Channel estimation method, channel estimation model training method and device

Abstract

The disclosure provides a channel estimation method, a training method and device of a channel estimation model, wherein the method comprises: performing channel estimation based on a received value of a reference signal to obtain a first channel estimation result of a first resource position occupied by the reference signal in a target channel; obtaining a first target fusion feature of the first resource position based on the first channel estimation result and the reference signal; and determining a second channel estimation result of a second resource position based on the first target fusion feature, wherein the second resource position is a resource position in the target channel that is not occupied by the reference signal. The algorithm amount of channel estimation of the second resource position is reduced, the complexity of the channel estimation algorithm is reduced, the efficiency and accuracy of the channel estimation are improved, the algorithm power consumption of the channel estimation is reduced, the compatibility of the channel estimation method for different types of channels is improved, and the scene adaptability and practicality of the channel estimation method are improved.

Description

Technical Field

[0001] This disclosure relates to the field of communication technology, and in particular to a channel estimation method, a training method and apparatus for a channel estimation model. Background Technology

[0002] In related technologies, people can rely on the results of channel estimation to perform equalization compensation and coherent demodulation on the signal received at the receiver, thereby realizing signal restoration at the receiver. Therefore, how to perform accurate channel estimation is very important. Summary of the Invention

[0003] This disclosure aims to at least partially address one of the technical problems in the related art.

[0004] Therefore, the first aspect of this disclosure proposes a channel estimation method.

[0005] The second aspect of this disclosure proposes a training method for a channel estimation model.

[0006] The third aspect of this disclosure proposes a channel estimation apparatus.

[0007] The fourth aspect of this disclosure proposes a training apparatus for a channel estimation model.

[0008] The fifth aspect of this disclosure proposes an electronic device.

[0009] The sixth aspect of this disclosure proposes a computer-readable storage medium.

[0010] The seventh aspect of this disclosure provides for a chip.

[0011] The first aspect of this disclosure proposes a channel estimation method, comprising: performing channel estimation based on the received value of a reference signal to obtain a first channel estimation result of a first resource location occupied by the reference signal in a target channel; obtaining a first target fusion feature of the first resource location based on the first channel estimation result and the reference signal; and determining a second channel estimation result of a second resource location based on the first target fusion feature, wherein the second resource location is a resource location in the target channel that is not occupied by the reference signal.

[0012] The second aspect of this disclosure proposes a training method for a channel estimation model, comprising: obtaining first model parameters of an initial channel estimation model using a candidate parameter calculation model to be trained and sample metadata of a sample signal; assigning values ​​to the initial channel estimation model based on the first model parameters to obtain a candidate channel estimation model to be trained; and jointly training the candidate channel estimation model and the candidate parameter calculation model based on the sample signal to obtain a trained target channel estimation model and a target parameter calculation model, wherein the target channel estimation model is used to implement the channel estimation method proposed in the first aspect above.

[0013] A third aspect of this disclosure provides a channel estimation apparatus, comprising: a first estimation module, configured to perform channel estimation based on received values ​​of a reference signal to obtain a first channel estimation result of a first resource location occupied by the reference signal in a target channel; a feature fusion module, configured to obtain a first target fusion feature of the first resource location based on the first channel estimation result and the reference signal; and a second estimation module, configured to determine a second channel estimation result of a second resource location based on the first target fusion feature, wherein the second resource location is a resource location in the target channel not occupied by the reference signal.

[0014] This disclosure provides a fourth aspect of a channel estimation model training apparatus, comprising: an acquisition module for obtaining first model parameters of an initial channel estimation model using candidate parameter calculation models to be trained and sample metadata of sample signals; an assignment module for assigning values ​​to the initial channel estimation model based on the first model parameters to obtain candidate channel estimation models to be trained; and a training module for jointly training the candidate channel estimation model and the candidate parameter calculation model based on the sample signals to obtain a trained target channel estimation model and a target parameter calculation model, wherein the target channel estimation model is used to implement the channel estimation apparatus proposed in the third aspect above.

[0015] This disclosure provides a fifth aspect of an electronic device, comprising: a processor; and a memory for storing executable instructions of the processor; wherein the processor is configured to execute instructions to implement the channel estimation method as described in the first aspect and / or the training method for the channel estimation model as described in the second aspect.

[0016] The sixth aspect of this disclosure provides a computer-readable storage medium, wherein when the instructions in the computer-readable storage medium are executed by a processor of an electronic device, the electronic device is able to perform the channel estimation method as described in the first aspect and / or the training method of the channel estimation model as described in the second aspect.

[0017] A seventh aspect of this disclosure provides a chip including one or more interface circuits and one or more processors; the interface circuits are configured to receive signals and send the signals to the processors, the signals including computer instructions stored in a memory, wherein when the processor executes the computer instructions, the chip causes the chip to perform the channel estimation method as described in the first aspect and / or the channel estimation model training method as described in the second aspect.

[0018] The channel estimation method, training method, and apparatus proposed in this disclosure obtain a first target fusion feature based on the first channel estimation result of the first resource location occupied by the reference signal, and then determine the second channel estimation result of the second resource location based on the first target fusion feature. This reduces the algorithmic complexity of the channel estimation for the second resource location, shortens the processing latency of the channel estimation, and improves the efficiency and accuracy of the channel estimation. It eliminates the need for pilot configuration for each resource unit in the resource location to be estimated, thus reducing the computational overhead of the channel estimation. It also eliminates the need to limit the reference signal, improving the compatibility of the channel estimation method with different types of channels, thereby improving the scenario adaptability and practicality of the channel estimation method.

[0019] The training process improves the robustness and stability of the trained target parameter calculation model and target channel estimation model, as well as the accuracy and quality of their outputs. Joint training of the parameter calculation model and channel estimation model enhances the adaptability of the model parameters output by the target parameter calculation model to the parameter assignment requirements of the target channel estimation model. Dynamic adjustment of the model parameters of the target channel estimation model is achieved, improving the compatibility and matching degree of the target channel estimation model for different types of channels. Furthermore, it enhances the accuracy, precision, and efficiency of the target channel estimation model when estimating different types of channels. This eliminates the need for separate resource development for different types of channels, reducing the model deployment cost of channel estimation, saving resource overhead, and optimizing the channel estimation method.

[0020] It should be understood that the description herein is not intended to identify key or essential features of the embodiments thereof, nor is it intended to limit the scope of this disclosure. Other features of this disclosure will become readily apparent from the following description. Attached Figure Description

[0021] The above and / or additional aspects and advantages of this disclosure will become apparent and readily understood from the following description of the embodiments taken in conjunction with the accompanying drawings, in which: Figure 1 This is a schematic flowchart of a channel estimation method according to an embodiment of the present disclosure; Figure 2This is a schematic flowchart of a channel estimation method according to another embodiment of the present disclosure; Figure 3 This is a flowchart illustrating a training method for a channel estimation model according to an embodiment of the present disclosure. Figure 4 This is a flowchart illustrating a training method for a channel estimation model according to another embodiment of the present disclosure. Figure 5 This is a schematic diagram of a channel estimation apparatus according to an embodiment of the present disclosure; Figure 6 This is a schematic diagram of a training apparatus for a channel estimation model according to an embodiment of the present disclosure; Figure 7 This is a schematic diagram of the structure of an electronic device according to an embodiment of the present disclosure; Figure 8 This is a schematic diagram of the structure of a chip according to an embodiment of the present disclosure. Detailed Implementation

[0022] Embodiments of this disclosure are described in detail below. Examples of these embodiments are illustrated in the accompanying drawings, wherein the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and intended to explain this disclosure, and should not be construed as limiting this disclosure.

