WO2023108436A1 - 信号检测方法和装置 - Google Patents

信号检测方法和装置 Download PDF

Info

Publication number
WO2023108436A1
WO2023108436A1 PCT/CN2021/138021 CN2021138021W WO2023108436A1 WO 2023108436 A1 WO2023108436 A1 WO 2023108436A1 CN 2021138021 W CN2021138021 W CN 2021138021W WO 2023108436 A1 WO2023108436 A1 WO 2023108436A1
Authority
WO
WIPO (PCT)
Prior art keywords
learning model
deep learning
training
parameter
matrix
Prior art date
Application number
PCT/CN2021/138021
Other languages
English (en)
French (fr)
Inventor
陈栋
池连刚
Original Assignee
北京小米移动软件有限公司
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 北京小米移动软件有限公司 filed Critical 北京小米移动软件有限公司
Priority to PCT/CN2021/138021 priority Critical patent/WO2023108436A1/zh
Publication of WO2023108436A1 publication Critical patent/WO2023108436A1/zh

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04B—TRANSMISSION
    • H04B17/00—Monitoring; Testing
    • H04B17/30—Monitoring; Testing of propagation channels
    • H04B17/391—Modelling the propagation channel
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04B—TRANSMISSION
    • H04B7/00—Radio transmission systems, i.e. using radiation field
    • H04B7/02—Diversity systems; Multi-antenna system, i.e. transmission or reception using multiple antennas
    • H04B7/04—Diversity systems; Multi-antenna system, i.e. transmission or reception using multiple antennas using two or more spaced independent antennas
    • H04B7/0413—MIMO systems

