤进行省略或替代。
在一些实施例中,步骤S4303、步骤S4304、步骤S4312是可选地,在不同实施例中可以对这些步骤中的一个或多个步骤进行省略或替代。
在一些实施例中,步骤S4303、步骤S4304是可选地,在不同实施例中可以对这些步骤中的一个或多个步骤进行省略或替代。
在一些实施例中,步骤S4312是可选地,在不同实施例中可以对这些步骤中的一个或多个步骤进行省略或替代。
图5A是根据本公开实施例示出的模型初始化方法的流程示意图。如图5A所示,本公开实施例涉及模型初始化方法可由第二通信设备102执行,示例性的,该第二通信设备102为终端,本实施例中的第一通信设备101为网络设备,上述方法可以包括但不限于如下步骤。
步骤S5101,发送第一指示信息。
在一些实施例中,上述第一指示信息用于指示第二通信设备102配置了第一模型,且具备使用第一模型完成CSI反馈工作的能力。在一些实施例中,上述第一模型为用于CSI反馈的模型,作为一种示例,第一模型可以是CSI反馈网络中的编码器模型,或者也可以是CSI反馈网络中的译码器模型。
步骤S5101的可选实现方式可以参见图3的步骤S3101的可选实现方式、及图3所涉及的实施例中其他关联部分,此处不再赘述。
步骤S5102,发送第一信息。
在一些实施例中,上述第一信息包括由第二通信设备102配置的第一模型的模型信息。
步骤S5102的可选实现方式可以参见图3的步骤S3102的可选实现方式、及图3所涉及的实施例中其他关联部分,此处不再赘述。
步骤S5103,接收第二指示信息。
在一些实施例中,上述第二指示信息用于指示第一通信设备101通知第二通信设备102进入模型初始化流程。
步骤S5103的可选实现方式可以参见图3的步骤S3103的可选实现方式、及图3所涉及的实施例中其他关联部分,此处不再赘述。
步骤S5104,发送第三指示信息。
在一些实施例中,上述第三指示信息用于指示第二通信设备102接受了模型初始化指令,进入模型初始化流程。可选的,第一通信设备101接收到第二通信设备102发送的上述第三指示信息,则可以说明第二通信设备102收到了第一通信设备101发送的上述第二指示信息,且接受了对第二通信设备上CSI反馈网络模型的初始化指令,进入模型初始化流程。
步骤S5104的可选实现方式可以参见图3的步骤S3104的可选实现方式、及图3所涉及的实施例中其他关联部分,此处不再赘述。
步骤S5105,接收导频信号和数据请求指示信息。
在一些实施例中,上述导频信号用于终端在当前时刻进行信道估计以得到CSI实时数据。
在一些实施例中,上述数据请求指示信息用于指示网络设备向终端请求CSI实时数据。
步骤S5105的可选实现方式可以参见图3的步骤S3105的可选实现方式、及图3所涉及的实施例中其他关联部分,此处不再赘述。
步骤S5106,基于上述导频信号在当前时刻进行信道估计,以得到CSI实时数据。
步骤S5106的可选实现方式可以参见图3的步骤S3105的可选实现方式、及图3所涉及的实施例中其他关联部分,此处不再赘述。
步骤S5107,发送上述CSI实时数据。
步骤S5107的可选实现方式可以参见图3的步骤S3105的可选实现方式、及图3所涉及的实施例中其他关联部分,此处不再赘述。
在一些实施例中,上述CSI实时数据用于网络设备基于该CSI实时数据进行下一步的模型选择。其的可选实现方式可以参见图3的步骤S3105、步骤S3106、步骤S3107、步骤S3108、步骤S3109和步骤S3110的可选实现方式、及图3所涉及的实施例中其他关联部分,此处不再赘述。
步骤S5108,接收第二信息。
在一些实施例中,上述第二信息可以为待部署在所述第二通信设备中的所述第一模型的模型部署信息。
在一些实施例中,上述第二信息可以承载在模型部署指示信息中发送。
在一些实施例中,上述模型部署指示信息包括上述第二信息,上述模型部署指示信息用于指示第二通信设备102可以进行模型部署,示例性的,可以指示第二通信设备102基于上述第二信息进行模型部署。
步骤S5108的可选实现方式可以参见图3的步骤S3111的可选实现方式、及图3所涉及的实施例中其他关联部分,此处不再赘述。
步骤S5109,收到上述第一CSI反馈网络中第一模型的模型信息,第二通信设备102激活该模型信息对应的第一模型。
步骤S5109的可选实现方式可以参见图3的步骤S3112的可选实现方式、及图3所涉及的实施例中其他关联部分,此处不再赘述。
步骤S5110,收到上述第一CSI反馈网络中第一模型的模型参数,或者收到上述第一CSI反馈网络中第一模型的模型参数和模型信息,第二通信设备102将该第一模型的模型参数进行去量化处理,并设定该第一模型的模型编号为该模型信息中的模型编号,激活第一模型。
步骤S5110的可选实现方式可以参见图3的步骤S3113的可选实现方式、及图3所涉及的实施例中其他关联部分,此处不再赘述。
步骤S5111,发送确认信息。
在一些实施例中,上述确认信息包括用于指示第二通信设备102完成模型初始化的确认指示信息。示例性的,上述确认指示信息可以用Indicatorcomplete表示。
步骤S5111的可选实现方式可以参见图3的步骤S3114的可选实现方式、及图3所涉及的实施例中其他关联部分,此处不再赘述。
本公开实施例所涉及的方法可以包括步骤S5101~步骤S5111中的至少一者。例如,步骤S5102+步骤S5105+步骤S5106+步骤S5107+步骤S5108+步骤S5109可以作为独立实施例来实施,步骤S5102+步骤S5105+步骤S5106+步骤S5107+步骤S5108+步骤S5110可以作为独立实施例来实施,步骤S5102+步骤S5105+步骤S5106+步骤S5107+步骤S5108+步骤S5109+步骤S5111可以作为独立实施例来实施,步骤S5102+步骤S5105+步骤S5106+步骤S5107+步骤S5108+步骤S5110+步骤S5111可以作为独立实施例来实施,步骤S5101+步骤S5102+步骤S5105+步骤S5106+步骤S5107+步骤S5108+步骤S5109可以作为独立实施例来实施,步骤S5101+步骤S5102+步骤S5105+步骤S5106+步骤S5107+步骤S5108+步骤S5110可以作为独立实施例来实施,步骤S5101+步骤S5102+步骤S5105+步骤S5106+步骤S5107+步骤S5108+步骤S5109+步骤S5111可以作为独立实施例来实施,步骤S5101+步骤S5102+步骤S5105+步骤S5106+步骤S5107+步骤S5108+步骤S5110+步骤S5111可以作为独立实施例来实施,步骤S5101+步骤S5102+步骤S5103+步骤S5104+步骤S5105+步骤S5106+步骤S5107+步骤S5108+步骤S5109可以作为独立实施例来实施,步骤S5101+步骤S5102+步骤S5103+步骤S5104+步骤S5105+步骤S5106+步骤S5107+步骤S5108+步骤S5110可以作为独立实施例来实施,步骤S5101+步骤S5102+步骤S5103+步骤S5104+步骤S5105+步骤S5106+步骤S5107+步骤S5108+步骤S5109+步骤S5111可以作为独立实施例来实施,步骤S5101+步骤S5102+步骤S5103+步骤S5104+步骤S5105+步骤S5106+步骤S5107+步骤S5108+步骤S5110+步骤S5111可以作为独立实施例来实施,但不限于此。
在一些实施例中,步骤S5101、步骤S5102、步骤S5103可以交换顺序或同时执行,步骤S5112、步骤S5113可以交换顺序或同时执行。