[0023] The following describes, with reference to the accompanying drawings, a channel estimation method, a channel estimation model training method, and an apparatus proposed in the embodiments of this disclosure.

[0024] It should be noted that the acquisition, storage, use, and processing of data in this disclosed technical solution comply with relevant laws and regulations and do not violate public order and good morals. It should also be noted that all data processed in this disclosure is explicitly authorized by users or relevant parties, and has undergone de-identification or anonymization processing before collection and use, and does not contain any personally identifiable information or user privacy content; all data is used solely for the purpose of channel estimation methods and / or channel estimation model training methods, ensuring that while achieving technical effects, data security and user privacy rights are fully protected.

[0025] Figure 1 This is a flowchart illustrating a channel estimation method according to an embodiment of the present disclosure, as shown below. Figure 1 As shown, the method includes: S101, based on the received value of the reference signal, perform channel estimation to obtain the first channel estimation result of the first resource location occupied by the reference signal in the target channel.

[0026] In this embodiment of the disclosure, the channel for which channel estimation is required can be determined as the target channel. A reference signal is transmitted on the target channel. The reference signal can be a demodulation reference signal (DMRS), a cell-specific reference signal (LTE CRS), a user-specific reference signal (UE-Specific Reference Signal (LTE URS), or other types of reference signals required for signal estimation. No specific limitation is made here.

[0027] In this embodiment of the disclosure, the resource location occupied by the reference signal in the target channel can be determined as the first resource location. In this scenario, the reference signal can be processed by the signal reception value acquisition algorithm in the related technology. Based on the result of the algorithm processing, the received value of the reference signal is obtained. Then, the received value of the reference signal can be processed by the channel estimation algorithm in the related technology to obtain the channel estimation result of the first resource location, which is used as the first channel estimation result.

[0028] S102, based on the first channel estimation result and the reference signal, the first target fusion feature of the first resource location is obtained.

[0029] In this embodiment of the disclosure, the first channel estimation result and the reference signal can be processed by the feature extraction algorithm in the related art, and the features of the first channel estimation result and the features of the reference signal can be obtained based on the result of the algorithm processing.

[0030] In this scenario, feature fusion algorithms from related technologies can be used to process the features of the first channel estimation result and the features of the reference signal. Based on the result of the algorithm processing, the fused features of the two can be obtained, and these fused features can be used as the first target fused features of the first resource location.

[0031] S103, based on the first target fusion features and the target signal at the second resource location, determine the second channel estimation result of the second resource location, wherein the second resource location is a resource location in the target channel that is not occupied by the reference signal.

[0032] In this embodiment of the disclosure, there is also a second resource location in the target channel. The resource location that the reference signal does not occupy in the target channel can be determined as the second resource location. In this scenario, the channel estimation of the second resource location can be performed based on the first channel estimation result of the first resource location, and the obtained channel estimation result of the second resource location can be determined as the second channel estimation result.

[0033] In this embodiment of the disclosure, the first target fusion feature of the first resource location carries feature information of the channel estimation result of the first resource location and feature information of the reference signal at the first resource location. In this scenario, the first target fusion feature can be processed by the analysis algorithm in the related technology. Based on the result of the algorithm processing, the relevant information required for channel estimation of the second resource location can be extracted from the first target fusion feature. Then, based on the information extracted in this part, the channel estimation algorithm in the related technology is used to perform channel estimation of the second resource location, thereby obtaining the channel estimation result of the second resource location, that is, the second channel estimation result.

[0034] It should be noted that this disclosure does not impose any restrictions on the execution sequence of steps S101 to S103. Figure 1 The example only demonstrates the sequential execution of steps S101 to S103.

[0035] The channel estimation method proposed in this disclosure obtains a first target fusion feature based on the first channel estimation result of the first resource location occupied by the reference signal, and then determines the second channel estimation result of the second resource location based on the first target fusion feature. This reduces the amount of algorithmic computation for channel estimation of the second resource location, lowers the complexity of the channel estimation algorithm, shortens the processing latency of channel estimation, and improves the efficiency and accuracy of channel estimation. It eliminates the need for pilot configuration for each resource unit in the resource location to be estimated, thus reducing the computational overhead of channel estimation. It also eliminates the need to limit the reference signal, improving the compatibility of the channel estimation method with different types of channels, thereby improving the scenario adaptability and practicality of the channel estimation method.

[0036] In the above embodiments, the acquisition of the first channel estimation result and the second channel estimation result can be combined with... Figure 2 To understand further, Figure 2 This is a flowchart illustrating a channel estimation method according to another embodiment of the present disclosure, as shown below. Figure 2 As shown, the method includes: S201, based on the first channel estimation result and the reference signal, obtain the first term representation of each of the multiple first resource units in the first resource location.

[0037] In this embodiment of the disclosure, each resource element (RE) included in the first resource location can be determined as a plurality of first resource elements included in the first resource location.

[0038] In this scenario, the token representation generation algorithm in the relevant technology can be used to process each first resource unit separately, thereby generating token representations for each of the multiple first resource units, which serve as the first token representations for each of the multiple first resource units.

[0039] In some possible implementations, based on the first channel estimation result, the first unit channel estimation result of each of the plurality of first resource elements is determined, and based on the reference signal, the first unit signal of each of the plurality of first resource elements is determined.

[0040] In this embodiment of the disclosure, the first channel estimation result can be processed by an algorithm for obtaining the unit channel estimation result of a resource unit in the related art, and the first unit channel estimation result on any first resource unit can be obtained based on the result of the algorithm processing.

[0041] Furthermore, the reference signal at the first resource location can be processed by the unit signal acquisition algorithm on the resource unit in the related technology to extract the signal on any first resource unit from the reference signal and use it as the first unit signal on any first resource unit.

[0042] In some possible implementations, for any first resource unit, a first term representation of the first resource unit is generated based on the first signal metadata of the first unit signal of the first resource unit and the first unit channel estimation result of the first resource unit. The first signal metadata includes at least one of the following: signal measurement data, signal time-frequency coordinate data, signal pattern data, signal scale data, signal transmission data, and signal physical data of the first unit signal.

[0043] In this embodiment of the disclosure, the first unit signal has corresponding metadata, which can be identified as the first signal metadata of the first unit signal. The first signal metadata may include at least one of the following: signal measurement data, signal time-frequency coordinate data, signal pattern data, signal scale data, signal transmission data, and signal physical data of the first unit signal.

[0044] As an example, the signal measurement data of the first unit signal includes, but is not limited to, the original measurement results of the first unit signal, the least squares (LS) results of the first unit signal, etc.

[0045] The signal time-frequency coordinate data of the first unit signal can be either absolute coordinates or relative coordinates.

[0046] The signal pattern data of the first unit signal includes, but is not limited to, the frequency domain density and starting point of the first unit signal, the symbol distribution of the time-domain orthogonal frequency division multiplexing (OFDM) technology, and information related to the comb structure.

[0047] The signal scale data of the first unit signal includes, but is not limited to: (1) the scale characteristics of the resource grid of the first unit signal: including but not limited to the number of resource blocks (RBs) or subcarriers and the number of OFDM symbols of the first unit signal; (2) the physical scale characteristics of the resource grid of the first unit signal: including but not limited to the subcarrier spacing, OFDM symbol duration, system bandwidth and signal bandwidth of the first unit signal.

[0048] The signal transmission data of the first unit signal includes, but is not limited to, the port number, layer number, antenna number, and reference signal sequence related data of the first unit signal.

[0049] The physical data of the first unit signal includes, but is not limited to, the signal-to-noise ratio (SNR), noise power, signal power, Doppler spread, time delay spread, time offset, and frequency offset of the first unit signal.