Definitions

  • the present disclosure relates to the field of communication technologies, and in particular to a signal detection method and device.
  • Massive millimeter-wave MIMO (multiple-input multiple-output) technology is regarded as a key technology in future wireless communication, and it is also a basic component of 5G wireless communication network.
  • massive millimeter wave MIMO has a larger number of receiving and transmitting antennas, which brings performance advantages, but at the same time, the complexity of signal detection also increases. Therefore, how to reduce the complexity of signal detection in massive millimeter-wave MIMO is an urgent problem to be solved.
  • Embodiments of the present disclosure provide a signal detection method and device, so as to reduce the complexity of signal detection.
  • an embodiment of the present disclosure provides a signal detection method, which includes: receiving a signal from a terminal device, performing channel estimation, and obtaining a channel matrix; performing data preprocessing on the channel matrix, and obtaining the first matrix and the first Vector; the first matrix and the first vector are input to the trained deep learning model to obtain an estimated detection signal; wherein, the trained deep learning model includes N trained convolutional network models; Wherein, N is a positive integer equal to the number of transmitting antennas.
  • the signal of the terminal equipment is received, the channel is estimated, and the channel matrix is obtained; the data preprocessing is performed on the channel matrix, and the first matrix and the first vector are obtained; the first matrix and the first vector are input into the trained
  • the deep learning model is obtained to estimate the detection signal; wherein, the trained deep learning model includes N trained convolutional network models; wherein, N is a positive integer and equal to the number of transmitting antennas.
  • an embodiment of the present disclosure provides another signal detection method, which includes: sending a signal and an intermediate parameter; wherein, the intermediate parameter is obtained by training a meta-learning parameter learning model based on network fusion.
  • the embodiment of the present disclosure provides a communication device, the communication device has some or all functions of the network side equipment in the method described in the first aspect above, for example, the function of the communication device may have part or all of the functions in the present disclosure.
  • the functions in all of the embodiments may also have the functions of implementing any one of the embodiments in the present disclosure independently.
  • the functions described above may be implemented by hardware, or may be implemented by executing corresponding software on the hardware.
  • the hardware or software includes one or more units or modules corresponding to the above functions.
  • the structure of the communication device may include a transceiver module and a processing module, and the processing module is configured to support the communication device to perform corresponding functions in the foregoing method.
  • the transceiver module is used to support communication between the communication device and other equipment.
  • the communication device may further include a storage module, which is used to be coupled with the transceiver module and the processing module, and stores necessary computer programs and data of the communication device.
  • the processing module may be a processor
  • the transceiver module may be a transceiver or a communication interface
  • the storage module may be a memory
  • the communication device includes: a transceiver module, configured to receive a signal from a terminal device; a processing module, configured to perform channel estimation to obtain a channel matrix; perform data preprocessing on the channel matrix to obtain the first A matrix and a first vector; the first matrix and the first vector are input to the trained deep learning model to obtain an estimated detection signal; wherein, the trained deep learning model includes N trained volumes Product network model; wherein, N is a positive integer and equal to the number of transmitting antennas; wherein, the intermediate parameters are obtained by training the meta-learning parameter learning model based on network fusion.
  • the embodiment of the present disclosure provides another communication device, which has some or all functions of the terminal device in the method example described in the second aspect above, for example, the function of the communication device may have part of the present disclosure Or the functions in all the embodiments may also have the function of implementing any one embodiment in the present disclosure alone.
  • the functions described above may be implemented by hardware, or may be implemented by executing corresponding software on the hardware.
  • the hardware or software includes one or more units or modules corresponding to the above functions.
  • the structure of the communication device may include a transceiver module.
  • the transceiver module is used to support communication between the communication device and other devices.
  • the communication device may further include a storage module, which is used to be coupled with the transceiver module and the processing module, and stores necessary computer programs and data of the communication device.
  • the communication device includes: a transceiver module, configured to send signals and intermediate parameters; wherein, the intermediate parameters are obtained by training a meta-learning parameter learning model based on network fusion.
  • an embodiment of the present disclosure provides a communication device, where the communication device includes a processor, and when the processor invokes a computer program in a memory, executes the method described in the first aspect above.
  • an embodiment of the present disclosure provides a communication device, where the communication device includes a processor, and when the processor invokes a computer program in a memory, it executes the method described in the second aspect above.
  • an embodiment of the present disclosure provides a communication device, the communication device includes a processor and a memory, and a computer program is stored in the memory; the processor executes the computer program stored in the memory, so that the communication device executes The method described in the first aspect above.
  • an embodiment of the present disclosure provides a communication device, the communication device includes a processor and a memory, and a computer program is stored in the memory; the processor executes the computer program stored in the memory, so that the communication device executes The method described in the second aspect above.
  • an embodiment of the present disclosure provides a communication device, the device includes a processor and an interface circuit, the interface circuit is used to receive code instructions and transmit them to the processor, and the processor is used to run the code instructions to make the The device executes the method described in the first aspect above.
  • an embodiment of the present disclosure provides a communication device, the device includes a processor and an interface circuit, the interface circuit is used to receive code instructions and transmit them to the processor, and the processor is used to run the code instructions to make the The device executes the method described in the second aspect above.
  • an embodiment of the present disclosure provides a communication system, the system includes the communication device described in the third aspect and the communication device described in the fourth aspect, or the system includes the communication device described in the fifth aspect and The communication device described in the sixth aspect, or, the system includes the communication device described in the seventh aspect and the communication device described in the eighth aspect, or, the system includes the communication device described in the ninth aspect and the communication device described in the tenth aspect the communication device described above.
  • an embodiment of the present invention provides a computer-readable storage medium for storing instructions used by the above-mentioned network-side equipment, and when the instructions are executed, the network-side equipment executes the above-mentioned first aspect. described method.
  • the embodiment of the present invention provides a readable storage medium, which is used to store the instructions used by the above-mentioned terminal equipment, and when the instructions are executed, the terminal equipment executes the method described in the above-mentioned second aspect .
  • the present disclosure further provides a computer program product including a computer program, which, when run on a computer, causes the computer to execute the method described in the first aspect above.
  • the present disclosure further provides a computer program product including a computer program, which, when run on a computer, causes the computer to execute the method described in the second aspect above.
  • the present disclosure provides a chip system, which includes at least one processor and an interface, configured to support the network side device to implement the functions involved in the first aspect, for example, determine or process the At least one of data and information.
  • the chip system further includes a memory, and the memory is configured to store necessary computer programs and data of the network side device.
  • the system-on-a-chip may consist of chips, or may include chips and other discrete devices.
  • the present disclosure provides a chip system
  • the chip system includes at least one processor and an interface, used to support the terminal device to implement the functions involved in the second aspect, for example, determine or process the data involved in the above method and at least one of information.
  • the chip system further includes a memory, and the memory is configured to store necessary computer programs and data of the terminal device.
  • the system-on-a-chip may consist of chips, or may include chips and other discrete devices.
  • the present disclosure provides a computer program that, when run on a computer, causes the computer to execute the method described in the first aspect above.
  • the present disclosure provides a computer program that, when run on a computer, causes the computer to execute the method described in the second aspect above.
  • FIG. 1 is an architecture diagram of a communication system provided by an embodiment of the present disclosure
  • FIG. 2 is a flowchart of a signal detection method provided by an embodiment of the present disclosure
  • Fig. 3 is a flowchart of another signal detection method provided by an embodiment of the present disclosure.
  • FIG. 4 is a flowchart of another signal detection method provided by an embodiment of the present disclosure.
  • FIG. 5 is a flow chart of another signal detection method provided by an embodiment of the present disclosure.
  • FIG. 6 is a flowchart of another signal detection method provided by an embodiment of the present disclosure.
  • FIG. 7 is a flowchart of another signal detection method provided by an embodiment of the present disclosure.
  • FIG. 8 is a structural diagram of a prediction gradient-based meta-learning model provided by an embodiment of the present disclosure.
  • FIG. 9 is a structural diagram of an LSTM-based meta-learning model provided by an embodiment of the present disclosure.
  • FIG. 10 is a flowchart of another signal detection method provided by an embodiment of the present disclosure.
  • Fig. 11 is a structural diagram of a communication device provided by an embodiment of the present disclosure.
  • Fig. 12 is a structural diagram of another communication device provided by an embodiment of the present disclosure.
  • FIG. 13 is a structural diagram of a chip provided by an embodiment of the present disclosure.
  • FIG. 1 is a schematic structural diagram of a communication system 10 provided by an embodiment of the present disclosure.
  • the communication system 10 may include, but is not limited to, a network side device and a terminal device.
  • the number and shape of the devices shown in FIG. More than one network side device, two or more terminal devices.
  • the communication system 10 shown in FIG. 1 includes one network side device 101 and one terminal device 102 as an example.
  • LTE long term evolution
  • 5th generation 5th generation
  • 5G new radio new radio, NR
  • side link in the embodiment of the present disclosure may also be referred to as a direct link or a direct link.
  • the network-side device 101 in the embodiment of the present disclosure is an entity on the network side for transmitting or receiving signals.
  • the network side device 101 may be an evolved base station (evolved NodeB, eNB), a transmission point (transmission reception point, TRP), a next generation base station (next generation NodeB, gNB) in an NR system, or a A base station or an access node in a wireless fidelity (wireless fidelity, WiFi) system, etc.
  • eNB evolved NodeB
  • TRP transmission reception point
  • gNB next generation base station
  • a base station or an access node in a wireless fidelity (wireless fidelity, WiFi) system etc.
  • the embodiments of the present disclosure do not limit the specific technology and specific device form adopted by the network side device.
  • the network side device may be composed of a central unit (CU) and a distributed unit (DU), wherein the CU may also be called a control unit, and the CU-
  • the structure of the DU can separate the protocol layers of network-side devices, such as base stations.
  • the functions of some protocol layers are centrally controlled by the CU, and the remaining part or all of the functions of the protocol layers are distributed in the DU, which is centrally controlled by the CU.
  • the terminal device 102 in the embodiment of the present disclosure is an entity on the user side for receiving or transmitting signals, such as a mobile phone.
  • the terminal device may also be called a terminal (terminal), user equipment (user equipment, UE), mobile station (mobile station, MS), mobile terminal (mobile terminal, MT), etc.
  • the terminal device can be a car with communication function, smart car, mobile phone, wearable device, tablet computer (Pad), computer with wireless transceiver function, virtual reality (virtual reality, VR) terminal, augmented reality (augmented reality, AR) terminals, wireless terminals in industrial control, wireless terminals in self-driving, wireless terminals in remote medical surgery, and smart grids Wireless terminals, wireless terminals in transportation safety, wireless terminals in smart city, wireless terminals in smart home, etc.
  • the embodiments of the present disclosure do not limit the specific technology and specific device form adopted by the terminal device.
  • massive millimeter-wave MIMO has a larger number of transceiver antennas, which brings performance advantages, but at the same time, the complexity of signal detection also increases.
  • MMSE minimum mean square error
  • an embodiment of the present disclosure provides a signal detection method to reduce the network calculation complexity of signal detection.
  • FIG. 2 is a flow chart of a signal detection method provided by an embodiment of the present disclosure.
  • the method is performed by a network side device, and the method may include but not limited to the following steps:
  • S21 Receive a signal from a terminal device, perform channel estimation, and obtain a channel matrix.
  • the terminal device information symbol stream is separated into N t parallel sub-streams, and after modulation, they are simultaneously transmitted through N t transmit antennas to achieve the maximum transmission rate.
  • the signals received by the N r receiving antennas of the network side equipment are input to the MIMO detector for detection.
  • the signal model of the MIMO system can be expressed as:
  • H represents N t ⁇ N r dimensional channel matrix
  • the signal received by the receiving end is the aliasing of signals transmitted by different transmitting antennas. Therefore, the role of the MIMO detector is to separate the aliased transmitted signals and restore the transmitted vector x when the received signal vector y and the MIMO channel matrix H are known.
  • a signal of a terminal device is received, channel estimation is performed, and a channel matrix H is obtained.
  • S22 Perform data preprocessing on the channel matrix to obtain a first matrix and a first vector.
  • performing data preprocessing on the channel matrix H is to perform QR decomposition on the channel matrix H:
  • QR decomposition is performed on the channel matrix H to obtain the first matrix R and the second matrix Q. Then according to the first matrix R and the second matrix Q, the expression of ⁇ y-Hx ⁇ 2 satisfies the following relationship:
  • S23 Input the first matrix and the first vector into the trained deep learning model to obtain an estimated detection signal; wherein, the trained deep learning model includes N trained convolutional network models; where N is a positive integer and is equal to the number of transmitting antennas.
  • the first matrix R and the first vector z obtained through the above calculation are input to the trained deep learning model to obtain an estimated detection signal.
  • the trained deep learning model includes N trained convolutional network models, where N is a positive integer equal to the number N t of transmitting antennas.
  • N is a positive integer equal to the number N t of transmitting antennas.
  • the trained deep learning model instead of the K-best detection algorithm to select the optimal path, perform maximum likelihood detection, obtain the optimal path and retain the path nodes , to estimate the detection signal in a further stage, which can reduce the computational complexity.
  • the signal of the terminal equipment is received, the channel is estimated, and the channel matrix is obtained; the data preprocessing is performed on the channel matrix, and the first matrix and the first vector are obtained; the first matrix and the first vector are input into the trained
  • the deep learning model is obtained to estimate the detection signal; wherein, the trained deep learning model includes N trained convolutional network models; wherein, N is a positive integer and equal to the number of transmitting antennas.
  • FIG. 3 is a flow chart of a signal detection method provided by an embodiment of the present disclosure.
  • the method is performed by the network side device, and the method may include but not limited to the following steps:
  • S31 Receive a signal from a terminal device, perform channel estimation, and obtain a channel matrix.
  • S32 Decompose the channel matrix to generate a first matrix and a second matrix.
  • S34 Input the first vector and the first matrix to N trained convolutional network models to perform maximum likelihood detection, and each trained convolutional network model outputs K path nodes corresponding to the target value of the current layer;
  • N trained convolutional network models correspond to N layers of the search tree, and the target value K is a positive integer.
  • the maximum likelihood detection problem can be transformed into a minimum path search problem for a weighted tree, the jth element of the emission vector x corresponds to the jth layer of the tree, and each node in the jth layer can be from the root node to the The path of a node is uniquely determined.
  • each trained convolutional network model corresponds to a layer of the search tree, and correspondingly outputs K path nodes of the target value of the current layer
  • Each trained convolutional network model for path selection includes several compound convolutional layers.
  • the first vector z and the first matrix R are used as input, and the input dimension is N t ⁇ N t , using a convolution layer with a convolution kernel size of m ⁇ m and a number of l, the activation function adopts the ReLU activation function, and the output layer’s
  • the dimension is the target value K, where N t , m, l, and K are all positive integers.
  • S35 Summarize and calculate the target value K path nodes of the current layer output by each trained convolutional network model, and generate an estimated detection signal.
  • Each element j of the emission vector x is assigned an expression for the path metric PD satisfying the following relationship:
  • PD(x j ) PD(x j+1 )+ ⁇ (x j )
  • the path metric is non-decreasing, and the maximum likelihood detection can be equivalent to searching for the leaf node with the smallest metric in the tree.
  • the target value K path nodes of the current layer output by each trained convolutional network model are summarized and calculated An estimated detection signal is generated. In this way, the signal is detected, an estimated detection signal is generated, and the computational complexity is reduced under the condition close to the maximum likelihood detection performance.
  • FIG. 4 is a flowchart of a signal detection method provided by an embodiment of the present disclosure.
  • the method is performed by the network side device, and the method may include but not limited to the following steps:
  • the terminal device obtains the intermediate parameters based on the meta-learning parameter model training based on the network fusion, initializes the parameters of the learning target value K by the number of transmitting antennas and the modulation order, and obtains the intermediate parameters based on the meta-learning parameter model training based on the network fusion Then send it to the network side device.
  • the meta-learning method has a better parameter training effect and a faster convergence speed.
  • a variety of well-behaved meta-learning networks are used, and the learning results of the meta-learning network model are fused using the network fusion method, so that It can achieve better performance and stronger network generalization ability.
  • the intelligent parameter model performs approximate fitting to obtain a first target value; wherein, the first target value is a positive integer; and a deep learning model is constructed according to the first target value.
  • the first target value needs to reach a certain value, and the first target value is also set in the deep learning model.
  • the demand of the upper-layer nodes is obviously smaller than that of the lower-layer nodes, so the K-best algorithm is canceled for each layer to fix the value of the first target value, and the value of the first target value is designed according to the law of change as
  • the expression in the form of the fitting function satisfies the following relationship:
  • a, b, and c are all learnable parameters
  • the first target value k is the number of path node layers.
  • the intelligent parameter model is approximately fitted to obtain the first target value, and an appropriate approximate fitting function is designed for the changed first target value k.
  • the intelligent parameter model is deployed on the network side device, and then the positive
  • the integer first target value is fed back to the deep learning model to participate in the construction of the deep model, which increases the performance of the detection network and obtains lower computational complexity.
  • S51 and S52 can be implemented independently, or can be implemented in combination with any other steps in the embodiment of the present disclosure, for example, in combination with S31 to S33 and/or S41 to S44 is implemented together, which is not limited in this embodiment of the present disclosure.