在一些实施例中,步骤S5101、步骤S5103、步骤S5104、步骤S5110、步骤S5111是可选地,在不同实施例中可以对这些步骤中的一个或多个步骤进行省略或替代。
在一些实施例中,步骤S5101、步骤S5103、步骤S5104、步骤S5109、步骤S5111是可选地,在不同实施例中可以对这些步骤中的一个或多个步骤进行省略或替代。
在一些实施例中,步骤S5103、步骤S5104、步骤S5110、步骤S5111是可选地,在不同实施例中可以对这些步骤中的一个或多个步骤进行省略或替代。
在一些实施例中,步骤S5103、步骤S5104、步骤S5109、步骤S5111是可选地,在不同实施例中可以对这些步骤中的一个或多个步骤进行省略或替代。
在一些实施例中,步骤S5101、步骤S5103、步骤S5104、步骤S5109是可选地,在不同实施例中可以对这些步骤中的一个或多个步骤进行省略或替代。
在一些实施例中,步骤S5101、步骤S5103、步骤S5104、步骤S5110是可选地,在不同实施例中可以对这些步骤中的一个或多个步骤进行省略或替代。
图5B是根据本公开实施例示出的模型初始化方法的流程示意图。如图5A所示,本公开实施例涉及模型初始化方法可由第二通信设备102执行,示例性的,该第二通信设备102为网络设备,本实施例中的第一通信设备101为终端,上述方法可以包括但不限于如下步骤。
步骤S5201,发送第一指示信息。
在一些实施例中,上述第一指示信息用于指示第二通信设备102配置了第一模型,且具备使用第一模型完成CSI反馈工作的能力。在一些实施例中,上述第一模型为用于CSI反馈的模型,作为一种示例,第一模型可以是CSI反馈网络中的编码器模型,或者也可以是CSI反馈网络中的译码器模型。
步骤S5201的可选实现方式可以参见图3的步骤S3101的可选实现方式、及图3所涉及的实施例中其他关联部分,此处不再赘述。
步骤S5202,发送第一信息。
在一些实施例中,上述第一信息包括由第二通信设备102配置的第一模型的模型信息。
步骤S5202的可选实现方式可以参见图3的步骤S3102的可选实现方式、及图3所涉及的实施例中其他关联部分,此处不再赘述。
步骤S5203,接收第二指示信息。
在一些实施例中,上述第二指示信息用于指示第一通信设备101通知第二通信设备102进入模型初始化流程。
步骤S5203的可选实现方式可以参见图3的步骤S3103的可选实现方式、及图3所涉及的实施例中其他关联部分,此处不再赘述。
步骤S5204,发送第三指示信息。
在一些实施例中,上述第三指示信息用于指示第二通信设备102接受了模型初始化指令,进入模型初始化流程。可选的,第一通信设备101接收到第二通信设备102发送的上述第三指示信息,则可以说明第二通信设备102收到了第一通信设备101发送的上述第二指示信息,且接受了对第二通信设备上CSI反馈网络模型的初始化指令,进入模型初始化流程。
步骤S5204的可选实现方式可以参见图3的步骤S3104的可选实现方式、及图3所涉及的实施例中其他关联部分,此处不再赘述。
步骤S5205,发送导频信号。
在一些实施例中,网络设备发送导频信号。可选的,网络设备向终端发送上述导频信号。终端接收网络设备发送的上述导频信号。
在一些实施例中,上述导频信号用于终端在当前时刻进行信道估计以得到CSI实时数据。
步骤S5205的可选实现方式可以参见图3的步骤S3105的可选实现方式、及图3所涉及的实施例中其他关联部分,此处不再赘述。
在一些实施例中,上述CSI实时数据用于终端进行下一步的模型选择。可选的,终端在当前时刻进行信道估计以得到CSI实时数据后,可以基于该CSI实时数据进行下一步的模型选择。其的可选实现方式可以参见图3的步骤S3106、步骤S3107、步骤S3108、步骤S3109、步骤S3110的可选实现方式、及图3所涉及的实施例中其他关联部分,此处不再赘述。
步骤S5206,接收第二信息。
在一些实施例中,上述第二信息可以为待部署在所述第二通信设备中的所述第一模型的模型部署信息。
在一些实施例中,上述第二信息可以承载在模型部署指示信息中发送。
在一些实施例中,上述模型部署指示信息包括上述第二信息,上述模型部署指示信息用于指示第二通信设备102可以进行模型部署,示例性的,可以指示第二通信设备102基于上述第二信息进行模型部署。
步骤S5206的可选实现方式可以参见图3的步骤S3111的可选实现方式、及图3所涉及的实施例中其他关联部分,此处不再赘述。
步骤S5207,收到上述第一CSI反馈网络中第一模型的模型信息,第二通信设备102激活该模型信息对应的第一模型。
步骤S5207的可选实现方式可以参见图3的步骤S3112的可选实现方式、及图3所涉及的实施例中其他关联部分,此处不再赘述。
步骤S5208,收到上述第一CSI反馈网络中第一模型的模型参数,或者收到上述第一CSI反馈网络中第一模型的模型参数和模型信息,第二通信设备102将该第一模型的模型参数进行去量化处理,并设定该第一模型的模型编号为该模型信息中的模型编号,激活第一模型。
步骤S5208的可选实现方式可以参见图3的步骤S3113的可选实现方式、及图3所涉及的实施例中其他关联部分,此处不再赘述。
步骤S5209,发送确认信息。
在一些实施例中,上述确认信息包括用于指示第二通信设备102完成模型初始化的确认指示信息。示例性的,上述确认指示信息可以用Indicatorcomplete表示。
步骤S5209的可选实现方式可以参见图3的步骤S3114的可选实现方式、及图3所涉及的实施例中其他关联部分,此处不再赘述。
本公开实施例所涉及的方法可以包括步骤S5201~步骤S5209中的至少一者。例如,步骤S5202+步骤S5205+步骤S5206+步骤S5207可以作为独立实施例来实施,步骤S5202+步骤S5205+步骤S5206+步骤S5208可以作为独立实施例来实施,步骤S5202+步骤S5205+步骤S5206+步骤S5207+步骤S5209可以作为独立实施例来实施,步骤S5202+步骤S5205+步骤S5206+步骤S5208+步骤S5209可以作为独立实施例来实施,步骤S5201+步骤S5202+步骤S5205+步骤S5206+步骤S5207可以作为独立实施例来实施,步骤S5201+步骤S5202+步骤S5205+步骤S5206+步骤S5208可以作为独立实施例来实施,步骤S5201+步骤S5202+步骤S5205+步骤S5206+步骤S5207+步骤S5209可以作为独立实施例来实施,步骤S5201+步骤S5202+步骤S5205+步骤S5206+步骤S5208+步骤S5209可以作为独立实施例来实施,步骤S5201+步骤S5202+步骤S5203+步骤S5204+步骤S5205+步骤S5206+步骤S5207+步骤S5208+步骤S5209可以作为独立实施例来实施,步骤S5201+步骤S5202+步骤S5203+步骤S5204+步骤S5205+步骤S5206+步骤S5207可以作为独立实施例来实施,步骤S5201+步骤S5202+步骤S5203+步骤S5204+步骤S5205+步骤S5206+步骤S5208可以作为独立实施例来实施,步骤S5201+步骤S5202+步骤S5203+步骤S5204+步骤S5205+步骤S5206+步骤S5207+步骤S5209可以作为独立实施例来实施,步骤S5201+步骤S5202+步骤S5203+步骤S5204+步骤S5205+步骤S5206+步骤S5208+步骤S5209可以作为独立实施例来实施,但不限于此。
在一些实施例中,步骤S5201、步骤S5202、步骤S5203可以交换顺序或同时执行,步骤S5207、步骤S5208可以交换顺序或同时执行。