[0050] In this scenario, a lexical representation generation algorithm from related technologies can be used to process the lexical representation based on the channel estimation result of the first unit on any first resource unit and the first signal metadata of the first unit signal. The lexical representation on the first resource unit is obtained based on the result of the algorithm processing and is used as the first lexical representation of the first resource unit.

[0051] S202, based on the first word representation of each of the multiple first resource units, the first target fusion feature of the first resource position is obtained.

[0052] In this embodiment of the disclosure, for any first resource unit's first lexical representation, the first lexical representation carries feature information of the first unit channel estimation result on the first resource unit and feature information of the first signal metadata of the first unit signal. In this scenario, by fusing the first lexical representations of multiple first resource units, a fused feature carrying feature information of the reference signal at the first resource location and feature information of the first channel estimation result obtained through the reference signal can be obtained. This fused feature can be determined as the first target fused feature of the first resource location.

[0053] In some possible implementations, the first token representations of each of the plurality of first resource units are encoded separately to obtain the first token embeddings of each of the plurality of first token representations.

[0054] In this embodiment of the disclosure, the first lexical representation of any first resource unit can be processed by an encoding algorithm in the related art, and the embedding vector of the first lexical representation of the first resource unit can be obtained based on the result of the algorithm processing, which is used as the first lexical embedding of the first lexical representation.

[0055] In some possible implementations, feature fusion using an attention mechanism is performed on multiple first word embeddings to obtain the fused first target fused feature.

[0056] In this embodiment of the disclosure, an attention-based feature fusion algorithm can be used to perform feature fusion processing on multiple first word embeddings. Based on the result of the algorithm processing, the fused features of multiple first word embeddings are obtained, which is the first target fused feature.

[0057] It should be noted that the above-mentioned fusion feature algorithm can be obtained through the self-attention mechanism in related technologies to obtain the first target fusion feature, or other algorithms that can achieve feature fusion of multiple first word embeddings can be used to obtain the first target fusion feature. No specific limitation is made here.

[0058] As an example, the acquisition of the first target fusion feature can be understood using the following formula:

[0059]

[0060] In the above formula, Z represents a word embedding matrix composed of multiple first word embeddings. , and Indicates attention weights, Indicates the first target fusion feature, Let B represent a single-head dimension, and let B represent the bias matrix, where B includes... For any two first word embeddings i and j, The relative temporal and frequency domain positions of the first word embedding i and the first word embedding j are used to determine their relative positions. , }get.

[0061] S203, based on the target signal occupying the second resource location, obtain the second term representation of each of the multiple second resource units in the second resource location.

[0062] In this embodiment of the disclosure, the signal occupying the second resource location in the target channel can be determined as the target signal, and the multiple resource units included in the second resource location can be determined as multiple second resource units in the second resource location.

[0063] In some possible implementations, for any second resource unit, a second unit signal on the second resource unit is determined based on the target signal, and a second lexical representation of the second resource unit is generated based on the second signal metadata of the second unit signal, wherein the second signal metadata includes at least one of the following: signal measurement data, signal time-frequency coordinate data, signal pattern data, signal scale data, signal transmission data, and signal physical data of the second unit signal.

[0064] In this embodiment of the disclosure, the target signal at the second resource location can be processed by an algorithm using the signal acquisition method of the resource unit in the related art, so as to extract the signal of any second resource unit from the target signal and use it as the second unit signal of the second resource unit.

[0065] In this embodiment of the disclosure, the second unit signal is deployed with corresponding metadata, which can be identified as the second signal metadata of the second unit signal.

[0066] As an example, the signal measurement data of the second unit signal includes, but is not limited to, the original measurement results of the second unit signal, the least squares results of the second unit signal, etc.

[0067] The signal pattern data of the second unit signal includes, but is not limited to, the frequency domain density and starting point of the second unit signal, OFDM symbol distribution, and comb structure related information.

[0068] The signal scale data of the second unit signal includes, but is not limited to: (1) the scale characteristics of the resource grid of the second unit signal: including but not limited to the number of RBs or subcarriers and the number of OFDM symbols of the second unit signal; (2) the physical scale characteristics of the resource grid of the second unit signal: including but not limited to the subcarrier spacing, OFDM symbol duration, system bandwidth and signal bandwidth of the second unit signal.

[0069] The signal transmission data of the second unit signal includes, but is not limited to, the port number, layer number, antenna number, and reference signal sequence related data of the second unit signal.

[0070] The physical data of the second unit signal includes, but is not limited to, the SNR, noise power, signal power, Doppler spread, time delay spread, time offset, and frequency offset of the second unit signal.

[0071] In this scenario, the second signal metadata of any second unit signal can be processed by the lexical representation acquisition algorithm in the relevant technology to obtain the lexical representation generated based on the second signal metadata of the second unit signal. This lexical representation is the second lexical representation of the second unit signal.

[0072] S204. Based on the first target fusion features and the second lexical representations of each of the multiple second resource units, the second channel estimation result of the second resource location is obtained.

[0073] In this embodiment of the disclosure, there is a certain degree of correlation between the channel state at the first resource location and the channel state at the second resource location. In this scenario, the first target fusion feature carrying the reference signal and channel state information at the first resource location can be analyzed accordingly, and then the channel estimation for the second resource location can be achieved based on the analysis results.

[0074] In some possible implementations, the second token representations of each of the plurality of second resource units are encoded to obtain the second token embeddings of each of the plurality of second token representations.

[0075] In this embodiment of the disclosure, the second lexical representation of any second resource unit can be encoded using an encoding algorithm in the related art, and the resulting encoded embedding vector can be determined as the second lexical embedding of the second lexical representation.

[0076] In some possible implementations, for any second word embedding, the second word embedding and the first target fusion feature are fused using an attention mechanism to obtain the second target fusion feature of the second resource unit to which the second word embedding belongs.

[0077] In this embodiment of the disclosure, for any second word embedding, the second word embedding and the first target fusion feature can be fused using a feature fusion algorithm based on an attention mechanism in related technologies. The resulting fusion feature can be determined as the fusion feature on the second resource unit corresponding to the second word embedding, and the fusion feature can be determined as the second target fusion feature of the second resource unit.

[0078] The feature fusion algorithm can be constructed based on the cross-attention mechanism in related technologies, or it can be constructed using other feature fusion algorithms that can embed the second word into the first target fusion feature. No specific limitation is made here.

[0079] As an example, the acquisition of the second target fusion features can be understood using the following formula:

[0080]

[0081] In the above formula, Let Q represent the first target fusion feature, and let Q represent the second word embedding. , and Y represents the attention mechanism weights, and Y represents the second objective fusion features. This indicates the relative positional offset between the second resource unit to which the second word embedding Q belongs and the first resource unit in the first resource position.

[0082] In some possible implementations, a second channel estimation result for the second resource location is obtained based on the second target fusion features of each of the multiple second resource units.

[0083] In this embodiment of the disclosure, for multiple second resource elements, the channel estimation results of each of the multiple second resource elements can be obtained, and the overall second channel estimation result of the second resource location can be obtained based on the integration of the channel estimation results of the multiple resource elements.

[0084] Alternatively, the second target fusion features of multiple second resource units can be integrated to obtain the overall second channel estimation result of the second resource location based on the integrated features.

[0085] As one possible implementation, for any second resource unit, based on preset feature dimension constraints, the second target fusion features of the second resource unit are integrated to obtain the second unit channel estimation result of the second resource unit. Based on the second unit channel estimation results of multiple second resource units, the second channel estimation result of the second resource location is obtained.