  • FIG. 5 is a flow chart of a signal detection method provided by an embodiment of the present disclosure.
  • the method is performed by a network side device, and the method may include but not limited to the following steps:
  • S51 Acquire a training data set; wherein, the training data set includes at least one set of training transmission signals and training reception signals.
  • At least one set of training transmission signals and training reception signals are simulated and generated to obtain a training data set.
  • S52 Input the training data set into the deep learning model, train the deep learning model, and generate a trained deep learning model.
  • the training transmission signal is input into the deep learning model to obtain a prediction signal; according to the prediction signal and the training reception signal, the deep learning model is updated to generate a trained deep learning model.
  • At least one set of training transmission signals and training reception signals are simulated and generated according to the channel matrix H obtained by channel estimation, and besides the training data set, a verification data set and a test data set are also included.
  • the verification data set includes at least one set of verification transmission signals and verification reception signals
  • the test data set includes at least one set of test transmission signals and test reception signals.
  • the trained deep learning model includes the target value K, and the method also includes:
  • the first target value is updated to obtain the target value K.
  • the intermediate parameters obtained by training the terminal device based on the meta-learning parameter learning model of network integration are approximately fitted through the intelligent parameter model to obtain the first target value, participate in the construction of the deep learning model, and then During the training process of the deep learning model, joint training is performed on the first target value to obtain the target value K, so that the deep learning model can learn the global optimum.
  • the trained deep learning model including the target value K is deployed on the network device.
  • FIG. 6 is a flow chart of a signal detection method provided by an embodiment of the present disclosure.
  • the method is performed by the network side device, and the method may include but not limited to the following steps:
  • S61 Send a model update instruction in response to a preset condition being met; wherein, the preset condition is that the change of the channel state information exceeds a certain range.
  • deep learning models may experience severe delays in collecting sufficient training data online, especially in low signal-to-noise ratio states, which brings difficulties to the training and deployment of deep learning models.
  • the network side device in response to the preset condition being satisfied, detects that the channel state information has changed, and evaluates the performance of the trained deep learning model in the current channel environment. When the performance is significantly reduced , to send the model update command to the terminal device.
  • the online update process of the network is designed.
  • the model update command is sent to the terminal device, so as to fully Utilizing the advantages of meta-learning in obtaining training data and fast updating, the meta-learning parameter learning model training based on network fusion at the terminal device obtains updated intermediate parameters.
  • the network side device receives the updated intermediate parameters obtained by the terminal device based on the network fusion meta-learning parameter learning model training; according to the updated intermediate parameters, the intelligent parameter model performs approximate fitting to obtain the second target value; wherein, the first The second target value is a positive integer; a deep learning model to be updated is constructed according to the second target value.
  • the update training data set is obtained; wherein, the update training data set includes at least one set of update training transmission signals and update training reception signals; the update training data set is input to the deep learning model to be updated, and the deep learning model to be updated Perform training to generate an updated trained deep learning model.
  • the deep learning model to be updated is constructed according to the second target value, and then the updated training data set is obtained; wherein, the updated training data set includes at least one set of updated Training the transmitting signal and updating the training receiving signal; inputting the updated training data set to the deep learning model to be updated, training the deep learning model to be updated, generating the updated trained deep learning model and constructing the deep learning model, and
  • the process of training the deep learning model is similar, and reference may be made to the relevant descriptions of the above embodiments, and details are not repeated here.
  • FIG. 7 is a flowchart of a signal detection method provided by an embodiment of the present disclosure.
  • the method is executed by the terminal device, and the method may include but not limited to the following steps:
  • S71 Send a signal and an intermediate parameter; wherein, the intermediate parameter is obtained by training a meta-learning parameter learning model based on network fusion.
  • the information symbol stream in the MIMO system, in order to obtain the maximum transmission rate, in the terminal device, the information symbol stream is separated into N t parallel sub-streams, and after modulation, the information symbol streams are sent simultaneously through N t transmitting antennas, and the terminal device sends the signal .
  • the terminal device is deployed with a network-integrated meta-learning parameter learning model.
  • the meta-learning method can quickly learn new tasks on the basis of acquiring existing "knowledge". The intention is to quickly learn through a small number of training examples Or a network model that adapts to a new environment and has the ability to learn to learn.
  • At least one first parameter is initialized according to the number of transmitting antennas and the modulation order, the first parameter is input to the meta-learning parameter learning model of network fusion, the intermediate parameters obtained by training are performed, and the intermediate parameters are sent to the network side device.
  • At least one first parameter is initialized according to the number of transmit antennas and the modulation order, for example, parameter a, parameter b, and parameter c are initialized.
  • parameter a, parameter b, and parameter c are all learnable parameters.
  • the network fusion meta-learning parameter learning model in the embodiments of the present disclosure includes a prediction gradient-based meta-learning model, an LSTM-based meta-learning model, and a network fusion model.
  • the first intermediate parameters are generated by inputting the first parameters into the predicted gradient-based meta-learning model.
  • the first parameter is input to the meta-learning model based on LSTM to generate the second intermediate parameter.
  • the meta-learning model based on the prediction gradient predicts the gradient by training a general-purpose neural network, and trains it through the regression problem of the quadratic equation in one variable, and the gradient descent of the obtained neural network optimizer is faster and more accurate. for quick learning.
  • LSTM is a special recurrent neural network that deletes or adds information to cell state C through a gate structure.
  • the parameter X can be learned through LSTM network training, the current learnable parameter is input, and the new update result is directly output.
  • the prediction gradient-based meta-learning model and the LSTM-based meta-learning model both show good performance, and the network fusion model is designed to fuse the learning results of meta-learning networks with different structures, so that it can achieve better performance than multiple Fine-tuned models yield better results.
  • the expression of the network fusion model satisfies the following relationship:
  • Y 1 is the output result of parameter learning based on the meta-learning model based on prediction gradient
  • Y 2 is the output result of parameter learning based on LSTM meta-learning model
  • W 1 and W 2 are learnable scalars, which can be learned through the network W 1 and W 2 .
  • the network fusion learning method is used through the network fusion model, which enhances the network generalization ability while improving the performance of the network, and enhances the ability to cope with channel changes.
  • FIG. 10 is a flowchart of a signal detection method provided by an embodiment of the present disclosure.
  • the method is executed by the terminal device, and the method may include but not limited to the following steps:
  • S101 Receive a model update instruction from a network side device.
  • S102 Re-initialize at least one second parameter according to the number of transmit antennas and the modulation order.
  • the terminal device receives the model update instruction sent by the network side device, re-initializes at least one second parameter, and then inputs the second parameter into the meta-learning parameter learning model of network integration, and performs training to obtain the updated intermediate parameter.
  • the method is the same as in the above embodiment, according to the number of transmitting antennas and the modulation order, initialize at least one first parameter; input the first parameter to the meta-learning parameter learning model of network fusion, perform the intermediate parameters obtained by training, and send the intermediate parameters.
  • a model update instruction is sent to the terminal device, and on the terminal device side, at least one second parameter is re-initialized according to the number of transmitting antennas and the modulation order, using
  • the meta-learning parameter learning model of network fusion can be retrained to update the intermediate parameters. While using the meta-learning method, it can use previous knowledge and experience to guide the learning of new tasks, and quickly update the deep learning model through a small number of training examples.
  • the methods provided in the embodiments of the present disclosure are introduced from the perspectives of the network side device and the terminal device respectively.
  • the network-side device and the terminal device may include a hardware structure and a software module, and realize the above-mentioned functions in the form of a hardware structure, a software module, or a hardware structure plus a software module.
  • a certain function among the above-mentioned functions may be implemented in the form of a hardware structure, a software module, or a hardware structure plus a software module.
  • FIG. 11 is a schematic structural diagram of a communication device 1 provided by an embodiment of the present disclosure.
  • the communication device 1 shown in FIG. 11 may include a transceiver module 11 and a processing module 12 .
  • the transceiver module 11 may include a sending module and/or a receiving module, the sending module is used to realize the sending function, the receiving module is used to realize the receiving function, and the sending and receiving module 11 can realize the sending function and/or the receiving function.
  • the communication device 1 may be a network-side device, or a device in the network-side device, or a device that can be matched with the network-side device.
  • the communication device 1 may be a terminal device, may also be a device in a terminal device, or may be a device that can be matched and used with the terminal device.
  • the communication device 1 is a network-side device: a transceiver module 11, configured to receive a signal from a terminal device.
  • the processing module 12 is used to perform channel estimation to obtain a channel matrix; perform data preprocessing on the channel matrix to obtain a first matrix and a first vector; input the first matrix and the first vector to the trained A deep learning model to obtain an estimated detection signal; wherein, the trained deep learning model includes N trained convolutional network models; wherein, N is a positive integer equal to the number of transmitting antennas.
  • the communication device 1 is a terminal device: the transceiver module 11 is used to send signals and intermediate parameters; wherein, the intermediate parameters are obtained by training a meta-learning parameter learning model based on network integration.
  • FIG. 12 is a schematic structural diagram of another communication device 1000 provided by an embodiment of the present disclosure.
  • the communication device 1000 may be a network-side device, or a terminal device, or a chip, a chip system, or a processor that supports the network-side device to implement the above method, or a chip or a chip system that supports the terminal device to implement the above method , or processor, etc.
  • the communication device 1000 may be used to implement the methods described in the foregoing method embodiments, and for details, refer to the descriptions in the foregoing method embodiments.
  • the communication device 1000 may be a network-side device, or a terminal device, or a chip, a chip system, or a processor that supports the network-side device to implement the above method, or a chip or a chip system that supports the terminal device to implement the above method , or processor, etc.
  • the device can be used to implement the methods described in the above method embodiments, and for details, refer to the descriptions in the above method embodiments.
  • the communication device 1000 may include one or more processors 1001 .
  • the processor 1001 may be a general purpose processor or a special purpose processor or the like. For example, it can be a baseband processor or a central processing unit.
  • the baseband processor can be used to process communication protocols and communication data
  • the central processing unit can be used to control communication devices (such as base stations, baseband chips, terminal equipment, terminal equipment chips, DU or CU, etc.) and execute computer programs , to process data for computer programs.
  • the communication device 1000 may further include one or more memories 1002, on which a computer program 1004 may be stored, and the memory 1002 executes the computer program 1004, so that the communication device 1000 executes the methods described in the foregoing method embodiments .
  • data may also be stored in the memory 1002 .
  • the communication device 1000 and the memory 1002 can be set separately or integrated together.
  • the communication device 1000 may further include a transceiver 1005 and an antenna 1006 .
  • the transceiver 1005 may be called a transceiver unit, a transceiver, or a transceiver circuit, etc., and is used to implement a transceiver function.
  • the transceiver 1005 may include a receiver and a transmitter, and the receiver may be called a receiver or a receiving circuit for realizing a receiving function; the transmitter may be called a transmitter or a sending circuit for realizing a sending function.
  • the communication device 1000 may further include one or more interface circuits 1007 .
  • the interface circuit 1007 is used to receive code instructions and transmit them to the processor 1001 .
  • the processor 1001 runs the code instructions to enable the communication device 1000 to execute the methods described in the foregoing method embodiments.
  • the communication device 1000 is a network side device: the transceiver 1005 is used to execute S21 in FIG. 2; S31 in FIG. 3; S41 in FIG. 4; S51 in FIG. 5; S61 in FIG. 6; the processor 1001 is used to execute S22, S23 in Fig. 2; S32, S33, S34, S35 in Fig. 3; S42 in Fig. 4; S52 in Fig. 5 .
  • the communication device 1000 is a terminal device: the transceiver 1005 is used to execute S71 in FIG. 7 ; S101 in FIG. 10 ; the processor 1001 is used to execute S102, S103, and S104 in FIG. 10 .
  • the processor 1001 may include a transceiver for implementing receiving and sending functions.
  • the transceiver may be a transceiver circuit, or an interface, or an interface circuit.
  • the transceiver circuits, interfaces or interface circuits for realizing the functions of receiving and sending can be separated or integrated together.
  • the above-mentioned transceiver circuit, interface or interface circuit may be used for reading and writing code/data, or the above-mentioned transceiver circuit, interface or interface circuit may be used for signal transmission or transmission.
  • the processor 1001 may store a computer program 1003, and the computer program 1003 runs on the processor 1001 to enable the communication device 1000 to execute the methods described in the foregoing method embodiments.
  • the computer program 1003 may be solidified in the processor 1001, and in this case, the processor 1001 may be implemented by hardware.
  • the communication device 1000 may include a circuit, and the circuit may implement the function of sending or receiving or communicating in the foregoing method embodiments.
  • the processors and transceivers described in this disclosure can be implemented on integrated circuits (integrated circuits, ICs), analog ICs, radio frequency integrated circuits (RFICs), mixed signal ICs, application specific integrated circuits (ASICs), printed circuit boards ( printed circuit board, PCB), electronic equipment, etc.
  • the processor and transceiver can also be fabricated using various IC process technologies such as complementary metal oxide semiconductor (CMOS), nMetal-oxide-semiconductor (NMOS), P-type Metal oxide semiconductor (positive channel metal oxide semiconductor, PMOS), bipolar junction transistor (bipolar junction transistor, BJT), bipolar CMOS (BiCMOS), silicon germanium (SiGe), gallium arsenide (GaAs), etc.
  • CMOS complementary metal oxide semiconductor
  • NMOS nMetal-oxide-semiconductor
  • PMOS P-type Metal oxide semiconductor
  • BJT bipolar junction transistor
  • BiCMOS bipolar CMOS
  • SiGe silicon germanium
  • GaAs gallium arsenide
  • the communication device described in the above embodiments may be a terminal device, but the scope of the communication device described in the present disclosure is not limited thereto, and the structure of the communication device may not be limited by FIG. 12 .
  • a communication device may be a stand-alone device or may be part of a larger device.
  • the communication device may be:
  • a set of one or more ICs may also include storage components for storing data and computer programs;
  • ASIC such as modem (Modem);
  • FIG. 13 is a structural diagram of a chip provided in an embodiment of the present disclosure.
  • the chip 1100 includes a processor 1101 and an interface 1103 .
  • the number of processors 1101 may be one or more, and the number of interfaces 1103 may be more than one.
  • Interface 1103 configured to receive code instructions and transmit them to the processor.
  • the processor 1101 is configured to run code instructions to execute the signal detection method as described in some of the above embodiments.
  • Interface 1103 configured to receive code instructions and transmit them to the processor.
  • the processor 1101 is configured to run code instructions to execute the signal detection method as described in some of the above embodiments.
  • the chip 1100 also includes a memory 1102 for storing necessary computer programs and data.
  • An embodiment of the present disclosure also provides a signal detection system, the system includes the aforementioned communication device as the network side device and the communication device as the terminal device in the embodiment of Figure 11, or the system includes the aforementioned embodiment of Figure 12 as the network side The communication device of the equipment and the communication device of the terminal equipment.
  • the present disclosure also provides a readable storage medium on which instructions are stored, and when the instructions are executed by a computer, the functions of any one of the above method embodiments are realized.
  • the present disclosure also provides a computer program product, which can realize the functions of any one of the above method embodiments when executed by a computer.
  • all or part of them may be implemented by software, hardware, firmware or any combination thereof.
  • software When implemented using software, it may be implemented in whole or in part in the form of a computer program product.
  • the computer program product comprises one or more computer programs. When the computer program is loaded and executed on the computer, all or part of the processes or functions according to the embodiments of the present disclosure will be generated.
  • the computer can be a general purpose computer, a special purpose computer, a computer network, or other programmable devices.
  • the computer program can be stored in or transmitted from one computer-readable storage medium to another computer-readable storage medium, for example, the computer program can be downloaded from a website, computer, server or data center Transmission to another website site, computer, server or data center by wired (such as coaxial cable, optical fiber, digital subscriber line (DSL)) or wireless (such as infrared, wireless, microwave, etc.).
  • the computer-readable storage medium may be any available medium that can be accessed by a computer, or a data storage device such as a server or a data center integrated with one or more available media.
  • the available medium may be a magnetic medium (for example, a floppy disk, a hard disk, a magnetic tape), an optical medium (for example, a high-density digital video disc (digital video disc, DVD)), or a semiconductor medium (for example, a solid state disk (solid state disk, SSD)) etc.
  • a magnetic medium for example, a floppy disk, a hard disk, a magnetic tape
  • an optical medium for example, a high-density digital video disc (digital video disc, DVD)
  • a semiconductor medium for example, a solid state disk (solid state disk, SSD)
  • At least one in the present disclosure can also be described as one or more, and a plurality can be two, three, four or more, and the present disclosure is not limited.
  • the technical feature is distinguished by "first”, “second”, “third”, “A”, “B”, “C” and “D”, etc.
  • the technical features described in the “first”, “second”, “third”, “A”, “B”, “C” and “D” have no sequence or order of magnitude among the technical features described.
  • each table in the present disclosure may be configured or predefined.
  • the values of the information in each table are just examples, and may be configured as other values, which are not limited in the present disclosure.
  • the corresponding relationship shown in some rows may not be configured.
  • appropriate deformation adjustments can be made based on the above table, for example, splitting, merging, and so on.
  • the names of the parameters shown in the titles of the above tables may also adopt other names understandable by the communication device, and the values or representations of the parameters may also be other values or representations understandable by the communication device.
  • other data structures can also be used, for example, arrays, queues, containers, stacks, linear tables, pointers, linked lists, trees, graphs, structures, classes, heaps, hash tables or hash tables can be used wait.
  • Predefinition in the present disclosure can be understood as definition, predefinition, storage, prestorage, prenegotiation, preconfiguration, curing, or prefiring.

