Semantic communication method and related device

Abstract

The present application provides a semantic communication method and a related device. Semantic features and channel gains are combined, and a transmitting-end signal processing function and a receiving-end signal processing function are optimized with the objective of minimizing an error between an original semantic signal and a reconstructed semantic signal, solving the technical problem of adaptively adjusting semantic signals on the basis of channel gains of different carriers, facilitating transmission of extracted semantic features over multiple carriers having different channel gains, and further solving the technical problem of how to migrate a codec trained under an additive white Gaussian noise channel to other fading channels for application, eliminating the disadvantage of requiring codec training for different statistical features or different types of channels, ensuring the semantic communication quality for non-orthogonal multi-users when the original codec is applied to a fading channel, and improving communication efficiency.

Description

Semantic communication method and related equipment Technical Field

[0001] The present application relates to the field of communication technology, and in particular to a semantic communication method with channel migration and related equipment. Background Art

[0002] Semantic communication is a new communication architecture that integrates user information needs and semantics into the communication process, significantly improving transmission efficiency by exploring semantic information. Unlike traditional communication methods, semantic communication focuses not on the accuracy of bit transmission but on the content of the transmission, enabling the receiver to correctly understand the meaning of the sender's information and reducing uncertainty about the information. By eliminating the need to preserve the fidelity of the original data or signal and instead extracting semantic information from the source, a more concise information representation is achieved with the goal of preserving the meaning. This can significantly reduce the required transmission channel bandwidth and improve end-to-end communication efficiency.

[0003] However, in the existing technology, power allocation and performance analysis of user semantic communication have not been carried out. In addition, the semantic codec has the limitation of relying on the channel type and the application generalization effect is poor, which will lead to a decline in the quality of semantic communication.

[0004] Summary of the Invention

[0005] In view of this, the purpose of this application is to propose a semantic communication method with channel migration and related equipment to solve or partially solve the above problems.

[0006] Based on the above objectives, this application provides a semantic communication method, including:

[0007] Obtaining first input information and second input information from a transmitting end, inputting the first input information into a pre-trained semantic encoder, extracting semantic features of the first input information, and obtaining a first semantic signal; inputting the second input information into a pre-trained semantic encoder, extracting semantic features of the second input information, and obtaining a second semantic signal;

[0008] Sending the first semantic signal and the second semantic signal to a first channel, and superimposing the signals through the first channel to obtain a third semantic signal;

[0009] performing transmitter signal processing on the first semantic signal and the second semantic signal according to a predetermined channel coefficient of a second channel to obtain a processed first semantic signal and a processed second semantic signal;

[0010] sending the processed first semantic signal and the processed second semantic signal to the second channel, performing signal superposition to obtain a fourth semantic signal, and performing receiving-end signal processing on the fourth semantic signal according to the channel coefficient to obtain a fifth semantic signal;

[0011] optimizing the fifth semantic signal with the goal of minimizing an error between the third semantic signal and the fifth semantic signal to obtain an optimized fifth semantic signal;

[0012] The optimized fifth semantic signal is input into a pre-trained semantic decoder for semantic decoding to obtain first decoding information and second decoding information; wherein, the first decoding information corresponds to the semantics of the first input information; and the second decoding information corresponds to the semantics of the second input information.

[0013] In a possible implementation, inputting the first input information into a pre-trained semantic encoder, extracting semantic features of the first input information, and obtaining a first semantic signal includes:

[0014] Converting the semantic feature of the first input information into a first complex number corresponding to a carrier;

[0015] The first complex number is normalized according to a preset power constraint condition to obtain the first semantic signal.

[0016] In a possible implementation, inputting the second input information into a pre-trained semantic encoder, extracting semantic features of the second input information, and obtaining a second semantic signal includes:

[0017] Converting the semantic feature of the second input information into a second complex number corresponding to a carrier;

[0018] The second complex number is normalized according to a preset power constraint condition to obtain the second semantic signal.

[0019] In a possible implementation, the first channel is an additive white Gaussian noise channel.

[0020] In a possible implementation, the second channel is a Rayleigh fading channel or a Ricean fading channel.

[0021] In a possible implementation, the method further includes: constructing the semantic encoder and the semantic decoder:

[0022] Acquire historical input messages of the sending end as a training set, extract semantic features of the training set through an initial semantic encoder, and obtain a first training signal;

[0023] Sending the first training signal to a receiving end through the first channel, parsing the first training signal through an initial semantic decoder to obtain a second training signal;

[0024] A loss function is constructed according to the first training signal and the second training signal, and the initial semantic encoder and the initial semantic decoder are trained by minimizing the loss function to obtain the pre-trained semantic encoder and the pre-trained semantic decoder.

[0025] In one possible implementation, performing transmitter signal processing on the first semantic signal and the second semantic signal according to a predetermined channel coefficient of the second channel to obtain the processed first semantic signal and the processed second semantic signal includes:

[0026] constructing a transmitting-end signal processing function according to a predetermined channel coefficient of a second channel;

[0027] Obtaining the processed first semantic signal according to the first semantic signal and the transmitting end signal processing function;

[0028] The processed second semantic signal is obtained according to the second semantic signal and the sending end signal processing function.