[0086] In this embodiment of the disclosure, the second target fusion feature on any second resource unit carries feature information of the second unit signal on the second resource unit, as well as feature information of the reference signal and channel estimation result on the first resource location. In this scenario, the channel estimation result of the second resource unit can be obtained by cross-fusion of the second target fusion feature on the second resource unit obtained by the above attention mechanism. The channel estimation result can be determined as the second unit channel estimation result of the second resource unit.

[0087] In this scenario, a dimension integration algorithm from a related technology can be used to perform dimension integration on the second target fusion feature of the second resource unit based on preset feature dimension constraints. Based on the result of the algorithm processing, the channel estimation result on the second resource unit is obtained, which is the channel estimation result of the second unit.

[0088] In this embodiment of the disclosure, the second channel estimation results of each of the multiple second resource units can be integrated using an integration algorithm for unit channel estimation results in the related art, and the overall second channel estimation result of the second resource location can be obtained based on the processing result of the integration algorithm.

[0089] As another possible implementation, the second target fusion features of each of the multiple second resource units are integrated to obtain the third target fusion features of the second resource location. Based on the preset feature dimension constraints, the third target fusion features are dimensionally integrated to obtain the second channel estimation result of the second resource location.

[0090] In this embodiment of the disclosure, the second target fusion features of each of the multiple second resource units can be processed by the feature fusion algorithm in the related technology. The fusion feature is obtained based on the result of the algorithm processing, and the fusion feature can be determined as the third target fusion feature of the second resource location.

[0091] In this scenario, a dimension integration algorithm from a related technology can be used to perform dimension integration on the fused features of the third target based on preset feature dimension constraints. The feature dimensions of the fused features of the third target can be adjusted based on the algorithm processing results, and the adjusted results can be used as the second channel estimation results for the second resource location.

[0092] It should be noted that the feature dimension adjustment algorithm proposed in this disclosure can be implemented by a multi-layer perceptron (MLP) in related technologies, a convolutional neural network (CNN) in related technologies, an attention mechanism algorithm in related technologies, or other algorithms that can achieve feature dimension adjustment. No specific limitation is made here.

[0093] It should be noted that this disclosure does not impose any restrictions on the execution sequence of steps S201 to S204. Figure 2 Example only, executing steps S201 to S204 in sequence. The channel estimation method proposed in this disclosure obtains the second target fusion feature on the second resource unit based on the unit signal on the second resource location and the first target fusion feature on the first resource location, so as to obtain the channel estimation result of the second resource location. It combines the reference signal feature of the first resource location, the feature of the channel estimation result and the feature of the unit signal of the second resource unit, which improves the accuracy and precision of the channel estimation of the second resource location, reduces the influence of redundant information on the channel estimation of the second resource location, eliminates the need to configure pilots for each second unit resource in the second resource location, reduces the cumbersomeness of the channel estimation operation and the complexity of the algorithm, and reduces the computational cost of channel estimation.

[0094] To implement the channel estimation method proposed in the above embodiments, this disclosure also proposes a training method for the channel estimation model, which can be combined with... Figure 3 understand, Figure 3 This is a flowchart illustrating a training method for a channel estimation model according to an embodiment of the present disclosure, as shown below. Figure 3 As shown, the method includes: S301, the first model parameters of the initial channel estimation model are obtained by calculating the model using the candidate parameters to be trained and the sample metadata of the sample signal.

[0095] In this embodiment of the disclosure, the model used to calculate some model parameters of the channel estimation model can be determined as the candidate parameter calculation model to be trained. The candidate parameter calculation model can be constructed based on the hypernetwork in the related technology, or it can be constructed based on the film in the related technology, or it can be constructed based on the model in the related technology that can realize dynamic calculation of model parameters. No specific limitation is made here.

[0096] In this embodiment of the disclosure, the signal used for model training can be determined as a sample signal, and the metadata of the sample signal can be determined as sample metadata. In this scenario, the sample metadata of the sample signal can be input into the candidate parameter calculation model, and the sample metadata can be processed by the candidate parameter calculation model to obtain the output result of the candidate parameter calculation model.

[0097] In this embodiment of the disclosure, the channel estimation model that requires model parameter assignment can be determined as the initial channel estimation model. In this scenario, the model parameters that the initial channel estimation model needs to be assigned can be obtained from the output results of the candidate parameter calculation model based on the assignment requirements of the model parameters of the initial channel estimation model. These model parameters can then be determined as the first model parameters of the initial channel estimation model.

[0098] As an example, the sample metadata of a sample signal may include, but is not limited to: frequency domain density, bias, comb form, time domain symbol set and number, pre-demodulation reference signal position, mapping type, demodulation reference signal type A position, sample signal scale data, physical channel type, sample signal transmission data, and sample signal physical data.

[0099] S302, assign values ​​to the initial channel estimation model based on the first model parameters to obtain the candidate channel estimation model to be trained.

[0100] In this embodiment of the disclosure, the initial channel estimation model can be assigned values ​​based on the first model parameters using a model assignment algorithm in the related art, and the channel estimation model after the first model parameters are assigned values ​​can be determined as a candidate channel estimation model to be trained.

[0101] S303, based on the sample signal, jointly train the candidate channel estimation model and the candidate parameter calculation model to obtain the trained target channel estimation model and target parameter calculation model.

[0102] The target channel estimation model is used to achieve the above. Figures 1 to 2 The channel estimation method proposed in the example.

[0103] In this embodiment of the disclosure, the candidate channel estimation model and the candidate parameter calculation model can be jointly trained based on sample signals using model training methods in related technologies, thereby realizing the iteration and optimization of the candidate channel estimation model and the candidate parameter calculation model.

[0104] Specifically, for any training round, if both the candidate channel estimation model and the candidate parameter calculation model meet the preset model training termination conditions after the training of that round, the joint training of the candidate channel estimation model and the candidate parameter calculation model can be terminated, and the trained model can be determined as the target channel estimation model and the target parameter calculation model.

[0105] It should be noted that the target channel estimation model deploys an attention mechanism network. When training the model based on multiple sample signals under different channel types, the attention mechanism network deployed in the target channel estimation model only participates in the iterative optimization of model parameters and does not make other adjustments.

[0106] It should be noted that this disclosure does not impose any restrictions on the execution sequence of steps S301 to S303. Figure 3 The example only demonstrates the sequential execution of steps S301 to S303.

[0107] The channel estimation model training method proposed in this disclosure improves the robustness and stability of the trained target parameter calculation model and target channel estimation model through the training process, as well as the accuracy and quality of their outputs. Based on the joint training of the parameter calculation model and the channel estimation model, it enhances the adaptability of the model parameters output by the target parameter calculation model to the parameter assignment requirements of the target channel estimation model, enabling dynamic adjustment of the model parameters of the target channel estimation model. This improves the compatibility and matching degree of the target channel estimation model for different types of channels, and enhances the accuracy, precision, and efficiency of the target channel estimation model when estimating different types of channels. It eliminates the need for separate resource development for different types of channels, reducing the model deployment cost of channel estimation, saving resource overhead, and optimizing the channel estimation method.

[0108] In the above embodiments, the acquisition of the target channel estimation model and the target parameter calculation model can be combined with... Figure 4 To understand further, Figure 4 This is a flowchart illustrating a training method for a channel estimation model according to another embodiment of the present disclosure, as shown below. Figure 4 As shown, the method includes: S401, the first model parameters of the initial channel estimation model are obtained by calculating the model using the candidate parameters to be trained and the sample metadata of the sample signal. The initial channel estimation model is then assigned values ​​based on the first model parameters to obtain the candidate channel estimation model to be trained.