Abstract

本公开实施例公开了一种信号检测方法和装置，应用于通信技术领域，其中，由网络侧设备执行的方法包括：接收终端设备的信号，进行信道估计，得到信道矩阵；对信道矩阵进行数据预处理，获取第一矩阵和第一向量；将第一矩阵和第一向量输入至训练好的深度学习模型，得到估计检测信号；其中，训练好的深度学习模型中包括N个训练好的卷积网络模型；其中，N为正整数且等于发射天线数。由此，能够降低信号检测的复杂度。

Description

信号检测方法和装置 技术领域
本公开涉及通信技术领域，尤其涉及一种信号检测方法和装置。
背景技术
大规模毫米波MIMO(multiple-input multiple-output，多输入多输出)技术被视为未来无线通信中的一项关键技术，也是5G无线通信网络中的基本组成部分。
其中，大规模毫米波MIMO相较于传统的MIMO技术，它的收发天线数较大，这带来了性能优势，但同时信号检测的复杂度也随之增大。因此，如何降低大规模毫米波MIMO的信号检测的复杂度是亟需解决的问题。
发明内容
本公开实施例提供一种信号检测方法和装置，以降低信号检测的复杂度。
第一方面，本公开实施例提供一种信号检测方法，该方法包括：接收终端设备的信号，进行信道估计，得到信道矩阵；对所述信道矩阵进行数据预处理，获取第一矩阵和第一向量；将所述第一矩阵和所述第一向量输入至训练好的深度学习模型，得到估计检测信号；其中，所述训练好的深度学习模型中包括N个训练好的卷积网络模型；其中，N为正整数且等于发射天线数。
通过实施本公开实施例，接收终端设备的信号，进行信道估计，得到信道矩阵；对信道矩阵进行数据预处理，获取第一矩阵和第一向量；将第一矩阵和第一向量输入至训练好的深度学习模型，得到估计检测信号；其中，训练好的深度学习模型中包括N个训练好的卷积网络模型；其中，N为正整数且等于发射天线数。由此，能够降低信号检测的复杂度。
第二方面，本公开实施例提供另一种信号检测方法，该方法包括：发送信号和中间参数；其中，基于网络融合的元学习参数学习模型训练得到所述中间参数。
第三方面，本公开实施例提供一种通信装置，该通信装置具有实现上述第一方面所述的方法中网络侧设备的部分或全部功能，比如通信装置的功能可具备本公开中的部分或全部实施例中的功能，也可以具备单独实施本公开中的任一个实施例的功能。所述功能可以通过硬件实现，也可以通过硬件执行相应的软件实现。所述硬件或软件包括一个或多个与上述功能相对应的单元或模块。
在一种实现方式中，该通信装置的结构中可包括收发模块和处理模块，所述处理模块被配置为支持通信装置执行上述方法中相应的功能。所述收发模块用于支持通信装置与其他设备之间的通信。所述通信装置还可以包括存储模块，所述存储模块用于与收发模块和处理模块耦合，其保存通信装置必要的计算机程序和数据。
作为示例，处理模块可以为处理器，收发模块可以为收发器或通信接口，存储模块可以为存储器。
在一种实现方式中，所述通信装置包括：收发模块，用于接收终端设备的信号；处理模块，用于进行信道估计，得到信道矩阵；对所述信道矩阵进行数据预处理，获取第一矩阵和第一向量；将所述第一矩阵和所述第一向量输入至训练好的深度学习模型，得到估计检测信号；其中，所述训练好的深度学习模型中包括N个训练好的卷积网络模型；其中，N为正整数且等于发射天线数；其中，基于网络融合的元学习参数学习模型训练得到所述中间参数。
第四方面，本公开实施例提供另一种通信装置，该通信装置具有实现上述第二方面所述的方法示例 中终端设备的部分或全部功能，比如通信装置的功能可具备本公开中的部分或全部实施例中的功能，也可以具备单独实施本公开中的任一个实施例的功能。所述功能可以通过硬件实现，也可以通过硬件执行相应的软件实现。所述硬件或软件包括一个或多个与上述功能相对应的单元或模块。
在一种实现方式中，该通信装置的结构中可包括收发模块。收发模块用于支持通信装置与其他设备之间的通信。所述通信装置还可以包括存储模块，所述存储模块用于与收发模块和处理模块耦合，其保存通信装置必要的计算机程序和数据。
在一种实现方式中，所述通信装置包括：收发模块，用于发送信号和中间参数；其中，基于网络融合的元学习参数学习模型训练得到所述中间参数。
第五方面，本公开实施例提供一种通信装置，该通信装置包括处理器，当该处理器调用存储器中的计算机程序时，执行上述第一方面所述的方法。
第六方面，本公开实施例提供一种通信装置，该通信装置包括处理器，当该处理器调用存储器中的计算机程序时，执行上述第二方面所述的方法。
第七方面，本公开实施例提供一种通信装置，该通信装置包括处理器和存储器，该存储器中存储有计算机程序；所述处理器执行该存储器所存储的计算机程序，以使该通信装置执行上述第一方面所述的方法。
第八方面，本公开实施例提供一种通信装置，该通信装置包括处理器和存储器，该存储器中存储有计算机程序；所述处理器执行该存储器所存储的计算机程序，以使该通信装置执行上述第二方面所述的方法。
第九方面，本公开实施例提供一种通信装置，该装置包括处理器和接口电路，该接口电路用于接收代码指令并传输至该处理器，该处理器用于运行所述代码指令以使该装置执行上述第一方面所述的方法。
第十方面，本公开实施例提供一种通信装置，该装置包括处理器和接口电路，该接口电路用于接收代码指令并传输至该处理器，该处理器用于运行所述代码指令以使该装置执行上述第二方面所述的方法。
第十一方面，本公开实施例提供一种通信系统，该系统包括第三方面所述的通信装置以及第四方面所述的通信装置，或者，该系统包括第五方面所述的通信装置以及第六方面所述的通信装置，或者，该系统包括第七方面所述的通信装置以及第八方面所述的通信装置，或者，该系统包括第九方面所述的通信装置以及第十方面所述的通信装置。
第十二方面，本发明实施例提供一种计算机可读存储介质，用于储存为上述网络侧设备所用的指令，当所述指令被执行时，使所述网络侧设备执行上述第一方面所述的方法。
第十三方面，本发明实施例提供一种可读存储介质，用于储存为上述终端设备所用的指令，当所述指令被执行时，使所述终端设备执行上述第二方面所述的方法。
第十四方面，本公开还提供一种包括计算机程序的计算机程序产品，当其在计算机上运行时，使得计算机执行上述第一方面所述的方法。
第十五方面，本公开还提供一种包括计算机程序的计算机程序产品，当其在计算机上运行时，使得计算机执行上述第二方面所述的方法。
第十六方面，本公开提供一种芯片系统，该芯片系统包括至少一个处理器和接口，用于支持网络侧设备实现第一方面所涉及的功能，例如，确定或处理上述方法中所涉及的数据和信息中的至少一种。在一种可能的设计中，所述芯片系统还包括存储器，所述存储器，用于保存网络侧设备必要的计算机程序和数据。该芯片系统，可以由芯片构成，也可以包括芯片和其他分立器件。
第十七方面，本公开提供一种芯片系统，该芯片系统包括至少一个处理器和接口，用于支持终端设 备实现第二方面所涉及的功能，例如，确定或处理上述方法中所涉及的数据和信息中的至少一种。在一种可能的设计中，所述芯片系统还包括存储器，所述存储器，用于保存终端设备必要的计算机程序和数据。该芯片系统，可以由芯片构成，也可以包括芯片和其他分立器件。
第十八方面，本公开提供一种计算机程序，当其在计算机上运行时，使得计算机执行上述第一方面所述的方法。
第十九方面，本公开提供一种计算机程序，当其在计算机上运行时，使得计算机执行上述第二方面所述的方法。
附图说明
为了更清楚地说明本公开实施例或背景技术中的技术方案，下面将对本公开实施例或背景技术中所需要使用的附图进行说明。
图1是本公开实施例提供的一种通信系统的架构图；
图2是本公开实施例提供的一种信号检测方法的流程图；
图3是本公开实施例提供的另一种信号检测方法的流程图；
图4是本公开实施例提供的又一种信号检测方法的流程图；
图5是本公开实施例提供的又一种信号检测方法的流程图；
图6是本公开实施例提供的又一种信号检测方法的流程图；
图7是本公开实施例提供的又一种信号检测方法的流程图；
图8是本公开实施例提供的一种基于预测梯度的元学习模型的结构图；
图9是本公开实施例提供的一种基于LSTM的元学习模型的结构图；
图10是本公开实施例提供的又一种信号检测方法的流程图；
图11是本公开实施例提供的一种通信装置的结构图；
图12是本公开实施例提供的另一种通信装置的结构图；
图13是本公开实施例提供的一种芯片的结构图。
具体实施方式
为了更好的理解本公开实施例公开的一种信号检测方法，下面首先对本公开实施例适用的通信系统进行描述。
请参见图1，图1为本公开实施例提供的一种通信系统10的架构示意图。该通信系统10可包括但不限于一个网络侧设备和一个终端设备，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网络侧设备，两个或两个以上的终端设备。图1所示的通信系统10以包括一个网络侧设备101和一个终端设备102为例。
需要说明的是，本公开实施例的技术方案可以应用于各种通信系统。例如：长期演进(long term evolution，LTE)系统、第五代(5th generation，5G)移动通信系统、5G新空口(new radio，NR)系统，或者其他未来的新型移动通信系统等。还需要说明的是，本公开实施例中的侧链路还可以称为直连链路或直通链路。
本公开实施例中的网络侧设备101是网络侧的一种用于发射或接收信号的实体。例如，网络侧设备101可以为演进型基站(evolved NodeB，eNB)、传输点(transmission reception point，TRP)、NR系 统中的下一代基站(next generation NodeB，gNB)、其他未来移动通信系统中的基站或无线保真(wireless fidelity，WiFi)系统中的接入节点等。本公开的实施例对网络侧设备所采用的具体技术和具体设备形态不做限定。本公开实施例提供的网络侧设备可以是由集中单元(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的结构可以将网络侧设备，例如基站的协议层拆分开，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。
本公开实施例中的终端设备102是用户侧的一种用于接收或发射信号的实体，如手机。终端设备也可以称为终端(terminal)、用户设备(user equipment，UE)、移动台(mobile station，MS)、移动终端(mobile terminal，MT)等。终端设备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。本公开的实施例对终端设备所采用的具体技术和具体设备形态不做限定。
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供的技术方案的限定，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，本公开实施例提供的技术方案对于类似的技术问题，同样适用。
下面结合附图对本公开所提供的信号检测方法和装置进行详细地介绍。
大规模毫米波MIMO相较于传统的MIMO技术，它的收发天线数较大，这带来了性能优势，但同时信号检测的复杂度也随之增大。
相关技术中，采用线性检测算法，如最小均方误差MMSE，MMSE是被广泛使用的经典检测方法之一。
对于MMSE，在毫米波MIMO系统中，由于信道环境复杂、信道估计误差高、以及大量非线性因素，信号检测性能往往差ML 3dB以上，对于MIMO检测中需要高阶矩阵求逆，高复杂度计算，其信号检测的性能还存在很大的提升空间。
基于此，本公开实施例中提供一种信号检测方法，以降低信号检测的网络计算复杂度。
请参见图2，图2是本公开实施例提供的一种信号检测方法的流程图。
如图2所示，该方法由网络侧设备执行，该方法可以包括但不限于如下步骤：
S21：接收终端设备的信号，进行信道估计，得到信道矩阵。
可以理解的是，MIMO系统中，在终端设备信息符号流被分离为N t路并行子流，在经过调制之后，通过N t根发射天线同时发射，以达到最大的传输速率。在网络侧设备N r根接收天线接收到信号被输入至MIMO检测器进行检测，在独立准静态瑞利平衰落信道条件下，MIMO系统的信号模型可以表示为：