[0029] In a possible implementation, performing receiving-end signal processing on the fourth semantic signal according to the channel coefficient to obtain a fifth semantic signal includes:

[0030] constructing a receiving-end signal processing function according to a predetermined channel coefficient of a second channel;

[0031] The fifth semantic signal is obtained according to the fourth semantic signal and the receiving end signal processing function.

[0032] In one possible implementation, optimizing the fifth semantic signal with the goal of minimizing the error between the third semantic signal and the fifth semantic signal to obtain the optimized fifth semantic signal includes:

[0033] Optimizing the transmitting-end signal processing function and the receiving-end signal processing function with the goal of minimizing the error between the third semantic signal and the fifth semantic signal;

[0034] The fifth semantic signal is adjusted according to the optimized sending-end processing function and the optimized receiving-end processing function to minimize the difference between the third semantic signal and the fifth semantic signal, thereby obtaining the optimized fifth semantic signal.

[0035] In a possible implementation, the transmitting end signal processing function is a carrier power allocation function; the receiving end signal processing function is a signal amplitude operation function;

[0036] The method further comprises:

[0037] Determining the channel migration power optimization problem according to the multi-carrier power allocation function, the signal amplitude operation function, and the goal of minimizing the error between the third semantic signal and the fifth semantic signal;

[0038] Performing a semantically similar dual transformation on the channel migration power optimization problem to obtain a channel migration power optimization model;

[0039] The analytical solution of the channel migration power optimization model is determined by the Lagrange dual method, and the carrier power allocation result of the transmitting end and the amplitude scaling parameter of the receiving end are obtained according to the analytical solution and the preset power constraint condition.

[0040] In a possible implementation, inputting the optimized fifth semantic signal into a pre-trained semantic decoder for semantic decoding to obtain first decoding information and second decoding information includes:

[0041] Inputting the optimized fifth semantic signal into the pre-trained semantic decoder, extracting the semantic features for the first semantic signal and the semantic features for the second semantic signal contained in the optimized fifth semantic signal;

[0042] Semantic recovery and information reconstruction are performed based on the semantic features of the first semantic signal and the semantic features of the second semantic signal to obtain the first decoding information and the second decoding information.

[0043] Based on the same purpose, this application also proposes a semantic communication device, comprising:

[0044] an encoding module configured to obtain first input information and second input information from a transmitting end, input the first input information into a pre-trained semantic encoder, extract semantic features of the first input information, and obtain a first semantic signal; and input the second input information into a pre-trained semantic encoder, extract semantic features of the second input information, and obtain a second semantic signal;

[0045] a first superposition module configured to send the first semantic signal and the second semantic signal to a first channel, and perform signal superposition through the first channel to obtain a third semantic signal;

[0046] an optimization module configured to perform transmitter signal processing on the first semantic signal and the second semantic signal according to a predetermined channel coefficient of a second channel to obtain a processed first semantic signal and a processed second semantic signal;

[0047] a second superposition module configured to send the processed first semantic signal and the processed second semantic signal to the second channel, perform signal superposition to obtain a fourth semantic signal, and perform receiving-end signal processing on the fourth semantic signal according to the channel coefficient to obtain a fifth semantic signal;

[0048] an optimization module configured to optimize the fifth semantic signal with the goal of minimizing an error between the third semantic signal and the fifth semantic signal to obtain an optimized fifth semantic signal;

[0049] The decoding module is configured to input the optimized fifth semantic signal into a pre-trained semantic decoder for semantic decoding to obtain first decoding information and second decoding information; wherein, the first decoding information corresponds to the semantics of the first input information; and the second decoding information corresponds to the semantics of the second input information.

[0050] Based on the above purpose, the present application also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and runnable on the processor, wherein when the processor executes the program, it implements the semantic communication method as described in any one of the above.

[0051] Based on the above purpose, the present application also provides a non-transitory computer-readable storage medium, which stores computer instructions, and the computer instructions are used to enable the computer to execute any one of the semantic communication methods described above.

[0052] Based on the above objectives, the present application provides a computer program product, including computer program instructions, which, when executed on a computer, enable the computer to execute any one of the above-mentioned semantic communication methods.

[0053] As can be seen from the above, the present application provides a semantic communication method and related equipment, which first obtains the first input information and the second input information of the sending end, inputs the first input information into a pre-trained semantic encoder, extracts the semantic features of the first input information, and obtains a first semantic signal; inputs the second input information into a pre-trained semantic encoder, extracts the semantic features of the second input information, and obtains a second semantic signal; sends the first semantic signal and the second semantic signal to the first channel, performs signal superposition through the first channel to obtain a third semantic signal, and performs sending end signal processing on the first semantic signal and the second semantic signal according to a predetermined channel coefficient of the second channel to obtain a processed first semantic signal. The processed first semantic signal and the processed second semantic signal are sent to the second channel for signal superposition to obtain a fourth semantic signal, and the fourth semantic signal is subjected to receiving-end signal processing according to the channel coefficient to obtain a fifth semantic signal, and the fifth semantic signal is optimized with the goal of minimizing the error between the third semantic signal and the fifth semantic signal to obtain an optimized fifth semantic signal; the optimized fifth semantic signal is input into a pre-trained semantic decoder for semantic decoding to obtain first decoding information and second decoding information; wherein, the first decoding information corresponds to the semantics of the first input information; and the second decoding information corresponds to the semantics of the second input information.