在一些实施例中,步骤S5201、步骤S5203、步骤S5204、步骤S5208、步骤S5209是可选地,在不同实施例中可以对这些步骤中的一个或多个步骤进行省略或替代。
在一些实施例中,步骤S5201、步骤S5203、步骤S5204、步骤S5207、步骤S5209是可选地,在不同实施例中可以对这些步骤中的一个或多个步骤进行省略或替代。
在一些实施例中,步骤S5203、步骤S5204、步骤S5208、步骤S5209是可选地,在不同实施例中可以对这些步骤中的一个或多个步骤进行省略或替代。
在一些实施例中,步骤S5203、步骤S5204、步骤S5207、步骤S5209是可选地,在不同实施例中可以对这些步骤中的一个或多个步骤进行省略或替代。
在一些实施例中,步骤S5201、步骤S5203、步骤S5204、步骤S5209是可选地,在不同实施例中可以对这些步骤中的一个或多个步骤进行省略或替代。
在一些实施例中,步骤S5201、步骤S5203、步骤S5204是可选地,在不同实施例中可以对这些步骤中的一个或多个步骤进行省略或替代。
在一些实施例中,步骤S5201是可选地,在不同实施例中可以对这些步骤中的一个或多个步骤进行省略或替代。
在一些实施例中,步骤S5203、步骤S5204是可选地,在不同实施例中可以对这些步骤中的一个或多个步骤进行省略或替代。
图5C是根据本公开实施例示出的模型初始化方法的流程示意图。如图5C所示,本公开实施例涉及模型初始化方法可由第二通信设备102执行,上述方法可以包括但不限于如下步骤。
步骤S5301,发送第一指示信息。
在一些实施例中,上述第一指示信息用于指示第二通信设备102配置了第一模型,且具备使用第一模型完成CSI反馈工作的能力。在一些实施例中,上述第一模型为用于CSI反馈的模型,作为一种示例,第一模型可以是CSI反馈网络中的编码器模型,或者也可以是CSI反馈网络中的译码器模型。
步骤S5301的可选实现方式可以参见图3的步骤S3101的可选实现方式、及图3所涉及的实施例中其他关联部分,此处不再赘述。
步骤S5302,发送第一信息。
在一些实施例中,上述第一信息包括由第二通信设备102配置的第一模型的模型信息。
步骤S5302的可选实现方式可以参见图3的步骤S3102的可选实现方式、及图3所涉及的实施例中其他关联部分,此处不再赘述。
步骤S5303,接收第二指示信息。
在一些实施例中,上述第二指示信息用于指示第一通信设备101通知第二通信设备102进入模型初始化流程。
步骤S5303的可选实现方式可以参见图3的步骤S3103的可选实现方式、及图3所涉及的实施例中其他关联部分,此处不再赘述。
步骤S5304,发送第三指示信息。
在一些实施例中,上述第三指示信息用于指示第二通信设备102接受了模型初始化指令,进入模型初始化流程。可选的,第一通信设备101接收到第二通信设备102发送的上述第三指示信息,则可以说明第二通信设备102收到了第一通信设备101发送的上述第二指示信息,且接受了对第二通信设备上CSI反馈网络模型的初始化指令,进入模型初始化流程。
步骤S5304的可选实现方式可以参见图3的步骤S3104的可选实现方式、及图3所涉及的实施例中其他关联部分,此处不再赘述。
步骤S5305,接收第二信息。
在一些实施例中,上述第二信息可以为待部署在所述第二通信设备中的所述第一模型的模型部署信息。
在一些实施例中,上述第二信息可以承载在模型部署指示信息中发送。
在一些实施例中,第二通信设备102接收上述模型部署指示信息。可选的,第二通信设备102接收第一通信设备101发送的上述模型部署指示信息。
在一些实施例中,上述模型部署指示信息包括第二信息,该第二信息为为待部署在第二通信设备中的第一模型的模型部署信息。其的可选实现方式可以参见图3的步骤S3111的可选实现方式、及图3所涉及的实施例中其他关联部分,此处不再赘述。
步骤S5306,收到第一CSI反馈网络中第一模型的模型信息,第二通信设备102激活该模型信息对应的第一模型。
在一些实施例中,上述第一CSI反馈网络的实现方式可以参见图3的步骤S3106-步骤S3109的可选实现方式、及图3所涉及的实施例中其他关联部分,此处不再赘述。
步骤S5306的可选实现方式可以参见图3的步骤S3112的可选实现方式、及图3所涉及的实施例中其他关联部分,此处不再赘述。
步骤S5307,收到第一CSI反馈网络中第一模型的模型参数,或者收到上述第一CSI反馈网络中第一模型的模型参数和模型信息,第二通信设备102将该第一模型的模型参数进行去量化处理,并设定该第一模型的模型编号为该模型信息中的模型编号,激活第一模型。
步骤S5307的可选实现方式可以参见图3的步骤S3113的可选实现方式、及图3所涉及的实施例中其他关联部分,此处不再赘述。
步骤S5308,发送确认信息。
在一些实施例中,上述确认信息包括用于指示第二通信设备102完成模型初始化的确认指示信息。示例性的,上述确认指示信息可以用Indicatorcomplete表示。
步骤S5308的可选实现方式可以参见图3的步骤S3114的可选实现方式、及图3所涉及的实施例中其他关联部分,此处不再赘述。
本公开实施例所涉及的方法可以包括步骤S5301~步骤S5308中的至少一者。例如,步骤S5302+步骤S5305+步骤S5306可以作为独立实施例来实施,步骤S5302+步骤S5305+步骤S5307可以作为独立实施例来实施,步骤S5301+步骤S5302+步骤S5305+步骤S5306可以作为独立实施例来实施,步骤S5301+步骤S5302+步骤S5305+步骤S5307可以作为独立实施例来实施,步骤S5302+步骤S5305+步骤S5306+步骤S5308可以作为独立实施例来实施,步骤S5302+步骤S5305+步骤S5307+步骤S5308可以作为独立实施例来实施,步骤S5301+步骤S5302+步骤S5305+步骤S5306+步骤S5308可以作为独立实施例来实施,步骤S5301+步骤S5302+步骤S5305+步骤S5307+步骤S5308可以作为独立实施例来实施,步骤S5302+步骤S5303+步骤S5304+步骤S5305+步骤S5306可以作为独立实施例来实施,步骤S5302+步骤S5303+步骤S5304+步骤S5305+步骤S5307可以作为独立实施例来实施,步骤S5301+步骤S5302+步骤S5303+步骤S5304+步骤S5305+步骤S5306可以作为独立实施例来实施,步骤S5301+步骤S5302+步骤S5303+步骤S5304+步骤S5305+步骤S5307可以作为独立实施例来实施,步骤S5301+步骤S5302+步骤S5303+步骤S5304+步骤S5305+步骤S5306+步骤S5308可以作为独立实施例来实施,步骤S5301+步骤S5302+步骤S5303+步骤S5304+步骤S5305+步骤S5307+步骤S5308可以作为独立实施例来实施,但不限于此。
在一些实施例中,步骤S5301、步骤S5302、步骤S5303可以交换顺序或同时执行,步骤S5312、步骤S5313可以交换顺序或同时执行。
在一些实施例中,步骤S5301、步骤S5303、步骤S5304、步骤S5307、步骤S5308是可选地,在不同实施例中可以对这些步骤中的一个或多个步骤进行省略或替代。
在一些实施例中,步骤S5301、步骤S5303、步骤S5304、步骤S5306、步骤S5308是可选地,在不同实施例中可以对这些步骤中的一个或多个步骤进行省略或替代。
在一些实施例中,步骤S5303、步骤S5304、步骤S5307、步骤S5308是可选地,在不同实施例中可以对这些步骤中的一个或多个步骤进行省略或替代。