y＝Hx+n
               (1)
其中，x＝[x 1，x 2，…，x Nt] T表示N t维发射向量，y＝[y 1，y 2，...，y Nr] T表示N r维接收向量，H表示N t×N r维信道矩阵，n＝[n 1，n 2，…，n Nr] T表示N r维加性白高斯噪声向量。
由于MIMO信道矩阵的作用，接收端接收到的信号是不同发射天线发射的信号的混叠。因此，MIMO检测器的作用就是在己知接收信号向量y和MIMO信道矩阵H的情况下，将混叠在一起的发射信号分 离开，恢复出发射向量x。
本公开实施例中，接收终端设备的信号，进行信道估计，得到信道矩阵H。
S22：对信道矩阵进行数据预处理，获取第一矩阵和第一向量。
在一些实施例中，对信道矩阵H进行数据预处理，为对信道矩阵H进行QR分解：
Figure PCTCN2021138021-appb-000001
本公开实施例中，对信道矩阵H进行QR分解，得到第一矩阵R和第二矩阵Q。之后根据第一矩阵R和第二矩阵Q，‖y-Hx‖ 2的表达式满足如下关系：
Figure PCTCN2021138021-appb-000002
进一步，得到第一向量z。
其中，第一向量z的表达式满足如下关系：
z＝Q 1 *y     (4)
S23：将第一矩阵和第一向量输入至训练好的深度学习模型，得到估计检测信号；其中，训练好的深度学习模型中包括N个训练好的卷积网络模型；其中，N为正整数且等于发射天线数。
本公开实施例中，在通过上述计算得到第一矩阵R和第一向量z输入至训练好的深度学习模型，以得到估计检测信号。
本公开实施例中，训练好的深度学习模型中包括N个训练好的卷积网络模型，其中，N为正整数且等于发射天线数N t。根据第一矩阵R和第一向量z，对于树搜索检测算法，采用训练好的深度学习模型替代K-best检测算法进行最优路径选择，进行最大似然检测，获取最优路径并保留路径节点，进一步阶段估计检测信号，能够降低计算复杂度。
通过实施本公开实施例，接收终端设备的信号，进行信道估计，得到信道矩阵；对信道矩阵进行数据预处理，获取第一矩阵和第一向量；将第一矩阵和第一向量输入至训练好的深度学习模型，得到估计检测信号；其中，训练好的深度学习模型中包括N个训练好的卷积网络模型；其中，N为正整数且等于发射天线数。由此，能够降低信号检测的复杂度。
请参见图3，图3是本公开实施例提供的一种信号检测方法的流程图。
如图3所示，该方法由网络侧设备执行，该方法可以包括但不限于如下步骤：
S31：接收终端设备的信号，进行信道估计，得到信道矩阵。
S32：对信道矩阵进行分解，生成第一矩阵和第二矩阵。
S33：根据第一矩阵和第二矩阵，获取第一向量。
本公开实施例中S31至S33的描述说明可以参见上述实施例中的S21至S22中的描述，此处不再赘述。
S34：将第一向量和第一矩阵输入至N个训练好的卷积网络模型，进行最大似然检测，每个训练好的卷积网络模型分别输出对应当前层的目标值K个路径节点；其中，N个训练好的卷积网络模型对应搜索树的N个层，目标值K为正整数。
本公开实施例中，将第一向量z＝Q 1 *y和第一矩阵R作为训练好的深度学习模型的输入，最大似然检测的表达式满足如下关系：
Figure PCTCN2021138021-appb-000003
可以看出，最大似然检测问题可以转化为对一个加权树的最小路径搜索问题，发射向量x的第j个元素对应树的第j层，第j层的每个节点可以由根节点到该节点的路径唯一确定。
本公开实施例中，每个训练好的卷积网络模型对应搜索树的一个层，对应输出当前层的目标值K个路径节点
Figure PCTCN2021138021-appb-000004
每个用于路径选择的训练好的卷积网络模型，包括若干个复合卷积层。第一向量z和第一矩阵R作为输入，输入维度为N t×N t，使用卷积核大小为m×m、个数为l的卷积层，激活函数采用ReLU激活函数，输出层的维度为目标值K，其中，N t，m，l，K均为正整数。
S35：汇总计算每个训练好的卷积网络模型输出的当前层的目标值K个路径节点，生成估计检测信号。
发射向量x的每个元素j被赋予路径度量PD的表达式满足如下关系：
PD(x j)＝PD(x j+1)+σ(x j)
           (6)
其中，
Figure PCTCN2021138021-appb-000005
在根节点处，当j＝N t时，σ(x j)＝0。
可以看出，从根节点沿任何路径到叶子节点，路径度量是非递减的，最大似然检测可以等价为搜索树中具有最小度量的叶子节点。
基于此，本公开实施例中，汇总计算每个训练好的卷积网络模型输出的当前层的目标值K个路径节点
Figure PCTCN2021138021-appb-000006
生成估计检测信号。由此，对信号进行检测，生成估计检测信号，在接近最大似然检测性能的条件下降低计算复杂度。
请参见图4，图4是本公开实施例提供的一种信号检测方法的流程图。
如图4所示，该方法由网络侧设备执行，该方法可以包括但不限于如下步骤：
S41：接收终端设备基于网络融合的元学习参数学习模型训练得到的中间参数。
本公开实施例中，在终端设备基于网络融合的元学习参数模型训练得到中间参数，通过发射天线数和调制阶数初始化学习目标值K的参数，基于网络融合的元学习参数模型训练得到中间参数之后发送至网络侧设备。
其中，元学习方法具有更好的参数训练效果和更快的收敛速度，本公开实施例中，使用多种表现良好的元学习网络，运用网络融合的方法融合元学习网络模型的学习结果，从而能够获得更好的性能和更强的网络泛化能力。
S42：根据中间参数，智能参数模型进行近似拟合，得到第一目标值；其中，第一目标值为正整数；根据第一目标值构建深度学习模型。
树搜索算法中的K-best检测要达到接近ML的性能，第一目标值需要达到一定值，深度学习模型中同样设置了第一目标值值。对于能取得最优路径的K值，上层节点的需求明显小于下层节点，因此取消K-best算法每层固定第一目标值的取值，将第一目标值的取值按照变化的规律设计为拟合函数形式的表达式满足如下关系：
k＝a*k b+c  (7)
其中a，b，c均为可学习的参数，第一目标值k为路径节点层数。
本公开实施例中，智能参数模型进行近似拟合，得到第一目标值，为变化的第一目标值k设计了合 适的近似拟合函数，智能参数模型部署在网络侧设备，之后将得到正整数第一目标值反馈给深度学习模型，参与深度模型的构建，在增加检测网络性能同时获得较低的计算复杂度。
需要说明的是，本公开实施例中S51与S52可以单独实施，也可以结合本公开实施例中的任何一个其他步骤一起被实施，例如结合本公开实施例中的S31至S33和/或S41至S44一起被实施，本公开实施例并不对此做出限定。
请参见图5，图5是本公开实施例提供的一种信号检测方法的流程图。
如图5所示，该方法由网络侧设备执行，该方法可以包括但不限于如下步骤：
S51：获取训练数据集；其中，训练数据集中包括至少一组训练发射信号和训练接收信号。
本公开实施例中，根据信道估计得到的信道矩阵H，模拟生成至少一组训练发射信号和训练接收信号，得到训练数据集。
S52：将训练数据集输入至深度学习模型，对深度学习模型进行训练，生成训练好的深度学习模型。
在一些实施例中，将训练发射信号输入至深度学习模型，得到预测信号；根据预测信号和训练接收信号，对深度学习模型进行更新，生成训练好的深度学习模型。
本公开实施例中，根据信道估计得到的信道矩阵H，模拟生成至少一组训练发射信号和训练接收信号，得到训练数据集之外，还包括验证数据集和测试数据集，可以理解的是，验证数据集中包括至少一组验证发射信号和验证接收信号，测试数据集中包括至少一组测试发射信号和测试接收信号。
在训练数据集对深度学习模型训练，得到训练好的深度学习模型之后，使用验证数据集对训练好的深度学习模型进行验证，以验证训练好的深度学习模型的信号检测效果，并进一步的，使用测试数据集对训练好的深度学习模型进行测试，以测试训练好的深度学习模型的信号检测效果是否达到预期的效果。
在一些实施例中，训练好的深度学习模型中包括目标值K，所述方法，还包括：
在对深度学习模型进行更新的过程中，对第一目标值进行更新，得到目标值K。
本公开实施例中，将在终端设备基于网络融合的元学习参数学习模型训练得到的中间参数，经过智能参数模型进行近似拟合，得到第一目标值，参与到深度学习模型的构建，之后在深度学习模型训练的过程中，对第一目标值进行联合训练，得到目标值K，使得深度学习模型能够学习到全局最优。
本公开实施例中，在对深度学习模型进行训练之后，将包括目标值K的训练好的深度学习模型部署在网络设备。
请参见图6，图6是本公开实施例提供的一种信号检测方法的流程图。
如图6所示，该方法由网络侧设备执行，该方法可以包括但不限于如下步骤：
S61：响应于预设条件被满足，发送模型更新指令；其中，预设条件为信道状态信息变化超出一定范围。
可以理解的是，深度学习模型在线收集足够的训练数据时可能会经历严重的延迟，特别是在低信噪比状态下，这为深度学习模型的训练和部署带来了困难。
本公开实施例中，响应于预设条件被满足，可以为在网络侧设备检测到信道状态信息发生变化，对训练好的深度学习模型在当前信道环境下的性能进行评估，在性能显著降低时，发送模型更新指令，发送至终端设备。
本公开实施例中，为了应对复杂多变的信道环境，设计网络在线更新流程，在预设条件被满足，需要对训练好的深度学习模型进行更新时，发送模型更新指令至终端设备，从而充分利用元学习在获得训 练数据和快速更新方面的优势，在终端设备处基于网络融合的元学习参数学习模型训练得到更新中间参数。
在一些实施例中，网络侧设备接收终端设备基于网络融合的元学习参数学习模型训练得到的更新中间参数；根据更新中间参数，智能参数模型进行近似拟合，得到第二目标值；其中，第二目标值为正整数；根据第二目标值构建待更新深度学习模型。
在一些实施例中，获取更新训练数据集；其中，更新训练数据集中包括至少一组更新训练发射信号和更新训练接收信号；将更新训练数据集输入至待更新深度学习模型，对待更新深度学习模型进行训练，生成更新后的训练好的深度学习模型。
本公开实施例中，在对训练好的深度学习模型进行更新的过程中，根据第二目标值构建待更新深度学习模型，之后获取更新训练数据集；其中，更新训练数据集中包括至少一组更新训练发射信号和更新训练接收信号；将更新训练数据集输入至待更新深度学习模型，对待更新深度学习模型进行训练，生成更新后的训练好的深度学习模型的方式与构建深度学习模型，以及对深度学习模型进行训练的过程相似，可以参见上述实施例的相关描述，此处不再赘述。
请参见图7，图7是本公开实施例提供的一种信号检测方法的流程图。
如图7所示，该方法由终端设备执行，该方法可以包括但不限于如下步骤：
S71：发送信号和中间参数；其中，基于网络融合的元学习参数学习模型训练得到中间参数。
本公开实施例中，MIMO系统，为了得到最大的传输速率，在终端设备，信息符号流分离为N t路并行子流，在经过调制后，通过N t根发射天线同时发送，终端设备发送信号。
本公开实施例中，终端设备部署有网络融合的元学习参数学习模型，元学习方法能够在获取已有“知识”的基础上快速学习新的任务，意图在于通过少量的训练实例设计能够快速学习或适应新环境的网络模型，具有学会学习的能力。
在一些实施例中，根据发射天线数和调制阶数，初始化至少一个第一参数，将第一参数输入至网络融合的元学习参数学习模型，进行训练得到的中间参数，并发送中间参数至网络侧设备。
本公开实施例中，根据发射天线数和调制阶数初始化至少一个第一参数，示例性的，初始化参数a，参数b，参数c。其中，参数a，参数b，参数c均为可学习的参数。
将可学习的参数a,b,c合并为网络融合的元学习参数学习模型的输入X的表达式满足如下关系：
X＝(a,b,c)
            (8)在一些实施例中，本公开实施例中的网络融合的元学习参数学习模型包括基于预测梯度的元学习模型、基于LSTM的元学习模型和网络融合模型。
在一些实施例中，将第一参数输入至基于预测梯度的元学习模型，生成第一中间参数。将第一参数输入至基于LSTM的元学习模型，生成第二中间参数。将第一中间值和第二中间值输入至网络融合模型，生成中间参数。