[0054] This application combines semantic features with channel gain, with the goal of minimizing the error between the original semantic signal and the reconstructed semantic signal, optimizes the signal processing function at the transmitting end and the signal processing function at the receiving end, solves the technical problem of adaptively adjusting the semantic signal according to the channel gain of different carriers, is conducive to the transmission of the extracted semantic features on multiple carriers with different channel gains, and also solves the technical problem of how to migrate the codec trained under the additive white Gaussian noise channel to other fading channel applications, eliminating the disadvantage of needing to train the codec for different statistical features or different types of channels, ensuring the semantic communication quality of non-orthogonal multi-users when the original codec is applied to the fading channel, and improving communication efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0055] In order to more clearly illustrate the technical solutions in this application or related technologies, the following briefly introduces the drawings required for use in the embodiments or related technical descriptions. Obviously, the drawings described below are merely embodiments of this application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without any creative work.

[0056] Figure 1 is a schematic diagram of the structure of the NOMA semantic communication system with channel migration provided in an embodiment of the present application.

[0057] FIG2 is a flow chart of a semantic communication method with channel migration provided in an embodiment of the present application.

[0058] Figure 3 is a flow chart of a channel-migratable semantic communication method applied to a NOMA system provided in an embodiment of the present application.

[0059] FIG4 is an expanded table of complex matrix variable elements provided in an embodiment of the present application.

[0060] Figure 5 is a schematic diagram of a channel-migratable semantic communication device applied to a NOMA system provided in an embodiment of the present application.

[0061] FIG6 is a schematic diagram of the hardware structure of an electronic device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0062] In order to make the objectives, technical solutions and advantages of this application more clear, this application is further described in detail below in combination with specific embodiments and with reference to the accompanying drawings.

[0063] It should be noted that, unless otherwise defined, the technical terms or scientific terms used in the embodiments of the present application should have the usual meanings understood by people with ordinary skills in the field to which this application belongs. The "first", "second" and similar words used in the embodiments of the present application do not indicate any order, quantity or importance, but are only used to distinguish different components. "Include" or "comprise" and similar words mean that the elements or objects appearing before the word cover the elements or objects listed after the word and their equivalents, without excluding other elements or objects. "Connect" or "connected" and similar words are not limited to physical or mechanical connections, but may include electrical connections, whether direct or indirect. "Up", "down", "left", "right" and the like are only used to indicate relative positional relationships. When the absolute position of the described object changes, the relative positional relationship may also change accordingly.

[0064] As described in the technical background section, the autoencoder model is an unsupervised neural network model in deep learning. It consists of an encoder and a decoder. It can learn the implicit features of input data and use these features to reconstruct the original input data. It is an important tool for implementing joint source-channel coding in the communications field. Autoencoders can be used for nonlinear dimensionality reduction of features and extract more effective new features. Autoencoders are an important tool for semantic communication and play a key role in extracting and reconstructing user semantic information in semantic communication systems.

[0065] Orthogonal Frequency Division Multiplexing (OFDM) is a multi-carrier modulation technology. OFDM enables parallel transmission of high-speed serial data. The main process of OFDM is to divide an entire frequency band into several subcarriers, and make each subcarrier orthogonal to each other so that they do not overlap in the spectrum, which can reduce interference. OFDM operates by inverse Fourier transforming the frequency domain signal into the time domain at the transmitting end and sending it by adding a cyclic prefix. The receiving end removes the cyclic prefix from the signal, then converts the time domain signal back to the frequency domain through Fourier transform and decodes the original transmitted signal.

[0066] Non-Orthogonal Multiple Access (NOMA) achieves resource reuse by allocating different powers to multiple users on the same time-frequency resource, thereby achieving the purpose of non-orthogonal multiple access. The main idea of ​​traditional NOMA technology is to use non-orthogonal transmission with different powers for multiple users at the transmitting end, actively introduce interference information, and eliminate interference through serial interference cancellation receivers at the receiving end. The power corresponding to each frequency domain subchannel and time domain symbol no longer supports only one user, but is shared by multiple users, thus generating multi-access interference on the shared channel. To overcome interference, NOMA technology uses serial interference cancellation technology at the receiving end to detect and eliminate interference from multiple users to ensure normal communication of the system.