在一些实施例中,步骤S5303、步骤S5304、步骤S5306、步骤S5308是可选地,在不同实施例中可以对这些步骤中的一个或多个步骤进行省略或替代。
在一些实施例中,步骤S5301、步骤S5303、步骤S5304、步骤S5306是可选地,在不同实施例中可以对这些步骤中的一个或多个步骤进行省略或替代。
在一些实施例中,步骤S5301、步骤S5303、步骤S5304、步骤S5307是可选地,在不同实施例中可以对这些步骤中的一个或多个步骤进行省略或替代。
图6是根据本公开实施例示出的模型初始化方法的交互示意图。如图6所示,本公开实施例涉及的方法可应用于通信系统100,上述方法包括但不限于如下步骤。
步骤S6101,第二通信设备102向第一通信设备101发送第一信息,第一信息包括由第二通信设备配置的第一模型的模型信息,第一模型为用于信道状态信息CSI反馈的模型。
步骤S6101的可选实现方式可以参见图3的步骤S3102的可选实现方式、及图3所涉及的实施例中其他关联部分,此处不再赘述。
步骤S6102,第一通信设备101基于第一信息,确定第二信息。
在一些实施例中,上述第二信息可以为待部署在第二通信设备102中的第一模型的模型部署信息。
在一些实施例中,第一通信设备101确定CSI实时数据;基于所述CSI实时数据和所述第一信息,确定第二信息。其的可选实现方式可以参见图3的步骤S3105、步骤S3106、步骤S3107、步骤S3108、步骤S3109、步骤S3110的可选实现方式、及图3所涉及的实施例中其他关联部分,此处不再赘述。
步骤S6103,第一通信设备101向第二通信设备102发送第二信息,以使第二通信设备102基于第二信息进行模型部署。
在一些实施例中,上述第二信息可以为待部署在所述第二通信设备中的所述第一模型的模型部署信息。
在一些实施例中,上述第二信息可以承载在模型部署指示信息中发送。
步骤S6103的可选实现方式可以参见图3的步骤S3111的可选实现方式、及图3所涉及的实施例中其他关联部分,此处不再赘述。
步骤S6104,第二通信设备102基于上述第二信息进行模型部署。
在一些实施例中,确定第二信息为第一CSI反馈网络中第一模型的模型信息,在第二通信设备上激活该模型信息所对应的第一模型,其中,第一CSI反馈网络是由第一通信设备基于CSI实时数据从第一通信设备所部署的CSI反馈网络中选择的。其的可选实现方式可以参见图3的步骤S3112的可选实现方式、及图3所涉及的实施例中其他关联部分,此处不再赘述。
在一些实施例中,确定第二信息包括经量化处理的模型参数,将模型参数进行去量化处理,并将去量化处理的模型参数部署在第二通信设备上,激活第一模型。其的可选实现方式可以参见图3的步骤S3113的可选实现方式、及图3所涉及的实施例中其他关联部分,此处不再赘述。
在一些实施例中,确定第二信息为第一CSI反馈网络中第一模型的模型信息,在第二通信设备上激活该模型信息所对应的第一模型,其中,第一CSI反馈网络是由第一通信设备基于CSI实时数据从第一通信设备所部署的CSI反馈网络中选择的。确定第二信息包括经量化处理的模型参数,将模型参数进行去量化处理,并将去量化处理的模型参数部署在第二通信设备上,激活第一模型。
在一些实施例中,上述方法可以包括上述与第一通信设备、第二通信设备等的实施例所述的方法,此处不再赘述。
图7是根据本公开实施例示出的模型初始化方法的示意图。如图7所示,本公开实施例的方法可应用于网络设备。上述方法包括:
步骤S7101,接收终端所上报的AI/ML模型应用请求指示Indicatorinit_request、用于CSI反馈的AI/ML模型类型、参数等信息。所述指示是指终端配置了AI/ML模型,且当前具备使用AI/ML模型完成CSI反馈工作的能力;所述信息包括:终端配置的CSI反馈网络模型的编号Model1,Model2,…,Modeln、模型结构等信息。
步骤S7102,向终端发送AI/ML模型初始化指示Indicatorinitialize、CSI实时数据请求指示Indicatordata,所述AI/ML模型初始化指示表示网络设备通知终端进入模型初始化流程,所述CSI实时数据请求指示表示网络设备向终端请求实时CSI数据以完成下一步的模型选择,所述实时数据指:终端在当前时刻通过信道估计所获取的下行CSI,能够在一定程度上代表当前信道环境下的CSI数据特征。
步骤S7103,接收终端发送的AI/ML模型初始化接受指示Indicatorinit_accept、实时的经量化处理的CSI实时数据。将接收到的数据进行去量化处理。
在一些实施例中,进行模型选择操作(所述操作为网络设备内容行为)的实现方式可如下:将CSI实时数据输入网络设备所部署的AI/ML模型,获取经AI/ML模型压缩、重建处理后的CSI输出值,计算CSI输出值与原CSI实时数据之间的误差,进而获取AI/ML模型实时反馈精确度。选择反馈精确度最高的CSI反馈网络模型并获取其模型编号Modelopt。
在一些实施例中,根据终端上报的终端配置的CSI反馈网络模型的编号Model1,Model2,…,Modeln,若终端配置的模型编号中包含Modelopt,网络设备向终端发送模型部署指示Indicatordeploy,所述指示表示终端可以进行模型部署,该指示携带模型编号Modelopt。
在一些实施例中,若终端配置的模型编号中不包含Modelopt,网络设备将Modelopt所对应的编码器模型参数进行量化处理,向终端发送模型部署指示Indicatordeploy、量化后的编码器模型参数和模型编号Modelopt,所述指示表示终端可以进行模型部署。
步骤S7104,接收终端发送的模型初始化确认信息,所述确认信息主要包括:终端已经完成模型初始化的确认指示Indicatorcomplete。
步骤S7105,使用网络设备所选择的模型编号Modelopt所对应的译码器模型进行CSI接收。
图8是根据本公开实施例示出的模型初始化方法的示意图。如图8所示,本公开实施例的方法可应用于终端。上述方法包括:
步骤S8101,向网络设备上报AI/ML模型应用请求指示Indicatorinit_request、用于CSI反馈的AI/ML模型类型、参数等信息。所述指示是指终端配置了AI/ML模型,且当前具备使用AI/ML模型完成CSI反馈工作的能力;所述信息包括:终端配置的CSI反馈网络模型的编号Model1,Model2,…,Modeln、模型结构等信息。
步骤S8102,接收网络设备发送的AI/ML模型初始化指示Indicatorinitialize、CSI实时数据请求指示Indicatordata。
步骤S8103,通过下行信道估计获取一定数量的实时CSI样本数据,进行CSI数据量化处理(属于终端内部行为)。
步骤S8104,向网络设备上报AI/ML模型初始化接受指示Indicatorinit_accept、实时的经量化处理的CSI样本数据。
步骤S8105,接收网络设备发送的模型部署指示Indicatordeploy、模型编号Modelopt或量化后的编码器模型参数。
在一些实施例中,若接收到模型编号Modelopt,激活Modelopt所对应的编码器模型;
在一些实施例中,若接收到编码器模型参数和模型编号Modelopt,将编码器模型参数进行去量化处理,同时设定其模型编号为Modelopt。