其中，基于预测梯度的元学习模型的结构图如图8所示。其中，θ为传递的元学习网络状态信息，
Figure PCTCN2021138021-appb-000007
为传递的预测梯度信息。
本公开实施例中，基于预测梯度的元学习模型通过训练一个通用的神经网络来预测梯度，通过一元二次方程的回归问题来训练，得到的神经网络优化器梯度下降更快更准，从而实现了快速学习。
其中，基于LSTM的元学习模型其结构图如图9所示。LSTM是一种特殊的循环神经网络，通过门结构对细胞状态C进行删除或者添加信息。通过LSTM网络训练可学习参数X，输入当前可学习参数，直接输出新的更新结果。
本公开实施例中，基于预测梯度的元学习模型和基于LSTM的元学习模型均表现出良好的性能，设计网络融合模型，将不同结构的元学习网络的学习结果融合，从而能够取得比多个精细调优的模型更好的结果。网络融合模型的表达式满足如下关系：
Y＝W 1Y 1+W 2Y 2  (9)
其中，Y 1为基于预测梯度的元学习模型进行参数学习的输出结果，Y 2为基于LSTM的元学习模型进行参数学习的输出结果，W 1与W 2是可学习的标量，可以通过网络学习W 1与W 2。
本公开实施例中，通过网络融合模型使用网络融合学习方法，在提高网络的性能的同时增强了网络泛化能力，增强了应对信道变化的能力。
请参见图10，图10是本公开实施例提供的一种信号检测方法的流程图。
如图10所示，该方法由终端设备执行，该方法可以包括但不限于如下步骤：
S101：接收网络侧设备的模型更新指令。
S102：根据发射天线数和调制阶数，重新初始化至少一个第二参数。
S103：将第二参数输入至网络融合的元学习参数学习模型，进行训练得到的更新中间参数。
S104：发送更新中间参数。
本公开实施例中，终端设备接收网络侧设备发送的模型更新指令，重新初始化至少一个第二参数，之后将第二参数输入至网络融合的元学习参数学习模型，进行训练得到的更新中间参数的方法，与上述实施例中根据发射天线数和调制阶数，初始化至少一个第一参数；将第一参数输入至网络融合的元学习参数学习模型，进行训练得到的中间参数，发送中间参数的过程相似，具体可以参见上述实施例的相关描述，此处不再赘述。
本公开实施例中，在对训练好的深度学习模型进行更新时，向终端设备发送模型更新指令，在终端设备侧，在根据发射天线数和调制阶数，重新初始化至少一个第二参数，利用网络融合的元学习参数学习模型，重新训练得到更新中间参数，能够在发挥元学习方法的同时，利用以往的知识经验来指导新任务的学习，通过少数训练实例快速更新深度学习模型。
上述本公开提供的实施例中，分别从网络侧设备、终端设备的角度对本公开实施例提供的方法进行了介绍。为了实现上述本公开实施例提供的方法中的各功能，网络侧设备和终端设备可以包括硬件结构、软件模块，以硬件结构、软件模块、或硬件结构加软件模块的形式来实现上述各功能。上述各功能中的某个功能可以以硬件结构、软件模块、或者硬件结构加软件模块的方式来执行。
请参见图11，为本公开实施例提供的一种通信装置1的结构示意图。图11所示的通信装置1可包括收发模块11和处理模块12。收发模块11可包括发送模块和/或接收模块，发送模块用于实现发送功能，接收模块用于实现接收功能，收发模块11可以实现发送功能和/或接收功能。
通信装置1可以是网络侧设备，也可以是网络侧设备中的装置，还可以是能够与网络侧设备匹配使用的装置。或者，通信装置1可以是终端设备，也可以是终端设备中的装置，还可以是能够与终端设备匹配使用的装置。
通信装置1为网络侧设备：收发模块11，用于接收终端设备的信号。
处理模块12，用于进行信道估计，得到信道矩阵；对所述信道矩阵进行数据预处理，获取第一矩 阵和第一向量；将所述第一矩阵和所述第一向量输入至训练好的深度学习模型，得到估计检测信号；其中，所述训练好的深度学习模型中包括N个训练好的卷积网络模型；其中，N为正整数且等于发射天线数。
通信装置1为终端设备：收发模块11，用于发送信号和中间参数；其中，基于网络融合的元学习参数学习模型训练得到中间参数。
请参见图12，图12是本公开实施例提供的另一种通信装置1000的结构示意图。通信装置1000可以是网络侧设备，也可以是终端设备，也可以是支持网络侧设备实现上述方法的芯片、芯片系统、或处理器等，还可以是支持终端设备实现上述方法的芯片、芯片系统、或处理器等。该通信装置1000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。
通信装置1000可以是网络侧设备，也可以是终端设备，也可以是支持网络侧设备实现上述方法的芯片、芯片系统、或处理器等，还可以是支持终端设备实现上述方法的芯片、芯片系统、或处理器等。该装置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。
通信装置1000可以包括一个或多个处理器1001。处理器1001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，执行计算机程序，处理计算机程序的数据。
可选的，通信装置1000中还可以包括一个或多个存储器1002，其上可以存有计算机程序1004，存储器1002执行所述计算机程序1004，以使得通信装置1000执行上述方法实施例中描述的方法。可选的，所述存储器1002中还可以存储有数据。通信装置1000和存储器1002可以单独设置，也可以集成在一起。
可选的，通信装置1000还可以包括收发器1005、天线1006。收发器1005可以称为收发单元、收发机、或收发电路等，用于实现收发功能。收发器1005可以包括接收器和发送器，接收器可以称为接收机或接收电路等，用于实现接收功能；发送器可以称为发送机或发送电路等，用于实现发送功能。
可选的，通信装置1000中还可以包括一个或多个接口电路1007。接口电路1007用于接收代码指令并传输至处理器1001。处理器1001运行所述代码指令以使通信装置1000执行上述方法实施例中描述的方法。
通信装置1000为网络侧设备：收发器1005用于执行图2中的S21；图3中的S31；图4中的S41；图5中的S51；图6中的S61；处理器1001用于执行图2中的S22、S23；图3中的S32、S33、S34、S35；图4中的S42；图5中的S52。
通信装置1000为终端设备：收发器1005用于执行图7中的S71；图10中的S101；处理器1001用于执行图10中的S102、S103、S104。
在一种实现方式中，处理器1001中可以包括用于实现接收和发送功能的收发器。例如该收发器可以是收发电路，或者是接口，或者是接口电路。用于实现接收和发送功能的收发电路、接口或接口电路可以是分开的，也可以集成在一起。上述收发电路、接口或接口电路可以用于代码/数据的读写，或者，上述收发电路、接口或接口电路可以用于信号的传输或传递。
在一种实现方式中，处理器1001可以存有计算机程序1003，计算机程序1003在处理器1001上运行，可使得通信装置1000执行上述方法实施例中描述的方法。计算机程序1003可能固化在处理器1001中，该种情况下，处理器1001可能由硬件实现。
在一种实现方式中，通信装置1000可以包括电路，所述电路可以实现前述方法实施例中发送或接 收或者通信的功能。本公开中描述的处理器和收发器可实现在集成电路(integrated circuit，IC)、模拟IC、射频集成电路RFIC、混合信号IC、专用集成电路(application specific integrated circuit，ASIC)、印刷电路板(printed circuit board，PCB)、电子设备等上。该处理器和收发器也可以用各种IC工艺技术来制造，例如互补金属氧化物半导体(complementary metal oxide semiconductor，CMOS)、N型金属氧化物半导体(nMetal-oxide-semiconductor，NMOS)、P型金属氧化物半导体(positive channel metal oxide semiconductor，PMOS)、双极结型晶体管(bipolar junction transistor，BJT)、双极CMOS(BiCMOS)、硅锗(SiGe)、砷化镓(GaAs)等。
以上实施例描述中的通信装置可以是终端设备，但本公开中描述的通信装置的范围并不限于此，而且通信装置的结构可以不受图12的限制。通信装置可以是独立的设备或者可以是较大设备的一部分。例如所述通信装置可以是：
(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，可选的，该IC集合也可以包括用于存储数据，计算机程序的存储部件；
(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。
对于通信装置可以是芯片或芯片系统的情况，请参见图13，为本公开实施例中提供的一种芯片的结构图。
芯片1100包括处理器1101和接口1103。其中，处理器1101的数量可以是一个或多个，接口1103的数量可以是多个。
对于芯片用于实现本公开实施例中网络侧设备的功能的情况：
接口1103，用于接收代码指令并传输至所述处理器。
处理器1101，用于运行代码指令以执行如上面一些实施例所述的信号检测方法。
对于芯片用于实现本公开实施例中终端设备的功能的情况：
接口1103，用于接收代码指令并传输至所述处理器。
处理器1101，用于运行代码指令以执行如上面一些实施例所述的信号检测方法。
可选的，芯片1100还包括存储器1102，存储器1102用于存储必要的计算机程序和数据。
本领域技术人员还可以了解到本公开实施例列出的各种说明性逻辑块(illustrative logical block)和步骤(step)可以通过电子硬件、电脑软件，或两者的结合进行实现。这样的功能是通过硬件还是软件来实现取决于特定的应用和整个系统的设计要求。本领域技术人员可以对于每种特定的应用，可以使用各种方法实现所述的功能，但这种实现不应被理解为超出本公开实施例保护的范围。
本公开实施例还提供一种信号检测系统，该系统包括前述图11实施例中作为网络侧设备的通信装置和作为终端设备的通信装置，或者，该系统包括前述图12实施例中作为网络侧设备的通信装置和作为终端设备的通信装置。
本公开还提供一种可读存储介质，其上存储有指令，该指令被计算机执行时实现上述任一方法实施例的功能。
本公开还提供一种计算机程序产品，该计算机程序产品被计算机执行时实现上述任一方法实施例的 功能。
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理解：本公开中涉及的第一、第二等各种数字编号仅为描述方便进行的区分，并不用来限制本公开实施例的范围，也表示先后顺序。
本公开中的至少一个还可以描述为一个或多个，多个可以是两个、三个、四个或者更多个，本公开不做限制。在本公开实施例中，对于一种技术特征，通过“第一”、“第二”、“第三”、“A”、“B”、“C”和“D”等区分该种技术特征中的技术特征，该“第一”、“第二”、“第三”、“A”、“B”、“C”和“D”描述的技术特征间无先后顺序或者大小顺序。
本公开中各表所示的对应关系可以被配置，也可以是预定义的。各表中的信息的取值仅仅是举例，可以配置为其他值，本公开并不限定。在配置信息与各参数的对应关系时，并不一定要求必须配置各表中示意出的所有对应关系。例如，本公开中的表格中，某些行示出的对应关系也可以不配置。又例如，可以基于上述表格做适当的变形调整，例如，拆分，合并等等。上述各表中标题示出参数的名称也可以采用通信装置可理解的其他名称，其参数的取值或表示方式也可以通信装置可理解的其他取值或表示方式。上述各表在实现时，也可以采用其他的数据结构，例如可以采用数组、队列、容器、栈、线性表、指针、链表、树、图、结构体、类、堆、散列表或哈希表等。
本公开中的预定义可以理解为定义、预先定义、存储、预存储、预协商、预配置、固化、或预烧制。
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 (22)