[0067] In the process of implementing this application, the applicant discovered that the prior art has studied the power allocation problem of traditional NOMA based on Shannon capacity, and reasonably allocated limited transmission power according to the different needs of different user services to balance the user's capacity and interruption probability. However, this type of research has not yet conducted power allocation and performance analysis on user semantic communication. The prior art also studied the image transmission problem of NOMA semantic communication, but the solution did not study and utilize channel information. In addition, some prior art has studied the channel-adaptive wireless image transmission method that combines the autoencoder model and OFDM technology, but only targets one type of distributed characteristic multipath fading channel, and is not designed for various fading channels. The application generalization effect is poor. When the channel type changes or a characteristic parameter of the channel changes, the originally trained autoencoder model is no longer applicable to the new channel, resulting in a decrease in the quality of semantic communication.

[0068] For this reason, the present application provides a channel-migratable semantic communication method and related equipment, first obtaining the first input information and the second input information of the transmitting end, inputting the first input information into a pre-trained semantic encoder, extracting the semantic features of the first input information, and obtaining a first semantic signal; inputting the second input information into a pre-trained semantic encoder, extracting the semantic features of the second input information, and obtaining a second semantic signal; sending the first semantic signal and the second semantic signal to the first channel, superimposing the signals through the first channel, and obtaining a third semantic signal; performing transmitting end signal processing on the first semantic signal and the second semantic signal according to a predetermined channel coefficient of the second channel, and obtaining a processed third semantic signal; A semantic signal and a processed second semantic signal are transmitted to a second channel for signal superposition to obtain a fourth semantic signal, and the fourth semantic signal is subjected to receiving-end signal processing according to a channel coefficient to obtain a fifth semantic signal, and the fifth semantic signal is optimized with the goal of minimizing the error between the third semantic signal and the fifth semantic signal to obtain an optimized fifth semantic signal; the optimized fifth semantic signal is input into a pre-trained semantic decoder for semantic decoding to obtain first decoding information and second decoding information; wherein, the first decoding information corresponds to the semantics of the first input information; and the second decoding information corresponds to the semantics of the second input information.

[0069] This application combines semantic features with channel gain, with the goal of minimizing the error between the original semantic signal and the reconstructed semantic signal, optimizes the signal processing function at the transmitting end and the signal processing function at the receiving end, solves the technical problem of adaptively adjusting the semantic signal according to the channel gain of different carriers, is conducive to the transmission of the extracted semantic features on multiple carriers with different channel gains, and also solves the technical problem of how to migrate the codec trained under the additive white Gaussian noise channel to other fading channel applications, eliminating the disadvantage of needing to train the codec for different statistical features or different types of channels, ensuring the semantic communication quality of non-orthogonal multi-users when the original codec is applied to the fading channel, and improving communication efficiency.

[0070] The technical solution of the present application is further described in detail below through specific examples.

[0071] Figure 1 is a schematic diagram of the structure of the NOMA semantic communication system with channel migration provided in an embodiment of the present application.

[0072] Referring to Figure 1, in a specific implementation, an uplink NOMA transmission system is studied. Multiple users send information to the base station using non-orthogonal multiple access. Each carrier carries semantic information of multiple users simultaneously. In this system, a semantic encoder is deployed for each transmitting user, and a semantic decoder is deployed for the receiving base station. In a NOMA semantic communication system with channel migration, the NOMA semantic encoder and decoder are trained under an AWGN channel and tested under other fading channels (such as Ricean fading and Rayleigh fading). By optimizing the power allocation of the semantic transmission signal and the processing of the received signal, the end-to-end semantic communication quality is guaranteed, and channel migration from an AWGN channel to a multi-carrier fading channel is achieved in the NOMA semantic communication system.

[0073] FIG2 is a flow chart of a semantic communication method with channel migration provided by an embodiment of the present application. Referring to FIG2 , the semantic communication method with channel migration provided by an embodiment of the present application includes the following steps:

[0074] In step S201, first input information and second input information of a sending end are obtained.

[0075] In step S202 , the first input information is input into a pre-trained semantic encoder, and semantic features of the first input information are extracted to obtain a first semantic signal.

[0076] In step S203, the second input information is input into a pre-trained semantic encoder, and semantic features of the second input information are extracted to obtain a second semantic signal.

[0077] In the specific implementation, it is assumed that there are N users in the semantic communication system, represented as Through the user's own semantic encoder From the input message M of the sender n Extract semantic features.

[0078] In an embodiment of the present application, it is assumed that there are two users, corresponding to two semantic encoders, and the input information of the two users is the first input information and the second input information respectively. The semantic features of the input information can be extracted through the corresponding pre-trained semantic encoders.

[0079] As an optional embodiment, the semantic features of the first input information may be converted into a first complex number corresponding to the carrier; then, the first complex number may be normalized according to a preset power constraint condition to obtain a first semantic signal.

[0080] As an optional embodiment, the semantic features of the second input information may be converted into a second complex number corresponding to the carrier; then, the second complex number may be normalized according to a preset power constraint condition to obtain a second semantic signal.