[0109] In this embodiment, the specific information regarding step S401 can be understood in conjunction with the relevant content in the above embodiments, and will not be repeated here.

[0110] S402, input the sample signal into the candidate channel estimation model to obtain the predicted channel estimation result output by the candidate channel estimation model.

[0111] In this embodiment of the disclosure, a sample signal can be input into a candidate channel estimation model, and the sample signal can be feature extracted and processed by the model algorithm deployed in the candidate channel estimation model. Based on the result of the algorithm processing, a channel estimation result based on the sample signal can be obtained.

[0112] The channel estimation result can be used as the predicted channel estimation result output by the candidate channel estimation model.

[0113] It should be noted that the model algorithm in the candidate channel estimation model can be deployed based on the relevant algorithm content of the channel estimation algorithm mentioned in the above embodiments, or it can be deployed using other algorithms capable of channel estimation. No specific limitation is made here.

[0114] S403, based on the predicted channel estimation results and the tag channel estimation results of the sample signals, optimize the candidate channel estimation model and the candidate parameter calculation model to obtain the target channel estimation model and the target parameter calculation model.

[0115] In some possible implementations, a first loss value is obtained based on the predicted channel estimation result and the tag channel estimation result, and the parameters of the candidate channel estimation model are adjusted based on the first loss value to obtain the parameter-adjusted candidate channel estimation model.

[0116] In this embodiment of the disclosure, the channel estimation result carried in the tag information of the sample signal can be determined as the tag channel estimation result. In this scenario, the predicted channel estimation result and the tag channel estimation result can be processed by the loss value algorithm in the related technology. Based on the result of the algorithm processing, the loss value of the predicted channel estimation result based on the tag channel estimation result is obtained as the first loss value.

[0117] The first loss value can be used as the model training loss of the candidate channel estimation model. In this scenario, the model parameters of the candidate channel estimation model can be adjusted and optimized based on the first loss value through the model parameter adjustment and optimization methods in related technologies, so as to obtain the adjusted candidate channel estimation model.

[0118] In some possible implementations, a second loss value is obtained based on the second model parameters and the first model parameters of the parameter-adjusted candidate channel estimation model, and the parameters of the candidate parameter calculation model are adjusted based on the second loss value to obtain the parameter-adjusted candidate parameter calculation model.

[0119] In this embodiment of the disclosure, for the candidate channel estimation model with adjusted parameters, the adjusted parameters corresponding to the first model parameters in the candidate channel estimation model with adjusted parameters can be obtained, and the adjusted parameters can be determined as the second model parameters.

[0120] In this scenario, the first model parameters and the second model parameters can be processed using loss value algorithms from related technologies. Based on the results of the algorithm processing, the loss value of the first model parameters based on the second model parameters is obtained, which is then used as the second loss value.

[0121] The second loss value can be determined as the model training loss of the candidate parameter calculation model. In this scenario, the model parameters of the candidate parameter calculation model can be adjusted and optimized based on the second loss value using the model parameter adjustment and optimization methods in related technologies, thereby obtaining the adjusted candidate parameter calculation model.

[0122] In some possible implementations, the process returns to obtain the next sample signal and the next sample metadata of the next sample signal, and continues to train the parameter-adjusted candidate channel estimation model and the parameter-adjusted candidate parameter calculation model until the training ends, thereby obtaining the target channel estimation model and the target parameter calculation model.

[0123] In this embodiment of the disclosure, for the parameter-adjusted candidate channel estimation model and the parameter-adjusted candidate parameter calculation model obtained after any training round, it is possible to identify and determine whether the two meet the preset model training termination conditions. When it is identified that neither of them meets the preset model training termination conditions, the next sample signal and the next sample metadata of the next sample signal can be obtained, and the parameter-adjusted candidate channel estimation model and the parameter-adjusted candidate parameter calculation model can be jointly trained in the next round based on the next sample signal and the next sample metadata.

[0124] Accordingly, if both the candidate channel estimation model and the candidate parameter calculation model after parameter adjustment meet the preset model training termination conditions, the joint training of the candidate channel estimation model and the candidate parameter calculation model can be terminated, and the model obtained after training can be determined as the trained target channel estimation model and target parameter calculation model.

[0125] It should be noted that the termination condition for model training can be based on the number of training epochs or the quality of the model output results; no specific limitation is made here.

[0126] It should be noted that this disclosure does not impose any restrictions on the execution sequence of steps S401 to S403. Figure 4 The example only demonstrates the sequential execution of steps S401 to S403.

[0127] The channel estimation model training method proposed in this disclosure improves the robustness and stability of the trained target parameter calculation model and target channel estimation model through the training process, as well as the accuracy and quality of their outputs. Based on the joint training of the parameter calculation model and the channel estimation model, it enhances the adaptability of the model parameters output by the target parameter calculation model to the parameter assignment requirements of the target channel estimation model, enabling dynamic adjustment of the model parameters of the target channel estimation model. This improves the compatibility and matching degree of the target channel estimation model for different types of channels, and enhances the accuracy, precision, and efficiency of the target channel estimation model when estimating different types of channels. It eliminates the need for separate resource development for different types of channels, reducing the model deployment cost of channel estimation, saving resource overhead, and optimizing the channel estimation method.

[0128] Corresponding to the channel estimation methods provided in the above embodiments, one embodiment of this disclosure also provides a channel estimation device. Since the channel estimation device provided in this disclosure corresponds to the channel estimation methods provided in the above embodiments, the implementation methods of the above channel estimation methods are also applicable to the channel estimation device provided in this disclosure, and will not be described in detail in the following embodiments.

[0129] Figure 5 This is a schematic diagram of the structure of a channel estimation device according to an embodiment of the present disclosure, as shown below. Figure 5 As shown, the channel estimation device 500 includes a first estimation module 51, a feature fusion module 52, and a second estimation module 53, wherein: The first estimation module 51 is used to perform channel estimation based on the received value of the reference signal to obtain a first channel estimation result of the first resource location occupied by the reference signal in the target channel; Feature fusion module 52 is used to obtain the first target fusion feature of the first resource location based on the first channel estimation result and the reference signal; The second estimation module 53 is used to determine the second channel estimation result of the second resource location based on the first target fusion characteristics, wherein the second resource location is the resource location in the target channel that is not occupied by the reference signal.

[0130] In this embodiment of the disclosure, the feature fusion module 52 is further configured to: obtain the first term representation of each of the multiple first resource units in the first resource location based on the first channel estimation result and the reference signal; and obtain the first target fusion feature of the first resource location based on the first term representation of each of the multiple first resource units.

[0131] In this embodiment of the disclosure, the feature fusion module 52 is further configured to: determine the first unit channel estimation result of each of the plurality of first resource units based on the first channel estimation result; determine the first unit signal of each of the plurality of first resource units based on the reference signal; and generate a first lexical representation of the first resource unit for any first resource unit based on the first signal metadata of the first unit signal of the first resource unit and the first unit channel estimation result of the first resource unit, wherein the first signal metadata includes at least one of the signal measurement data, signal time-frequency coordinate data, signal pattern data, signal scale data, signal transmission data and signal physical data of the first unit signal.

[0132] In this embodiment of the disclosure, the feature fusion module 52 is further configured to: encode the first word representation of each of the multiple first resource units to obtain the first word embedding of each of the multiple first word representations; and perform feature fusion using an attention mechanism on the multiple first word embeddings to obtain the fused first target fused feature.

[0133] In this embodiment of the disclosure, the second estimation module 53 is further configured to: obtain the second word representation of each of the multiple second resource units in the second resource location based on the target signal occupying the second resource location; and obtain the second channel estimation result of the second resource location based on the first target fusion feature and the second word representation of each of the multiple second resource units.