需要说明的是,在一些实施例中,终端接收到模型编号Modelopt后,激活第一模型,网络设备在完成模型选择后,也需要将部署于网络设备上与该第一模型(如编码器模型)关联的第二模型(如译码器模型)进行激活。在一些实施例中,终端接收到模型编号Modelopt和编码器模型参数,并将编码器模型参数进行去量化处理,设定模型编号之后,也应将该编码器模型进行激活,网络设备在完成模型选择后,也需要将部署于网络设备上与该编码器模型关联的译码器模型进行激活。由此,可以通过激活的编码器和译码器实现CSI反馈。
步骤S8106,向网络设备发送模型初始化确认信息,所述确认信息主要包括:终端已经完成模型初始化的确认指示Indicatorcomplete。
步骤S8107,使用网络设备所选择的模型编号Modelopt所对应的编码器模型进行CSI发送。
本公开实施例还提出用于实现以上任一方法的装置,例如,提出一装置,上述装置包括用以实现以上任一方法中第一通信设备所执行的各步骤的单元或模块。再如,还提出另一装置,包括用以实现以上任一方法中第二通信设备所执行的各步骤的单元或模块。
应理解以上装置中各单元或模块的划分仅是一种逻辑功能的划分,在实际实现时可以全部或部分集成到一个物理实体上,也可以物理上分开。此外,装置中的单元或模块可以以处理器调用软件的形式实现:例如装置包括处理器,处理器与存储器连接,存储器中存储有指令,处理器调用存储器中存储的指令,以实现以上任一方法或实现上述装置各单元或模块的功能,其中处理器例如为通用处理器,例如中央处理单元(Central Processing Unit,CPU)或微处理器,存储器为装置内的存储器或装置外的存储器。或者,装置中的单元或模块可以以硬件电路的形式实现,可以通过对硬件电路的设计实现部分或全部单元或模块的功能,上述硬件电路可以理解为一个或多个处理器;例如,在一种实现中,上述硬件电路为专用集成电路(application-specific integrated circuit,ASIC),通过对电路内元件逻辑关系的设计,实现以上部分或全部单元或模块的功能;再如,在另一种实现中,上述硬件电路为可以通过可编程逻辑器件(programmable logic device,PLD)实现,以现场可编程门阵列(Field Programmable Gate Array,FPGA)为例,其可以包括大量逻辑门电路,通过配置文件来配置逻辑门电路之间的连接关系,从而实现以上部分或全部单元或模块的功能。以上装置的所有单元或模块可以全部通过处理器调用软件的形式实现,或全部通过硬件电路的形式实现,或部分通过处理器调用软件的形式实现,剩余部分通过硬件电路的形式实现。
在本公开实施例中,处理器是具有信息处理能力的电路,在一种实现中,处理器可以是具有指令读取与运行能力的电路,例如中央处理单元(Central Processing Unit,CPU)、微处理器、图形处理器(graphics processing unit,GPU)(可以理解为微处理器)、或数字信息处理器(digital signal processor,DSP)等;在另一种实现中,处理器可以通过硬件电路的逻辑关系实现一定功能,上述硬件电路的逻辑关系是固定的或可以重构的,例如处理器为专用集成电路(application-specific integrated circuit,ASIC)或可编程逻辑器件(programmable logic device,PLD)实现的硬件电路,例如FPGA。在可重构的硬件电路中,处理器加载配置文档,实现硬件电路配置的过程,可以理解为处理器加载指令,以实现以上部分或全部单元或模块的功能的过程。此外,还可以是针对人工智能设计的硬件电路,其可以理解为ASIC,例如神经网络处理单元(Neural Network Processing Unit,NPU)、张量处理单元(Tensor Processing Unit,TPU)、深度学习处理单元(Deep learning Processing Unit,DPU)等。
图9A是本公开实施例提出的第一终端的结构示意图。如图9A所示,第一通信设备9100可以包括:收发模块9101、处理模块9102等中的至少一者。在一些实施例中,上述收发模块,收发模块,用于接收第二通信设备发送的第一信息,所述第一信息包括由所述第二通信设备配置的第一模型的模型信息,所述第一模型为用于信道状态信息CSI反馈的模型;处理模块,用于基于所述第一信息,确定第二信息;所述收发模块,还用于向所述第二通信设备发送模型部署指示信息,所述模型部署指示信息包括所述第二信息,所述模型部署指示信息用于指示所述第二通信设备基于所述第二信息进行模型部署。可选地,上述收发模块用于执行以上任一方法中第一通信设备101执行的发送和/或接收等通信步骤(例如步骤S3103、步骤S3111,但不限于此)中的至少一者,此处不再赘述。可选地,上述处理模块用于执行以上任一方法中第二通信设备101执行的其他步骤(例如步骤S3105、步骤S3106、步骤S3107、步骤S3108、步骤S3109、步骤S3110、步骤S3203,但不限于此)中的至少一者,此处不再赘述。
图9B是本公开实施例提出的第二通信设备的结构示意图。如图9B所示,第二通信设备9200可以包括:收发模块9201、处理模块9202等中的至少一者。在一些实施例中,上述收发模块,用于向第一通信设备发送第一信息,所述第一信息包括由所述第二通信设备配置的第一模型的模型信息,所述第一模型为用于信道状态信息CSI反馈的模型;所述收发模块,还用于接收所述第一通信设备发送的第二信息,所述第二信息为所述第一通信设备基于所述第一信息确定的,所述第二信息为待部署在所述第二通信设备中的所述第一模型的模型部署信息;上述处理模块,用于基于所述第二信息进行模型部署。可选地,上述收发模块用于执行以上任一方法中第二通信设备102执行的发送和/或接收等通信步骤(例如步骤S3101、步骤S3102、步骤S3104、步骤S3114,但不限于此)中的至少一者,此处不再赘述。可选地,上述处理模块用于执行以上任一方法中第二通信设备102执行的其他步骤(如步骤S3112、步骤S3113,但不限于此)中的至少一者,此处不再赘述。
在一些实施例中,收发模块可以包括发送模块和/或接收模块,发送模块和接收模块可以是分离的,也可以集成在一起。可选地,收发模块可以与收发器相互替换。
在一些实施例中,处理模块可以是一个模块,也可以包括多个子模块。可选地,上述多个子模块分别执行处理模块所需执行的全部或部分步骤。可选地,处理模块可以与处理器相互替换。
图10A是本公开实施例提出的通信设备1100的结构示意图。通信设备1100可以是上述第一通信设备,也可以是上述第二通信设备,也可以是支持第一通信设备实现以上任一方法的芯片、芯片系统、或处理器等,还可以是支持第二通信设备实现以上任一方法的芯片、芯片系统、或处理器等。通信设备1100可用于实现上述方法实施例中描述的方法,具体可以参见上述方法实施例中的说明。
如图10A所示,通信设备1100包括一个或多个处理器1101。处理器1101可以是通用处理器或者专用处理器等,例如可以是基带处理器或中央处理器。基带处理器可以用于对通信协议以及通信数据进行处理,中央处理器可以用于对通信装置(如,基站、基带芯片,终端设备、终端设备芯片,DU或CU等)进行控制,执行程序,处理程序的数据。通信设备1100用于执行以上任一方法。
在一些实施例中,通信设备1100还包括用于存储指令的一个或多个存储器1102。可选地,全部或部分存储器1102也可以处于通信设备1100之外。
在一些实施例中,通信设备1100还包括一个或多个收发器1103。在通信设备1100包括一个或多个收发器1103时,收发器1103执行上述方法中的发送和/或接收等通信步骤(例如步骤S3103、步骤S3111、步骤S3101、步骤S3102、步骤S3104、步骤S3114,但不限于此)中的至少一者,处理器1101执行其他步骤(例如步骤S3105、步骤S3106、步骤S3107、步骤S3108、步骤S3109、步骤S3110、步骤S3203、步骤S3112、步骤S3113,但不限于此)中的至少一者。