[0081] Specifically, to facilitate system transmission, the semantic features are represented as complex numbers corresponding to the carrier, i.e. Where K is the number of subcarriers, L is the number of OFDM-NOMA symbols, and each symbol carries K semantic features for transmission. represents the semantic features of user n at the kth subcarrier of the lth OFDM-NOMA symbol. The semantic features of each user’s OFDM-NOMA symbol on all subcarriers are expressed as X n，l , subject to the average power constraint of each user P n Constraint, X n，l Send the normalized semantic signal T n，l , expressed as:

[0082] Step S204: Send the first semantic signal and the second semantic signal to a first channel, and perform signal superposition through the first channel to obtain a third semantic signal.

[0083] Specifically, after the semantic signal passes through the AWGN channel, all semantic signals are superimposed, which can be expressed as:

[0084] Among them, T n ={T n，1 ,…,T n，L}, is the noise matrix, whose element w l，k The variance is The independent and identically distributed complex Gaussian noise term is

[0085] It should be noted that the number of semantic encoders at the transmitting end is not limited to two, and can be flexibly set according to the number of users and actual carrier requirements.

[0086] As an optional embodiment, a semantic encoder and a semantic decoder can be constructed in the following manner: first, the historical input messages of the sending end are obtained as a training set, and the semantic features of the training set are extracted through the initial semantic encoder to obtain a first training signal; the first training signal is sent to the receiving end through a first channel, and the first training signal is parsed through the initial semantic decoder to obtain a second training signal; a loss function is constructed based on the first training signal and the second training signal, and the initial semantic encoder and the initial semantic decoder are trained by minimizing the loss function to obtain a pre-trained semantic encoder and a pre-trained semantic decoder.

[0087] In a specific implementation, the semantic decoder φ deployed at the receiving end base station D The received superposition signal Y is reconstructed and expressed as in is the matrix combination of all user reconstruction information. The loss function between the reconstruction information and the original information is:

[0088] The NOMA semantic encoder / decoder of the system is trained by minimizing the loss function, and the trained encoder / decoder is represented as

[0089] Step S205 , performing transmitter signal processing on the first semantic signal and the second semantic signal according to a predetermined channel coefficient of the second channel to obtain a processed first semantic signal and a processed second semantic signal.

[0090] Refer to Figure 3, which is a flow chart of the channel-migratable semantic communication method applied to the NOMA system provided in an embodiment of the present application.

[0091] In specific implementation, the codec group trained under the AWGN channel cannot be directly applied to the multi-carrier fading channel for the time being. In the NOMA semantic communication system provided in this application, a method for optimizing the multi-carrier power allocation according to the channel gain under the fading channel is proposed, and the semantic signals of the transmitting and receiving ends are processed so that the codec group can also be applied to the fading channel.

[0092] As an optional embodiment, a transmitting-end signal processing function is constructed based on the predetermined channel coefficient of the second channel; a processed first semantic signal is obtained based on the first semantic signal and the transmitting-end signal processing function; and a processed second semantic signal is obtained based on the second semantic signal and the transmitting-end signal processing function.

[0093] Specifically, when the channel type changes, the semantic codec Γ with fixed parameters * The semantic signal T will continue to be generated, but it cannot be used directly as a transmission signal, otherwise the communication quality will be reduced due to the change of the channel. In order to avoid this phenomenon, it is necessary to follow the feedback type C b The channel coefficient of the channel is used to process the semantic signal T, and the new signal obtained is the transmission signal. The first semantic signal and the second semantic signal are subjected to the power allocation function f t (·) After signal processing, the optimized semantic signal is obtained, which is expressed as:

[0094] in, It can represent the processed first semantic signal and the processed second semantic signal, is the channel fading matrix coefficient.

[0095] Step S206: Send the processed first semantic signal and the processed second semantic signal to the second channel, perform signal superposition, and obtain a fourth semantic signal.

[0096] Step S207: Perform receiving-end signal processing on the fourth semantic signal according to the channel coefficient to obtain a fifth semantic signal.

[0097] As an optional embodiment, the superimposed signal of the processed first semantic signal and the processed second semantic signal after passing through the fading channel is expressed as:

[0098] Then through type C b channel, and the received signal Y c , and process the received signal to obtain the input decoder The new signal is expressed as: new =f r (Y c , C b )

[0099] As an optional embodiment, the fourth semantic signal is subjected to receiving-end signal processing according to the channel coefficient to obtain a fifth semantic signal, and a receiving-end signal processing function can be constructed according to the predetermined channel coefficient of the second channel; the fifth semantic signal is obtained according to the fourth semantic signal and the receiving-end signal processing function.

[0100] Specifically, taking the application in NOMA system as an example, it is necessary to design the multi-carrier power allocation function f at the transmitting end t (·) and the receiving end signal amplitude operation function f r (·) expression, the transmitting end multi-carrier power allocation function f t (·) is expressed as:

[0101] in, and They are oriented and The power distribution results.

[0102] The design of the amplitude operation function at the receiving end adheres to the principle of simple parameters and introduces an amplitude scaling parameter α, f r (·) is expressed as:

[0103] Refer to Figure 4, which is a complex matrix variable element expansion table provided in an embodiment of the present application.