[0134] In this embodiment of the disclosure, the second estimation module 53 is further configured to: for any second resource unit, determine the second unit signal on the second resource unit based on the target signal, and generate a second lexical representation of the second resource unit based on the second signal metadata of the second unit signal, wherein the second signal metadata includes at least one of the signal measurement data, signal time-frequency coordinate data, signal pattern data, signal scale data, signal transmission data, and signal physical data of the second unit signal.

[0135] In this embodiment of the disclosure, the second estimation module 53 is further configured to: encode the second lexical representations of the plurality of second resource units to obtain the second lexical embeddings of the plurality of second lexical representations; for any second lexical embedding, perform feature fusion using an attention mechanism on the second lexical embedding and the first target fusion feature to obtain the second target fusion feature of the second resource unit to which the second lexical embedding belongs; and obtain the second channel estimation result of the second resource location based on the second target fusion features of the plurality of second resource units.

[0136] In this embodiment of the disclosure, the second estimation module 53 is further configured to: for any second resource unit, based on preset feature dimension constraints, integrate the second target fusion features of the second resource unit to obtain the second unit channel estimation result of the second resource unit; and based on the second unit channel estimation results of multiple second resource units, obtain the second channel estimation result of the second resource location.

[0137] In this embodiment of the disclosure, the second estimation module 53 is further configured to: integrate the second target fusion features of each of the multiple second resource units to obtain the third target fusion features of the second resource location; and integrate the third target fusion features based on preset feature dimension constraints to obtain the second channel estimation result of the second resource location.

[0138] The channel estimation apparatus proposed in this disclosure obtains a first target fusion feature based on the first channel estimation result of the first resource location occupied by the reference signal, and then determines the second channel estimation result of the second resource location based on the first target fusion feature. This reduces the amount of algorithmic computation for channel estimation of the second resource location, lowers the complexity of the channel estimation algorithm, shortens the processing latency of channel estimation, and improves the efficiency and accuracy of channel estimation. It eliminates the need for pilot configuration for each resource unit in the resource location to be estimated, reducing the computational overhead of channel estimation. It also eliminates the need to limit the reference signal, improving the compatibility of the channel estimation method with different types of channels, thereby improving the scenario adaptability and practicality of the channel estimation method.

[0139] Corresponding to the channel estimation model training methods provided in the above embodiments, an embodiment of this disclosure also provides a channel estimation model training apparatus. Since the channel estimation model training apparatus provided in this disclosure corresponds to the channel estimation model training methods provided in the above embodiments, the implementation methods of the above channel estimation model training methods are also applicable to the channel estimation model training apparatus provided in this disclosure, and will not be described in detail in the following embodiments.

[0140] Figure 6 This is a schematic diagram of the structure of a training device for a channel estimation model according to an embodiment of the present disclosure, as shown below. Figure 6As shown, the training device 600 for the channel estimation model includes an acquisition module 61, an assignment module 62, and a training module 63, wherein: The acquisition module 61 is used to calculate the model and sample metadata of the sample signal through the candidate parameters to be trained, and obtain the first model parameters of the initial channel estimation model. The assignment module 62 is used to assign values ​​to the initial channel estimation model based on the first model parameters to obtain the candidate channel estimation model to be trained. Training module 63 is used to jointly train the candidate channel estimation model and the candidate parameter calculation model based on sample signals, so as to obtain the trained target channel estimation model and target parameter calculation model. The target channel estimation model is used to implement the above... Figure 5 The channel estimation device proposed in the embodiment.

[0141] In this embodiment of the disclosure, the training module 63 is further configured to: input the sample signal into the candidate channel estimation model to obtain the predicted channel estimation result output by the candidate channel estimation model; and optimize the candidate channel estimation model and the candidate parameter calculation model based on the predicted channel estimation result and the tag channel estimation result of the sample signal to obtain the target channel estimation model and the target parameter calculation model.

[0142] In this embodiment of the present disclosure, the training module 63 is further configured to: obtain a first loss value based on the predicted channel estimation result and the tag channel estimation result, and adjust the parameters of the candidate channel estimation model based on the first loss value to obtain a parameter-adjusted candidate channel estimation model; obtain a second loss value based on the second model parameters and the first model parameters of the parameter-adjusted candidate channel estimation model, and adjust the parameters of the candidate parameter calculation model based on the second loss value to obtain a parameter-adjusted candidate parameter calculation model; return to obtain the next sample signal and the next sample metadata of the next sample signal, and continue to train the parameter-adjusted candidate channel estimation model and the parameter-adjusted candidate parameter calculation model until the training ends, thereby obtaining the target channel estimation model and the target parameter calculation model.

[0143] The training device for the channel estimation model proposed in this disclosure improves the robustness and stability of the trained target parameter calculation model and target channel estimation model through the training process, as well as the accuracy and quality of their outputs. Based on the joint training of the parameter calculation model and the channel estimation model, it enhances the adaptability of the model parameters output by the target parameter calculation model to the parameter assignment requirements of the target channel estimation model, enabling dynamic adjustment of the model parameters of the target channel estimation model. This improves the compatibility and matching degree of the target channel estimation model for different types of channels, and enhances the accuracy, precision, and efficiency of the target channel estimation model when estimating different types of channels. It eliminates the need for separate resource development for different types of channels, reducing the model deployment cost of channel estimation, saving resource overhead, and optimizing the channel estimation method.

[0144] To achieve the above embodiments, this disclosure also provides an electronic device, a computer-readable storage medium, and a computer program product.

[0145] Figure 7 This is a schematic diagram of the structure of an electronic device according to an embodiment of the present disclosure. For example, the electronic device 700 may be a vehicle, mobile phone, computer, digital broadcasting terminal, messaging device, game console, tablet device, medical device, fitness equipment, personal digital assistant, etc.

[0146] Reference Figure 7 The electronic device 700 may include one or more of the following components: processing component 702, memory 704, power component 706, multimedia component 708, audio component 710, input / output (I / O) interface 712, sensor component 714, and communication component 716.

[0147] Processing component 702 typically controls the overall operation of electronic device 700, such as operations associated with display, telephone calls, data communication, camera operation, and recording operations. Processing component 702 may include one or more processors 720 to execute instructions to complete all or part of the steps of the channel estimation method described above. Furthermore, processing component 702 may include one or more modules to facilitate channel estimation between processing component 702 and other components. For example, processing component 702 may include a multimedia module to facilitate channel estimation between multimedia component 708 and processing component 702.

[0148] Memory 704 is configured to store various types of data to support the operation of electronic device 700. Examples of this data include instructions for any application or method operating on electronic device 700, contact data, phonebook data, messages, pictures, videos, etc. Memory 704 can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as Static Random Access Memory (SRAM), Electrically Erasable Programmable Read-Only Memory (EEPROM), Erasable Programmable Read-Only Memory (EPROM), Programmable Read-Only Memory (PROM), Read-Only Memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk.

[0149] Power component 706 provides power to various components of electronic device 700. Power component 706 may include a power management system, one or more power supplies, and other components associated with generating, managing, and distributing power to electronic device 700.

[0150] Multimedia component 708 includes a screen that provides an output interface between electronic device 700 and user. In some embodiments, the screen may include a Liquid Crystal Display (LCD) and a Touch Panel (TP). If the screen includes a Touch Panel, the screen may be implemented as a touchscreen to receive input signals from the user. The Touch Panel includes one or more touch sensors to sense touches, swipes, and gestures on the Touch Panel. The touch sensors may sense not only the boundaries of touch or swipe actions but also the duration and pressure associated with the touch or swipe operation. In some embodiments, multimedia component 708 includes a front-facing camera and / or a rear-facing camera. When electronic device 700 is in an operating mode, such as a shooting mode or video mode, the front-facing camera and / or rear-facing camera may receive external multimedia data. Each front-facing camera and rear-facing camera may be a fixed optical lens system or have focal length and optical zoom capabilities.