  1. 一种信号检测方法，其特征在于，所述方法由网络侧设备执行，包括：
    接收终端设备的信号，进行信道估计，得到信道矩阵；
    对所述信道矩阵进行数据预处理，获取第一矩阵和第一向量；
    将所述第一矩阵和所述第一向量输入至训练好的深度学习模型，得到估计检测信号；其中，所述训练好的深度学习模型中包括N个训练好的卷积网络模型；其中，N为正整数且等于发射天线数。
  2. 根据权利要求1所述的方法，其特征在于，所述对所述信道矩阵进行数据预处理，获取第一矩阵和第一向量，包括：
    对所述信道矩阵进行分解，生成所述第一矩阵和第二矩阵；
    根据所述第一矩阵和所述第二矩阵，获取所述第一向量。
  3. 根据权利要求1或2所述的方法，其特征在于，所述将所述第一矩阵和所述第一向量输入至训练好的深度学习模型，生成估计检测信号，包括：
    将所述第一向量和所述第一矩阵输入至N个训练好的卷积网络模型，进行最大似然检测，每个训练好的卷积网络模型分别输出对应当前层的目标值K个路径节点；其中，N个训练好的卷积网络模型对应搜索树的N个层，目标值K为正整数；
    汇总计算每个训练好的卷积网络模型输出的当前层的目标值K个路径节点，生成所述估计检测信号。
  4. 根据权利要求1至3中任一项所述的方法，其特征在于，所述方法，还包括：
    接收终端设备基于网络融合的元学习参数学习模型训练得到的中间参数；
    根据所述中间参数，智能参数模型进行近似拟合，得到第一目标值；其中，所述第一目标值为正整数；
    根据所述第一目标值构建深度学习模型。
  5. 根据权利要求1至4中任一项所述的方法，其特征在于，所述方法，还包括：
    获取训练数据集；其中，所述训练数据集中包括至少一组训练发射信号和训练接收信号；
    将所述训练数据集输入至深度学习模型，对深度学习模型进行训练，生成训练好的深度学习模型。
  6. 根据权利要求5所述的方法，其特征在于，所述将所述训练数据集输入至深度学习模型，对深度学习模型进行训练，生成训练好的深度学习模型，包括：
    将所述训练发射信号输入至深度学习模型，得到预测信号；
    根据所述预测信号和所述训练接收信号，对深度学习模型进行更新，生成训练好的深度学习模型。
  7. 根据权利要求6所述的方法，其特征在于，所述训练好的深度学习模型中包括目标值K，所述方法，还包括：
    在对深度学习模型进行更新的过程中，对所述第一目标值进行更新，得到所述目标值K。
  8. 根据权利要求1至7中任一项所述的方法，其特征在于，所述方法，还包括：
    响应于预设条件被满足，发送模型更新指令；其中，所述预设条件为信道状态信息变化超出一定范围。
  9. 根据权利要求8所述的方法，其特征在于，所述方法，还包括：
    接收终端设备基于网络融合的元学习参数学习模型训练得到的更新中间参数；
    根据所述更新中间参数，智能参数模型进行近似拟合，得到第二目标值；其中，所述第二目标值为正整数；
    根据所述第二目标值构建待更新深度学习模型。
  10. 根据权利要求9所述的方法，其特征在于，所述方法，还包括：
    获取更新训练数据集；其中，所述更新训练数据集中包括至少一组更新训练发射信号和更新训练接收信号；
    将所述更新训练数据集输入至待更新深度学习模型，对待更新深度学习模型进行训练，生成更新后的训练好的深度学习模型。
  11. 一种信号检测方法，其特征在于，所述方法由终端设备执行，包括：
    发送信号和中间参数；其中，基于网络融合的元学习参数学习模型训练得到所述中间参数。
  12. 根据权利要求11所述的方法，其特征在于，所述方法，还包括：
    根据发射天线数和调制阶数，初始化至少一个第一参数；
    将所述第一参数输入至网络融合的元学习参数学习模型，进行训练得到的中间参数。
  13. 根据权利要求11所述的方法，其特征在于，所述将所述可学习参数输入至网络融合的元学习参数学习模型，进行训练得到的中间值k，包括：
    将所述第一参数输入至基于预测梯度的元学习模型，生成第一中间参数；
    将所述第一参数输入至基于LSTM的元学习模型，生成第二中间参数；
    将所述第一中间值和所述第二中间值输入至网络融合模型，生成所述中间参数。
  14. 根据权利要求11至13中任一项所述的方法，其特征在于，所述方法，还包括：
    接收网络侧设备的模型更新指令；
    根据发射天线数和调制阶数，重新初始化至少一个第二参数；
    将所述第二参数输入至网络融合的元学习参数学习模型，进行训练得到的更新中间参数；
    发送所述更新中间参数。
  15. 一种通信装置，其特征在于，包括：
    收发模块，用于接收终端设备的信号；
    处理模块，用于进行信道估计，得到信道矩阵；对所述信道矩阵进行数据预处理，获取第一矩阵和 第一向量；将所述第一矩阵和所述第一向量输入至训练好的深度学习模型，得到估计检测信号；其中，所述训练好的深度学习模型中包括N个训练好的卷积网络模型；其中，N为正整数且等于发射天线数。
  16. 一种通信装置，其特征在于，包括：
    收发模块，用于发送信号和中间参数。
  17. 一种通信装置，其特征在于，所述装置包括处理器和存储器，所述存储器中存储有计算机程序，所述处理器执行所述存储器中存储的计算机程序，以使所述装置执行如权利要求1至10中任一项所述的方法。
  18. 一种通信装置，其特征在于，所述装置包括处理器和存储器，所述存储器中存储有计算机程序，所述处理器执行所述存储器中存储的计算机程序，以使所述装置执行如权利要求11至14中任一项所述的方法。
  19. 一种通信装置，其特征在于，包括：处理器和接口电路；
    所述接口电路，用于接收代码指令并传输至所述处理器；
    所述处理器，用于运行所述代码指令以执行如权利要求1至10中任一项所述的方法。
  20. 一种通信装置，其特征在于，包括：处理器和接口电路；
    所述接口电路，用于接收代码指令并传输至所述处理器；
    所述处理器，用于运行所述代码指令以执行如权利要求11至14中任一项所述的方法。
  21. 一种计算机可读存储介质，用于存储有指令，当所述指令被执行时，使如权利要求1至10中任一项所述的方法被实现。
  22. 一种计算机可读存储介质，用于存储有指令，当所述指令被执行时，使如权利要求11至14中任一项所述的方法被实现。
PCT/CN2021/138021 2021-12-14 2021-12-14 信号检测方法和装置 WO2023108436A1 (zh)