[0104] Step S208 : optimizing the fifth semantic signal with the goal of minimizing the error between the third semantic signal and the fifth semantic signal to obtain an optimized fifth semantic signal.

[0105] As an optional embodiment, the transmitting-end signal processing function and the receiving-end signal processing function can be optimized with the goal of minimizing the error between the third semantic signal and the fifth semantic signal; the fifth semantic signal can be adjusted according to the optimized transmitting-end processing function and the optimized receiving-end processing function to minimize the difference between the third semantic signal and the fifth semantic signal, thereby obtaining the optimized fifth semantic signal.

[0106] Specifically, it is known that the trained decoder The signal Y can be reconstructed to make the reconstructed information The error with the original information M is the smallest, so when the signal Y new The smaller the error of the sum signal Y, the smaller the error of the sum signal Y. b The higher the channel information reconstruction quality.

[0107] The channel-migratable semantic communication system proposed in this application processes the signal processing function f at the sending end. r and the receiving end signal processing function f r The design principle is to meet the communication constraints Minimize the signal Y according to the channel feedback information under the condition new The sum signal Y error is expressed as:

[0108] Taking the application in the NOMA system as an example, the channel migration power optimization problem can be determined based on the multi-carrier power allocation function, the signal amplitude operation function and the goal of minimizing the error between the third semantic signal and the fifth semantic signal; the channel migration power optimization problem is subjected to a semantically similar dual transformation to obtain a channel migration power optimization model; the analytical solution of the channel migration power optimization model is determined by the Lagrangian dual method, and the carrier power allocation result at the transmitting end and the amplitude scaling parameters at the receiving end are obtained based on the analytical solution and the preset power constraints.

[0109] In a specific implementation, the channel migration power optimization problem can be refined and modeled according to the complex matrix variable element expansion table:

[0110] in,

[0111] Then, according to the meaning of the optimization problem (P1), a semantically similar dual transformation is performed on the problem to obtain a new problem (P2) that is similar to the original problem, which is expressed as:

[0112] in, is the noise-free term corresponding to y, such as Problem (P2) is a weighted sum power minimization problem. The weighting coefficient β n is the ratio factor that balances the transmission power of each user and is defined as constraint and The purpose is to ensure that the objective function d1 of the problem (P1) is approximately minimized.

[0113] Finally, the Lagrangian dual method is used to obtain the analytical solution of power distribution. and The Lagrange multiplier λ l，k and μ l，k By solving the partial derivatives of the variables simultaneously to zero, the analytical solution expression of power distribution is obtained as follows:

[0114] in,

[0115] According to the power allocation analytical solution and power constraints, calculate the amplitude scaling parameter α under each carrier power constraint n，l , the expression is:

[0116] And take the maximum value to meet the power constraints of all carriers, and obtain the amplitude scaling parameter α=max{α 1，1 ,…,α n，l ,…,α N，L According to the result of the carrier power distribution at the transmitting end and And the receiving end amplitude scaling parameter α, the optimized Y new .

[0117] Step S209: input the optimized fifth semantic signal into a pre-trained semantic decoder for semantic decoding to obtain first decoding information and second decoding information; wherein, the first decoding information corresponds to the semantics of the first input information; and the second decoding information corresponds to the semantics of the second input information.

[0118] As an optional embodiment, the optimized fifth semantic signal is input into a pre-trained semantic decoder to extract the semantic features for the first semantic signal and the semantic features for the second semantic signal contained in the optimized fifth semantic signal; semantic recovery and information reconstruction are performed based on the semantic features for the first semantic signal and the semantic features for the second semantic signal to obtain first decoding information and second decoding information.

[0119] As can be seen from the above, the present application provides a channel-migratable semantic communication method and related equipment, which first obtains the first input information and the second input information of the sending end, inputs the first input information into a pre-trained semantic encoder, extracts the semantic features of the first input information, and obtains a first semantic signal; inputs the second input information into a pre-trained semantic encoder, extracts the semantic features of the second input information, and obtains a second semantic signal; sends the first semantic signal and the second semantic signal to the first channel, performs signal superposition through the first channel, and obtains a third semantic signal; performs sending end signal processing on the first semantic signal and the second semantic signal according to a predetermined channel coefficient of the second channel, and obtains a processed signal. The first semantic signal and the processed second semantic signal are sent to the second channel for signal superposition to obtain a fourth semantic signal, and the fourth semantic signal is subjected to receiving-end signal processing according to the channel coefficient to obtain a fifth semantic signal, and the fifth semantic signal is optimized with the goal of minimizing the error between the third semantic signal and the fifth semantic signal to obtain an optimized fifth semantic signal; the optimized fifth semantic signal is input into a pre-trained semantic decoder for semantic decoding to obtain first decoding information and second decoding information; wherein, the first decoding information corresponds to the semantics of the first input information; and the second decoding information corresponds to the semantics of the second input information.