[0151] Audio component 710 is configured to output and / or input audio signals. For example, audio component 710 includes a microphone (MIC) configured to receive external audio signals when electronic device 700 is in an operating mode, such as call mode, recording mode, and voice recognition mode. The received audio signals may be further stored in memory 704 or transmitted via communication component 716. In some embodiments, audio component 710 also includes a speaker for outputting audio signals.

[0152] I / O interface 712 provides an interface between processing component 702 and peripheral interface modules, such as keyboards, click wheels, buttons, etc. These buttons may include, but are not limited to, home buttons, volume buttons, power buttons, and lock buttons.

[0153] Sensor assembly 714 includes one or more sensors for providing state assessments of various aspects of electronic device 700. For example, sensor assembly 714 may detect the on / off state of electronic device 700, the relative positioning of components such as the display and keypad of electronic device 700, changes in position of electronic device 700 or a component of electronic device 700, the presence or absence of user contact with electronic device 700, orientation or acceleration / deceleration of electronic device 700, and temperature changes of electronic device 700. Sensor assembly 714 may include a proximity sensor configured to detect the presence of nearby objects without any physical contact. Sensor assembly 714 may also include an optical sensor, such as a complementary metal-oxide-semiconductor (CMOS) or charge-coupled device (CCD) image sensor, for use in imaging applications. In some embodiments, sensor assembly 714 may also include an accelerometer, gyroscope, magnetometer, pressure sensor, or temperature sensor.

[0154] Communication component 716 is configured to facilitate wired or wireless communication between electronic device 700 and other devices. Electronic device 700 can access wireless networks based on communication standards, such as WiFi, 4G, or 7G, or combinations thereof. In one exemplary embodiment, communication component 716 receives broadcast signals or broadcast-related information from an external broadcast management system via a broadcast channel. In one exemplary embodiment, communication component 716 also includes a Near Field Communication (NFC) module to facilitate short-range communication. For example, the NFC module may be implemented based on Radio Frequency Identification (RFID), Infrared Data Association (IrDA), Ultra-Wideband (UWB), Bluetooth, and other technologies.

[0155] In an exemplary embodiment, the electronic device 700 may be implemented by one or more application-specific integrated circuits (ASICs), digital signal processors (DSPs), digital signal processing devices (DSPDs), programmable logic devices (PLDs), field-programmable gate arrays (FPGAs), controllers, microcontrollers, microprocessors, or other electronic components to perform the channel estimation method described above.

[0156] In an exemplary embodiment, a non-transitory computer-readable storage medium including instructions is also provided, such as a memory 704 including instructions, which can be executed by a processor 720 of an electronic device 700 to complete the channel estimation method described above. For example, the non-transitory computer-readable storage medium may be a read-only memory (ROM), a random access memory (RAM), a compact disc read-only memory (CD-ROM), magnetic tape, a floppy disk, and an optical data storage device, etc.

[0157] To implement the above embodiments, this disclosure also proposes a computer-readable storage medium storing computer program instructions that, when executed by a processor, implement the steps of the channel estimation method provided in this disclosure.

[0158] Alternatively, the computer-readable storage medium may be ROM, random access memory (RAM), CD-ROM, magnetic tape, floppy disk, and optical data storage device, etc.

[0159] To implement the above embodiments, this disclosure also proposes a chip including an interface circuit and a processing circuit coupled to each other. The interface circuit is used to input or output signals, and the processing circuit is configured to implement the steps of the channel estimation method provided in this disclosure.

[0160] Figure 8 This is a schematic diagram of the structure of a chip according to an embodiment of this disclosure. See also... Figure 8 The diagram shown is a schematic representation of the structure of chip 800, but is not limited to this.

[0161] Chip 800 includes processing circuitry 801, which is configured to perform any of the channel estimation methods described above.

[0162] In some embodiments, the chip 800 further includes one or more interface circuits 802. Optionally, the interface circuit 802 is connected to the memory 803, and the interface circuit 802 can be used to receive signals from the memory 803 or other devices, and the interface circuit 802 can be used to send signals to the memory 803 or other devices. For example, the interface circuit 802 can read instructions stored in the memory 803 and send the instructions to the processing circuit 801.

[0163] In some embodiments, the interface circuit 802 performs at least one of the communication steps such as sending and / or receiving in the above method, and the processing circuit 801 performs other steps.

[0164] In some embodiments, the terms interface circuit, interface, transceiver pin, transceiver, etc., can be used interchangeably.

[0165] In some embodiments, chip 800 further includes one or more memories 803 for storing instructions. Optionally, all or part of the memories 803 may be located outside of chip 800.

[0166] To implement the above embodiments, this disclosure also proposes a computer program product, including a computer program that, when executed by a processor, implements the steps of the channel estimation method provided in this disclosure.

[0167] Other embodiments of this disclosure will readily occur to those skilled in the art upon consideration of the specification and practice of the invention disclosed herein. This disclosure is intended to cover any variations, uses, or adaptations of this disclosure that follow the general principles of this disclosure and include common knowledge or customary techniques in the art not disclosed herein. The specification and examples are to be considered exemplary only, and the true scope and spirit of this disclosure are indicated by the following claims.

[0168] It should be understood that this disclosure is not limited to the precise structures described above and shown in the accompanying drawings, and various modifications and changes can be made without departing from its scope. The scope of this disclosure is limited only by the appended claims.

Claims

1. A channel estimation method, characterized in that, The method includes: Channel estimation is performed based on the received value of the reference signal to obtain a first channel estimation result of the first resource location occupied by the reference signal in the target channel; Based on the first channel estimation result and the reference signal, the first target fusion feature of the first resource location is obtained; Based on the first target fusion feature, a second channel estimation result is determined for the second resource location, wherein the second resource location is a resource location in the target channel that is not occupied by the reference signal.

2. The method according to claim 1, characterized in that, The step of obtaining the target fusion feature of the first resource location based on the first channel estimation result and the reference signal includes: Based on the first channel estimation result and the reference signal, the first word representation of each of the multiple first resource units in the first resource location is obtained; Based on the first lexical representation of each of the plurality of first resource units, the first target fusion feature of the first resource location is obtained.

3. The method according to claim 2, characterized in that, The step of obtaining the first term representation of each of the multiple first resource units in the first resource location based on the first channel estimation result and the reference signal includes: Based on the first channel estimation result, the first unit channel estimation result of each of the plurality of first resource units is determined; Based on the reference signal, the first unit signal of each of the plurality of first resource units is determined; For any first resource unit, a first term representation of the first resource unit is generated based on the first signal metadata of the first unit signal of the first resource unit and the first unit channel estimation result of the first resource unit. The first signal metadata includes at least one of the following: signal measurement data, signal time-frequency coordinate data, signal pattern data, signal scale data, signal transmission data, and signal physical data of the first unit signal.

4. The method according to claim 2, characterized in that, The step of obtaining the first target fusion feature of the first resource location based on the first lexical representation of each of the plurality of first resource units includes: Encode the first word representation of each of the plurality of first resource units to obtain the first word embedding of each of the plurality of first word representations; The first target fusion feature is obtained by performing feature fusion on multiple first word embeddings using an attention mechanism.