在一些实施例中,收发器可以包括接收器和/或发送器,接收器和发送器可以是分离的,也可以集成在一起。可选地,收发器、收发单元、收发机、收发电路等术语可以相互替换,发送器、发送单元、发送机、发送电路等术语可以相互替换,接收器、接收单元、接收机、接收电路等术语可以相互替换。
在一些实施例中,通信设备1100可以包括一个或多个接口电路1104。可选地,接口电路1104与存储器1102连接,接口电路1104可用于从存储器1102或其他装置接收信号,可用于向存储器1102或其他装置发送信号。例如,接口电路1104可读取存储器1102中存储的指令,并将该指令发送给处理器1101。
以上实施例描述中的通信设备1100可以是上述第一通信设备或者上述第二通信设备,但本公开中描述的通信设备1100的范围并不限于此,通信设备1100的结构可以不受图10A的限制。通信设备可以是独立的设备或者可以是较大设备的一部分。例如所述通信设备可以是:1)独立的集成电路IC,或芯片,或,芯片系统或子系统;(2)具有一个或多个IC的集合,可选地,上述IC集合也可以包括用于存储数据,程序的存储部件;(3)ASIC,例如调制解调器(Modem);(4)可嵌入在其他设备内的模块;(5)接收机、终端设备、智能终端设备、蜂窝电话、无线设备、手持机、移动单元、车载设备、网络设备、云设备、人工智能设备等等;(6)其他等等。
图10B是本公开实施例提出的芯片1200的结构示意图。对于通信设备1100可以是芯片或芯片系统的情况,可以参见图10B所示的芯片1200的结构示意图,但不限于此。
芯片1200包括一个或多个处理器1201,芯片1200用于执行以上任一方法。
在一些实施例中,芯片1200还包括一个或多个接口电路1202。可选地,接口电路1202与存储器1203连接,接口电路1202可以用于从存储器1203或其他装置接收信号,接口电路1202可用于向存储器1203或其他装置发送信号。例如,接口电路1202可读取存储器1203中存储的指令,并将该指令发送给处理器1201。
在一些实施例中,接口电路1202执行上述方法中的发送和/或接收等通信步骤(例如步骤S3103、步骤S3111、步骤S3101、步骤S3102、步骤S3104、步骤S3114,但不限于此)中的至少一者,处理器1201执行其他步骤(例如步骤S3105、步骤S3106、步骤S3107、步骤S3108、步骤S3109、步骤S3110、步骤S3203、步骤S3112、步骤S3113,但不限于此)中的至少一者。
在一些实施例中,接口电路、接口、收发管脚、收发器等术语可以相互替换。
在一些实施例中,芯片1200还包括用于存储指令的一个或多个存储器1203。可选地,全部或部分存储器1203可以处于芯片1200之外。
本公开还提出存储介质,上述存储介质上存储有指令,当上述指令在通信设备1100上运行时,使得通信设备1100执行以上任一方法。可选地,上述存储介质是电子存储介质。可选地,上述存储介质是计算机可读存储介质,但不限于此,其也可以是其他装置可读的存储介质。可选地,上述存储介质可以是非暂时性(non-transitory)存储介质,但不限于此,其也可以是暂时性存储介质。
本公开还提出程序产品,上述程序产品被通信设备1100执行时,使得通信设备1100执行以上任一方法。可选地,上述程序产品是计算机程序产品。
本公开还提出计算机程序,当其在计算机上运行时,使得计算机执行以上任一方法。
在上述实施例中,可以全部或部分地通过软件、硬件、固件或者其任意组合来实现。当使用软件实现时,可以全部或部分地以计算机程序产品的形式实现。所述计算机程序产品包括一个或多个计算机程序。在计算机上加载和执行所述计算机程序时,全部或部分地产生按照本公开实施例所述的流程或功能。所述计算机可以是通用计算机、专用计算机、计算机网络、或者其他可编程装置。所述计算机程序可以存储在计算机可读存储介质中,或者从一个计算机可读存储介质向另一个计算机可读存储介质传输,例如,所述计算机程序可以从一个网站站点、计算机、服务器或数据中心通过有线(例如同轴电缆、光纤、数字用户线(digital subscriber line,DSL))或无线(例如红外、无线、微波等)方式向另一个网站站点、计算机、服务器或数据中心进行传输。所述计算机可读存储介质可以是计算机能够存取的任何可用介质或者是包含一个或多个可用介质集成的服务器、数据中心等数据存储设备。所述可用介质可以是磁性介质(例如,软盘、硬盘、磁带)、光介质(例如,高密度数字视频光盘(digital video disc,DVD))、或者半导体介质(例如,固态硬盘(solid state disk,SSD))等。
本领域普通技术人员可以意识到,结合本文中所公开的实施例描述的各示例的单元及算法步骤,能够以电子硬件、或者计算机软件和电子硬件的结合来实现。这些功能究竟以硬件还是软件方式来执行,取决于技术方案的特定应用和设计约束条件。专业技术人员可以对每个特定的应用来使用不同方法来实现所描述的功能,但是这种实现不应认为超出本公开的范围。
所属领域的技术人员可以清楚地了解到,为描述的方便和简洁,上述描述的系统、装置和单元的具体工作过程,可以参考前述方法实施例中的对应过程,在此不再赘述。
以上所述,仅为本公开的具体实施方式,但本公开的保护范围并不局限于此,任何熟悉本技术领域的技术人员在本公开揭露的技术范围内,可轻易想到变化或替换,都应涵盖在本公开的保护范围之内。因此,本公开的保护范围应以所述权利要求的保护范围为准。
Claims (29)
- 一种模型初始化方法,其特征在于,包括:第一通信设备接收第二通信设备发送的第一信息,所述第一信息包括由所述第二通信设备配置的第一模型的模型信息,所述第一模型为用于信道状态信息CSI反馈的模型;基于所述第一信息,确定第二信息,其中,所述第二信息为待部署在所述第二通信设备中的所述第一模型的模型部署信息;向所述第二通信设备发送所述第二信息,以使所述第二通信设备基于所述第二信息进行模型部署。
- 如权利要求1所述的方法,其特征在于,所述基于所述第一信息,确定第二信息,包括:确定CSI实时数据;基于所述CSI实时数据和所述第一信息,确定所述第二信息。
- 如权利要求2所述的方法,其特征在于,所述第一通信设备为网络设备,所述第二通信设备为终端,所述第一模型为CSI反馈网络中的编码器模型。
- 如权利要求3所述的方法,其特征在于,所述确定CSI实时数据,包括:向所述终端发送导频信号和数据请求指示信息,所述导频信号用于所述终端在当前时刻进行信道估计以得到CSI实时数据,所述数据请求指示信息用于指示所述网络设备向所述终端请求所述CSI实时数据;接收所述终端发送的经量化处理的CSI实时数据;将接收到的所述经量化处理的CSI实时数据进行去量化处理,得到所述CSI实时数据。
- 如权利要求2所述的方法,其特征在于,所述第一通信设备为终端,所述第二通信设备为网络设备,所述第一模型为CSI反馈网络中的译码器模型。
- 如权利要求5所述的方法,其特征在于,所述确定CSI实时数据,包括:接收所述网络设备发送的导频信号;在当前时刻基于所述导频信号进行信道估计,得到所述CSI实时数据。