[0120] This application combines semantic features with channel gain, with the goal of minimizing the error between the original semantic signal and the reconstructed semantic signal, optimizes the signal processing function at the transmitting end and the signal processing function at the receiving end, solves the technical problem of adaptively adjusting the semantic signal according to the channel gain of different carriers, is conducive to the transmission of the extracted semantic features on multiple carriers with different channel gains, and also solves the technical problem of how to migrate the codec trained under the additive white Gaussian noise channel to other fading channel applications, eliminating the disadvantage of needing to train the codec for different statistical features or different types of channels, ensuring the semantic communication quality of non-orthogonal multi-users when the original codec is applied to the fading channel, and improving communication efficiency.

[0121] It should be noted that the method of the embodiment of the present application can be performed by a single device, such as a computer or server. The method of this embodiment can also be applied in a distributed scenario and performed by multiple devices working together. In such a distributed scenario, one of the multiple devices may only perform one or more steps of the method of the embodiment of the present application, and the multiple devices will interact with each other to complete the method.

[0122] It should be noted that the above description is limited to some embodiments of the present application. Other embodiments are within the scope of the appended claims. In some cases, the actions or steps recited in the claims may be performed in an order different from that described in the above embodiments and still achieve the desired results. Furthermore, the processes depicted in the accompanying drawings do not necessarily require the specific order or sequential order shown to achieve the desired results. In certain embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0123] Refer to Figure 5, which is a schematic diagram of a semantic communication device with channel migration provided in an embodiment of the present application.

[0124] Based on the same inventive concept, corresponding to the method provided in any of the above embodiments, the present application further provides a semantic communication device with channel migration, the device comprising:

[0125] The encoding module 501 is configured to obtain first input information and second input information from a transmitting end, input the first input information into a pre-trained semantic encoder, extract semantic features of the first input information, and obtain a first semantic signal; and input the second input information into a pre-trained semantic encoder, extract semantic features of the second input information, and obtain a second semantic signal;

[0126] A first superposition module 502 is configured to send the first semantic signal and the second semantic signal to a first channel, and perform signal superposition through the first channel to obtain a third semantic signal;

[0127] an optimization module 503 configured to perform transmitter signal processing on the first semantic signal and the second semantic signal according to a predetermined channel coefficient of a second channel to obtain a processed first semantic signal and a processed second semantic signal;

[0128] A second superposition module 504 is configured to send the processed first semantic signal and the processed second semantic signal to the second channel, perform signal superposition to obtain a fourth semantic signal, and perform receiving-end signal processing on the fourth semantic signal according to the channel coefficient to obtain a fifth semantic signal;

[0129] an optimization module 505 configured to optimize the fifth semantic signal with the goal of minimizing the error between the third semantic signal and the fifth semantic signal to obtain an optimized fifth semantic signal;

[0130] The decoding module 506 is configured to input the optimized fifth semantic signal into a pre-trained semantic decoder for semantic decoding to obtain first decoding information and second decoding information; wherein, the first decoding information corresponds to the semantics of the first input information; and the second decoding information corresponds to the semantics of the second input information.

[0131] Optionally, the encoding module 501 is further configured to:

[0132] Converting the semantic feature of the first input information into a first complex number corresponding to a carrier;

[0133] The first complex number is normalized according to a preset power constraint condition to obtain the first semantic signal.

[0134] Optionally, the encoding module 501 is further configured to:

[0135] Converting the semantic feature of the second input information into a second complex number corresponding to a carrier;

[0136] The second complex number is normalized according to a preset power constraint condition to obtain the second semantic signal.

[0137] Optionally, the first channel is an additive white Gaussian noise channel.

[0138] Optionally, the second channel is a Rayleigh fading channel or a Ricean fading channel.

[0139] Optionally, the encoding module 501 is further configured to:

[0140] Construct the semantic encoder and the semantic decoder:

[0141] Acquire historical input messages of the sending end as a training set, extract semantic features of the training set through an initial semantic encoder, and obtain a first training signal;

[0142] Sending the first training signal to a receiving end through the first channel, parsing the first training signal through an initial semantic decoder to obtain a second training signal;

[0143] A loss function is constructed according to the first training signal and the second training signal, and the initial semantic encoder and the initial semantic decoder are trained by minimizing the loss function to obtain the pre-trained semantic encoder and the pre-trained semantic decoder.

[0144] Optionally, the optimization module 503 is further configured to:

[0145] constructing a transmitting-end signal processing function according to a predetermined channel coefficient of a second channel;

[0146] Obtaining the processed first semantic signal according to the first semantic signal and the transmitting end signal processing function;

[0147] The processed second semantic signal is obtained according to the second semantic signal and the sending end signal processing function.

[0148] Optionally, the optimization module 503 is further configured to:

[0149] constructing a receiving-end signal processing function according to a predetermined channel coefficient of a second channel;

[0150] The fifth semantic signal is obtained according to the fourth semantic signal and the receiving end signal processing function.