Priority Applications (1)

Application Number Priority Date Filing Date Title
PCT/CN2021/138021 WO2023108436A1 (zh) 2021-12-14 2021-12-14 信号检测方法和装置

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
PCT/CN2021/138021 WO2023108436A1 (zh) 2021-12-14 2021-12-14 信号检测方法和装置

Publications (1)

Publication Number Publication Date
WO2023108436A1 true WO2023108436A1 (zh) 2023-06-22

Family

ID=86774932

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2021/138021 WO2023108436A1 (zh) 2021-12-14 2021-12-14 信号检测方法和装置

Country Status (1)

Country Link
WO (1) WO2023108436A1 (zh)

Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101674160A (zh) * 2009-10-22 2010-03-17 复旦大学 多输入多输出无线通信系统信号检测方法及装置
CN103532609A (zh) * 2013-10-08 2014-01-22 上海交通大学 基于K-Best算法的MIMO信号检测方法及检测器
CN112261685A (zh) * 2020-10-15 2021-01-22 东南大学 一种基于卷积神经网络的mimo信号检测方法

Patent Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101674160A (zh) * 2009-10-22 2010-03-17 复旦大学 多输入多输出无线通信系统信号检测方法及装置
CN103532609A (zh) * 2013-10-08 2014-01-22 上海交通大学 基于K-Best算法的MIMO信号检测方法及检测器
CN112261685A (zh) * 2020-10-15 2021-01-22 东南大学 一种基于卷积神经网络的mimo信号检测方法

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
ZHANG JING; HE YUNFENG; LI YU-WEN; WEN CHAO-KAI; JIN SHI: "Meta Learning-Based MIMO Detectors: Design, Simulation, and Experimental Test", IEEE TRANSACTIONS ON WIRELESS COMMUNICATIONS, IEEE SERVICE CENTER, PISCATAWAY, NJ., US, vol. 20, no. 2, 20 October 2020 (2020-10-20), US , pages 1122 - 1137, XP011837367, ISSN: 1536-1276, DOI: 10.1109/TWC.2020.3030882 *

Similar Documents

Publication Publication Date Title
US20230342593A1 (en) Neural network training method and related apparatus
US8831128B2 (en) MIMO communication system signal detection method
WO2023070586A1 (zh) 物理下行共享信道pdsch配置方法及装置
WO2023108436A1 (zh) 信号检测方法和装置
WO2023019596A1 (zh) 基于轨道角动量oam的通信方法及其装置
CN115997354A (zh) 波束指示方法及装置
US10872045B1 (en) Method and apparatus for symbol detection
CN111510187B (zh) 用于接收信号的检测的装置和方法
WO2020087293A1 (zh) 一种通信接收机及信号处理的方法
WO2023019585A1 (zh) 预编码模型训练方法、预编码方法和装置
WO2024036631A1 (zh) 一种信息反馈方法、装置、设备及存储介质
US20240049188A1 (en) Gradient transmission method and related apparatus
WO2022262687A1 (zh) 一种数据处理方法及装置
WO2023201668A1 (zh) 导频位置确定模型的获取、导频位置确定方法及其装置
WO2023150918A1 (zh) 一种波束管理的方法及其装置
WO2023115579A1 (zh) 一种人工智能ai服务提供方法和装置
WO2024011543A1 (zh) 基向量的选择指示上报方法和装置
WO2022213326A1 (zh) 一种数据接收方法及装置
WO2023245683A1 (zh) 基向量类型的指示方法和装置
EP4220484A1 (en) Communication method, apparatus and system
WO2024065547A1 (zh) 指示上报方法及其装置
WO2023216873A1 (zh) 一种信号传输方法及装置
WO2023283785A1 (zh) 信号处理的方法及接收机
WO2023050151A1 (zh) 资源配置方法、装置及存储介质
WO2023274101A1 (zh) 一种获得预编码矩阵的方法及装置

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: 21967576

Country of ref document: EP

Kind code of ref document: A1