- 如权利要求2所述的方法,其特征在于,所述基于所述CSI实时数据和所述第一信息,确定所述第二信息,包括:将所述CSI实时数据输入所述第一通信设备所部署的CSI反馈网络,其中,所述CSI反馈网络包括所述第一模型和第二模型,所述CSI反馈网络通过所述第一模型和所述第二模型对所述CSI实时数据进行压缩和重建处理;获取所述CSI反馈网络输出的CSI输出值;根据所述CSI输出值与所述CSI实时数据,确定所述CSI反馈网络的反馈精确度;根据所述反馈精确度,从所述第一通信设备所部署的CSI反馈网络中选择第一CSI反馈网络;基于所述第一CSI反馈网络中第一模型的模型信息和所述第二通信设备配置的第一模型的模型信息,确定所述第二信息。
- 如权利要求7所述的方法,其特征在于,所述第一通信设备所部署的CSI反馈网络为多个;所述根据所述反馈精确度,从所述第一通信设备所部署的CSI反馈网络中选择第一CSI反馈网络,包括:从多个所述CSI反馈网络中,选择所述反馈精确度最高的CSI反馈网络确定为所述第一CSI反馈网络。
- 如权利要求7或8所述的方法,其特征在于,所述基于所述第一CSI反馈网络中第一模型的模型信息和所述第二通信设备配置的第一模型的模型信息,确定所述第二信息,包括以下任一项:将所述第一CSI反馈网络中第一模型的模型信息确定为所述第二信息,其中,所述第二通信设备配置的第一模型的模型信息中包含所述第一CSI反馈网络中第一模型的模型信息;将所述第一CSI反馈网络中第一模型的模型参数进行量化处理,并将经量化处理的所述模型参数确定为所述第二信息,其中,所述第二通信设备配置的第一模型的模型信息中不包含所述第一CSI反馈网络中第一模型的模型信息。
- 如权利要求9所述的方法,其特征在于,所述将经量化处理的所述模型参数确定为所述第二信息,包括:将所述第一CSI反馈网络中第一模型的模型信息和经量化处理的所述模型参数,确定为所述第二信息。
- 如权利要求1-10中任一项所述的方法,其特征在于,所述方法还包括:接收所述第二通信设备发送的第一指示信息,所述第一指示信息用于指示所述第二通信设备配置了所述第一模型,且具备使用所述第一模型完成CSI反馈工作的能力。
- 如权利要求1-11中任一项所述的方法,其特征在于,所述方法还包括:向所述第二通信设备发送第二指示信息,所述第二指示信息用于指示所述第一通信设备通知所述第二通信设备进入模型初始化流程;接收所述第二通信设备发送的第三指示信息,所述第三指示信息用于指示所述第二通信设备接受了模型初始化指令,进入模型初始化流程。
- 如权利要求1-12中任一项所述的方法,其特征在于,所述方法还包括:接收所述第二通信设备发送的确认信息,所述确认信息包括用于指示所述第二通信设备完成模型初始化的确认指示信息。
- 如权利要求1-13中任一项所述的方法,其特征在于,所述模型信息包括模型编号和/或模型结构。
- 一种模型初始化方法,其特征在于,包括:第二通信设备向第一通信设备发送第一信息,所述第一信息包括由所述第二通信设备配置的第一模型的模型信息,所述第一模型为用于信道状态信息CSI反馈的模型;接收所述第一通信设备发送的第二信息,所述第二信息为所述第一通信设备基于所述第一信息确定的,所述第二信息为待部署在所述第二通信设备中的所述第一模型的模型部署信息;基于所述第二信息进行模型部署。
- 如权利要求15所述的方法,其特征在于,所述第一通信设备为网络设备,所述第二通信设备为终端,所述第一模型为CSI反馈网络中的编码器模型。
- 如权利要求16所述的方法,其特征在于,所述方法还包括:接收所述网络设备发送的导频信号和数据请求指示信息,所述导频信号用于所述终端在当前时刻进行信道估计以得到CSI实时数据,所述数据请求指示信息用于指示所述网络设备向所述终端请求所述CSI实时数据;基于所述导频信号进行信道估计,得到所述CSI实时数据;将所述CSI实时数据进行量化处理;将经量化处理的所述CSI实时数据发送给所述网络设备,其中,所述CSI实时数据用于所述网络设备结合所述第一信息确定所述第二信息。
- 如权利要求15所述的方法,其特征在于,所述第一通信设备为终端,所述第二通信设备为网络设备,所述第一模型为CSI反馈网络中的译码器模型。
- 如权利要求18所述的方法,其特征在于,所述方法还包括:向所述终端发送导频信号,其中,所述导频信号用于所述终端在当前时刻进行信道估计以得到CSI实时数据,所述CSI实时数据用于所述终端结合所述第一信息确定所述第二信息。
- 如权利要求15-19中任一项所述的方法,其特征在于,所述基于所述第二信息进行模型部署,包括以下任一项:确定所述第二信息为第一CSI反馈网络中第一模型的模型信息,在所述第二通信设备上激活所述模型信息所对应的第一模型,其中,所述第一CSI反馈网络是由所述第一通信设备基于CSI实时数据从所述第一通信设备所部署的CSI反馈网络中选择的;确定所述第二信息包括经量化处理的所述模型参数,将所述模型参数进行去量化处理,并将去量化处理的所述模型参数部署在所述第二通信设备上,激活所述部署的模型。
- 如权利要求15-20中任一项所述的方法,其特征在于,所述方法还包括:向所述第一通信设备发送第一指示信息,所述第一指示信息用于指示所述第二通信设备配置了所述第一模型,且具备使用所述第一模型完成CSI反馈工作的能力。
- 如权利要求15-21中任一项所述的方法,其特征在于,所述方法还包括:接收所述第一通信设备发送的第二指示信息,所述第二指示信息用于指示所述第一通信设备通知所述第二通信设备进入模型初始化流程;向所述第一通信设备发送第三指示信息,所述第三指示信息用于指示所述第二通信设备接受了模型初始化指令,进入模型初始化流程。
- 如权利要求15-22中任一项所述的方法,其特征在于,所述方法还包括:向所述第一通信设备发送确认信息,所述确认信息包括用于指示所述第二通信设备完成模型初始化的确认指示信息。
- 如权利要求15-23中任一项所述的方法,其特征在于,所述模型信息包括模型编号和/或模型结构。
- 一种第一通信装置,其特征在于,包括:收发模块,用于接收第二通信设备发送的第一信息,所述第一信息包括由所述第二通信设备配置的第一模型的模型信息,所述第一模型为用于信道状态信息CSI反馈的模型;处理模块,用于基于所述第一信息,确定第二信息,其中,所述第二信息为待部署在所述第二通信设备中的所述第一模型的模型部署信息;所述收发模块,还用于向所述第二通信设备发送所述第二信息,以使所述第二通信设备基于所述第二信息进行模型部署。
- 一种第二通信装置,其特征在于,包括:收发模块,用于向第一通信设备发送第一信息,所述第一信息包括由所述第二通信设备配置的第一模型的模型信息,所述第一模型为用于信道状态信息CSI反馈的模型;所述收发模块,还用于接收所述第一通信设备发送的第二信息,所述第二信息为所述第一通信设备基于所述第一信息确定的,所述第二信息为待部署在所述第二通信设备中的所述第一模型的模型部署信息;处理模块,用于基于所述第二信息进行模型部署。
- 一种通信系统,其特征在于,包括:第一通信设备,被配置为执行如权利要求1-14中任一项所述的模型初始化方法;第二通信设备,被配置为执行如权利要求15-24中任一项所述的模型初始化方法。
- 一种通信设备,其特征在于,包括:一个或多个处理器;其中,所述处理器用于调用指令以使得所述通信设备执行权利要求1-14、15-24中任一项所述的模型初始化方法。
- 一种存储介质,所述存储介质存储有指令,其特征在于,当所述指令在通信设备上运行 时,使得所述通信设备执行权利要求1-14、15-24中任一项所述的模型初始化方法。
Priority Applications (2)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| PCT/CN2023/108752 WO2025020009A1 (zh) | 2023-07-21 | 2023-07-21 | 一种模型初始化方法及其装置 |