5. The method according to claim 1, characterized in that, The second channel estimation result for determining the second resource location based on the first target fusion feature, wherein the second resource location is the unoccupied resource location of the reference signal in the target channel, includes: Based on the target signal occupying the second resource location, obtain the second word representation of each of the multiple second resource units in the second resource location; Based on the first target fusion feature and the second lexical representation of each of the plurality of second resource units, the second channel estimation result of the second resource location is obtained.

6. The method according to claim 5, characterized in that, The step of obtaining the second word representation of each of the multiple second resource units in the second resource location based on the target signal occupying the second resource location includes: For any second resource unit, based on the target signal, a second unit signal on the second resource unit is determined, and based on the second signal metadata of the second unit signal, a second lexical representation of the second resource unit is generated, wherein the second signal metadata includes at least one of the following: signal measurement data, signal time-frequency coordinate data, signal pattern data, signal scale data, signal transmission data, and signal physical data of the second unit signal.

7. The method according to claim 5, characterized in that, The second channel estimation result for the second resource location is obtained based on the first target fusion feature and the second lexical representation of each of the plurality of second resource units, including: Encode the second token representation of each of the plurality of second resource units to obtain the second token embedding of each of the plurality of second token representations; For any second word embedding, feature fusion using an attention mechanism is performed on the second word embedding and the first target fusion feature to obtain the second target fusion feature of the second resource unit to which the second word embedding belongs; Based on the second target fusion features of each of the plurality of second resource units, the second channel estimation result of the second resource location is obtained.

8. The method according to claim 7, characterized in that, The process of obtaining the second channel estimation result for the second resource location based on the second target fusion features of each of the plurality of second resource units includes: For any second resource unit, based on preset feature dimension constraints, the second target fusion features of the second resource unit are integrated by feature dimension to obtain the second unit channel estimation result of the second resource unit; Based on the second-unit channel estimation results of each of the multiple second resource units, the second channel estimation result of the second resource location is obtained.

9. The method according to claim 7, characterized in that, The process of obtaining the second channel estimation result for the second resource location based on the second target fusion features of each of the plurality of second resource units includes: The second target fusion features of each of the plurality of second resource units are integrated to obtain the third target fusion feature of the second resource location; Based on preset feature dimension constraints, the third target fusion features are dimensionally integrated to obtain the second channel estimation result of the second resource location.

10. A training method for a channel estimation model, characterized in that, The method includes: The first model parameters of the initial channel estimation model are obtained by calculating the model and sample metadata of the sample signal using the candidate parameters to be trained. The initial channel estimation model is assigned values ​​based on the first model parameters to obtain the candidate channel estimation model to be trained. The candidate channel estimation model and the candidate parameter calculation model are jointly trained based on the sample signal to obtain the trained target channel estimation model and target parameter calculation model, wherein the target channel estimation model is used to implement the channel estimation method according to any one of claims 1-9.

11. The method according to claim 10, characterized in that, The step of jointly training the candidate channel estimation model and the candidate parameter calculation model based on the sample signal to obtain the trained target channel estimation model and target parameter calculation model includes: The sample signal is input into the candidate channel estimation model to obtain the predicted channel estimation result output by the candidate channel estimation model; Based on the predicted channel estimation results and the tag channel estimation results of the sample signals, the candidate channel estimation model and the candidate parameter calculation model are optimized to obtain the target channel estimation model and the target parameter calculation model.

12. The method according to claim 11, characterized in that, The optimization of the candidate channel estimation model and the candidate parameter calculation model based on the predicted channel estimation result and the tag channel estimation result of the sample signal to obtain the target channel estimation model and the target parameter calculation model includes: Based on the predicted channel estimation result and the tag channel estimation result, a first loss value is obtained, and the parameters of the candidate channel estimation model are adjusted based on the first loss value to obtain the parameter-adjusted candidate channel estimation model. Based on the second model parameters of the candidate channel estimation model after parameter adjustment and the first model parameters, a second loss value is obtained, and the candidate parameter calculation model is adjusted based on the second loss value to obtain the candidate parameter calculation model after parameter adjustment. Return to obtain the next sample signal and the next sample metadata of the next sample signal, and continue to train the candidate channel estimation model and the candidate parameter calculation model after parameter adjustment until the training ends, so as to obtain the target channel estimation model and the target parameter calculation model.

13. A channel estimation device, characterized in that, The device includes: The first estimation module is used to perform channel estimation based on the received value of the reference signal to obtain a first channel estimation result of the first resource location occupied by the reference signal in the target channel; The feature fusion module is used to obtain the first target fusion feature of the first resource location based on the first channel estimation result and the reference signal; The second estimation module is used to determine the second channel estimation result of the second resource location based on the first target fusion feature, wherein the second resource location is the resource location in the target channel where the reference signal is not occupied.

14. The apparatus according to claim 13, characterized in that, The feature fusion module is also used for: Based on the first channel estimation result and the reference signal, the first word representation of each of the multiple first resource units in the first resource location is obtained; Based on the first lexical representation of each of the plurality of first resource units, the first target fusion feature of the first resource location is obtained.

15. The apparatus according to claim 13, characterized in that, The second estimation module is further configured to: Based on the target signal occupying the second resource location, obtain the second word representation of each of the multiple second resource units in the second resource location; Based on the first target fusion feature and the second lexical representation of each of the plurality of second resource units, the second channel estimation result of the second resource location is obtained.

16. A training apparatus for a channel estimation model, characterized in that, The device includes: The acquisition module is used to calculate the model and sample metadata of the sample signal through the candidate parameters to be trained, and obtain the first model parameters of the initial channel estimation model. The assignment module is used to assign values ​​to the initial channel estimation model based on the first model parameters to obtain the candidate channel estimation model to be trained. The training module is used to jointly train the candidate channel estimation model and the candidate parameter calculation model based on the sample signal to obtain the trained target channel estimation model and target parameter calculation model, wherein the target channel estimation model is used to implement the channel estimation device according to any one of claims 13-15.

17. The apparatus according to claim 16, characterized in that, The training module is also used for: The sample signal is input into the candidate channel estimation model to obtain the predicted channel estimation result output by the candidate channel estimation model; Based on the predicted channel estimation results and the tag channel estimation results of the sample signals, the candidate channel estimation model and the candidate parameter calculation model are optimized to obtain the target channel estimation model and the target parameter calculation model.

18. The apparatus according to claim 17, characterized in that, The training module is also used for: Based on the predicted channel estimation result and the tag channel estimation result, a first loss value is obtained, and the parameters of the candidate channel estimation model are adjusted based on the first loss value to obtain the parameter-adjusted candidate channel estimation model. Based on the second model parameters of the candidate channel estimation model after parameter adjustment and the first model parameters, a second loss value is obtained, and the candidate parameter calculation model is adjusted based on the second loss value to obtain the candidate parameter calculation model after parameter adjustment. Return to obtain the next sample signal and the next sample metadata of the next sample signal, and continue to train the candidate channel estimation model and the candidate parameter calculation model after parameter adjustment until the training ends, so as to obtain the target channel estimation model and the target parameter calculation model.

19. An electronic device, characterized in that, include: processor; Memory used to store the processor's executable instructions; The processor is configured to execute instructions to implement the method as described in any one of claims 1-9 and / or 10-12.

20. A computer-readable storage medium, wherein when the instructions in the computer-readable storage medium are executed by a processor of an electronic device, the electronic device is enabled to perform the method as described in any one of claims 1-9 and / or 10-12.

21. A chip, characterized in that, It includes one or more interface circuits and one or more processors; the interface circuits are used to receive signals and send the signals to the processors, the signals including computer instructions stored in a memory, which, when executed by the processor, cause the chip to perform the steps of the method according to any one of claims 1-9 and / or 10-12.

Images