| CN202380098480.7A CN121359493A (zh) | 2023-07-21 | 2023-07-21 | 一种模型初始化方法及其装置 |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| PCT/CN2023/108752 WO2025020009A1 (zh) | 2023-07-21 | 2023-07-21 | 一种模型初始化方法及其装置 |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| WO2025020009A1 true WO2025020009A1 (zh) | 2025-01-30 |
Family
ID=94373875
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| PCT/CN2023/108752 Pending WO2025020009A1 (zh) | 2023-07-21 | 2023-07-21 | 一种模型初始化方法及其装置 |
Country Status (2)
| Country | Link |
|---|---|
| CN (1) | CN121359493A (zh) |
| WO (1) | WO2025020009A1 (zh) |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111356171A (zh) * | 2018-12-21 | 2020-06-30 | 华为技术有限公司 | 一种信道状态信息csi上报的配置方法和通信装置 |
| CN115443643A (zh) * | 2022-08-03 | 2022-12-06 | 北京小米移动软件有限公司 | 确定用于压缩信道状态信息的压缩模型的方法、装置及存储介质 |
| WO2023283782A1 (zh) * | 2021-07-12 | 2023-01-19 | 北京小米移动软件有限公司 | 一种信道状态反馈的方法及其装置 |
| WO2023102045A1 (en) * | 2021-11-30 | 2023-06-08 | Interdigital Patent Holdings, Inc. | Pre-processing for csi compression in wireless systems |
-
2023
- 2023-07-21 WO PCT/CN2023/108752 patent/WO2025020009A1/zh active Pending
- 2023-07-21 CN CN202380098480.7A patent/CN121359493A/zh active Pending
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111356171A (zh) * | 2018-12-21 | 2020-06-30 | 华为技术有限公司 | 一种信道状态信息csi上报的配置方法和通信装置 |
| WO2023283782A1 (zh) * | 2021-07-12 | 2023-01-19 | 北京小米移动软件有限公司 | 一种信道状态反馈的方法及其装置 |
| WO2023102045A1 (en) * | 2021-11-30 | 2023-06-08 | Interdigital Patent Holdings, Inc. | Pre-processing for csi compression in wireless systems |
| CN115443643A (zh) * | 2022-08-03 | 2022-12-06 | 北京小米移动软件有限公司 | 确定用于压缩信道状态信息的压缩模型的方法、装置及存储介质 |
Also Published As
| Publication number | Publication date |
|---|---|
| CN121359493A (zh) | 2026-01-16 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN116941188B (zh) | 信息处理方法及装置、通信设备、通信系统、存储介质 | |
| WO2025060110A1 (zh) | 传输csi的方法、终端、网络设备、通信系统及存储介质 | |
| WO2025166665A1 (zh) | 通信方法、核心网设备、终端、通信系统及存储介质 | |
| WO2024234195A1 (zh) | 信息处理方法及装置、通信设备、通信系统、存储介质 | |
| WO2025081378A1 (zh) | 信息处理方法、终端、网络设备、通信系统及存储介质 | |
| WO2025020009A1 (zh) | 一种模型初始化方法及其装置 | |
| WO2025030294A1 (zh) | 信息处理方法及装置、通信设备、通信系统、存储介质 | |
| WO2024234201A1 (zh) | 通信方法及装置、通信设备、通信系统、存储介质 | |
| WO2025015484A1 (zh) | 模型切换更新方法及装置 | |
| WO2025000400A1 (zh) | 数据采集方法及装置、设备、存储介质 | |
| WO2026000215A1 (zh) | 模型性能监测方法、设备、系统及存储介质 | |
| WO2025236179A1 (zh) | 码本确定方法、装置以及存储介质 | |
| CN117280829A (zh) | 通信方法及装置、通信设备、通信系统、存储介质 | |
| CN117256185A (zh) | 通信方法及装置、通信设备、通信系统、存储介质 | |
| WO2026016138A1 (zh) | 模型性能监测方法、设备和存储介质 | |
| WO2025065294A1 (zh) | 模型监测方法及装置、通信设备、通信系统、存储介质 | |
| WO2025081439A1 (zh) | 通信方法、设备和存储介质 | |
| WO2025081374A1 (zh) | 通信方法、终端、网络设备、通信系统及存储介质 | |
| WO2025208307A1 (zh) | 信息发送方法及装置 | |
| WO2025208308A1 (zh) | 信息处理方法及装置 | |
| WO2025123365A1 (zh) | 通信方法、网络设备、终端、通信系统及存储介质 | |
| WO2026025378A1 (zh) | 数据传输方法、第一设备、第二设备及系统和存储介质 | |
| WO2026065071A1 (zh) | 通信方法、设备、系统及存储介质 | |
| WO2025184789A1 (zh) | 信息长度确定方法、装置以及存储介质 | |
| WO2025030276A1 (zh) | 信息的处理方法、终端及网络设备 |
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: 23946103 Country of ref document: EP Kind code of ref document: A1 |
|
| NENP | Non-entry into the national phase |
Ref country code: DE |