[0151] Optionally, the transmitting end signal processing function is a carrier power allocation function; the receiving end signal processing function is a signal amplitude operation function;

[0152] Optionally, the optimization module 503 is further configured to:

[0153] Optimizing the transmitting-end signal processing function and the receiving-end signal processing function with the goal of minimizing the error between the third semantic signal and the fifth semantic signal;

[0154] The fifth semantic signal is adjusted according to the optimized sending-end processing function and the optimized receiving-end processing function to minimize the difference between the third semantic signal and the fifth semantic signal, thereby obtaining the optimized fifth semantic signal.

[0155] Optionally, the optimization module 503 is further configured to:

[0156] Determining the channel migration power optimization problem according to the multi-carrier power allocation function, the signal amplitude operation function, and the goal of minimizing the error between the third semantic signal and the fifth semantic signal;

[0157] Performing a semantically similar dual transformation on the channel migration power optimization problem to obtain a channel migration power optimization model;

[0158] The analytical solution of the channel migration power optimization model is determined by the Lagrange dual method, and the carrier power allocation result of the transmitting end and the amplitude scaling parameter of the receiving end are obtained according to the analytical solution and the preset power constraint condition.

[0159] Optionally, the decoding module 506 is further configured to:

[0160] Inputting the optimized fifth semantic signal into the pre-trained semantic decoder, extracting the semantic features for the first semantic signal and the semantic features for the second semantic signal contained in the optimized fifth semantic signal;

[0161] Semantic recovery and information reconstruction are performed based on the semantic features of the first semantic signal and the semantic features of the second semantic signal to obtain the first decoding information and the second decoding information.

[0162] For the convenience of description, the above system is described as being divided into various modules according to their functions. Of course, when implementing this application, the functions of each module can be implemented in the same or multiple software and / or hardware.

[0163] The system of the above embodiment is used to implement the corresponding method in any of the above embodiments, and has the beneficial effects of the corresponding method embodiment, which will not be described in detail here.

[0164] Based on the same inventive concept, corresponding to any of the above-mentioned embodiments and methods, the present application also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the method described in any of the above embodiments when executing the program.

[0165] FIG6 shows a more specific schematic diagram of the hardware structure of an electronic device provided in this embodiment. The device may include: a processor 610, a memory 620, an input / output interface 630, a communication interface 640, and a bus 650. The processor 610, the memory 620, the input / output interface 630, and the communication interface 640 are communicatively connected to each other within the device via the bus 650.

[0166] The processor 610 can be implemented using a general-purpose CPU (Central Processing Unit), a microprocessor, an application-specific integrated circuit (ASIC), or one or more integrated circuits, and is used to execute relevant programs to implement the technical solutions provided in the embodiments of this specification.

[0167] The memory 620 can be implemented in the form of ROM (Read Only Memory), RAM (Random Access Memory), static storage devices, dynamic storage devices, etc. The memory 620 can store an operating system and other application programs. When the technical solutions provided in the embodiments of this specification are implemented through software or firmware, the relevant program code is stored in the memory 620 and is called and executed by the processor 610.

[0168] The input / output interface 630 is used to connect an input / output module to implement information input and output. The input / output module can be configured as a component in the device (not shown in the figure) or can be externally connected to the device to provide corresponding functions. Input devices may include a keyboard, mouse, touch screen, microphone, various sensors, etc., and output devices may include a display, speaker, vibrator, indicator light, etc.

[0169] The communication interface 640 is used to connect to a communication module (not shown) to enable communication between the device and other devices. The communication module can communicate via a wired method (such as USB, network cable, etc.) or a wireless method (such as mobile network, WiFi, Bluetooth, etc.).

[0170] The bus 650 comprises a pathway for transmitting information between the various components of the device, such as the processor 610 , the memory 620 , the input / output interface 630 , and the communication interface 640 .

[0171] It should be noted that although the above device only shows the processor 610, the memory 620, the input / output interface 630, the communication interface 640, and the bus 650, in a specific implementation, the device may also include other components necessary for normal operation. In addition, it will be understood by those skilled in the art that the above device may only include the components necessary to implement the embodiments of this specification, and does not necessarily include all the components shown in the figure.

[0172] The electronic device of the above embodiment is used to implement the corresponding method in any of the above embodiments, and has the beneficial effects of the corresponding method embodiment, which will not be described in detail here.

[0173] Based on the same inventive concept, corresponding to any of the above-mentioned embodiments and methods, the present application also provides a non-transitory computer-readable storage medium, wherein the non-transitory computer-readable storage medium stores computer instructions, and the computer instructions are used to enable the computer to execute the method described in any of the above embodiments.

[0174] The above-mentioned non-transitory computer-readable storage medium can be any available medium or data storage device that can be accessed by a computer, including but not limited to magnetic storage (such as floppy disks, hard disks, magnetic tapes, magneto-optical disks (MO)), optical storage (such as CDs, DVDs, BDs, HVDs, etc.), and semiconductor storage (such as ROMs, EPROMs, EEPROMs, non-volatile memories (NAND FLASH), solid-state drives (SSDs)), etc.

[0175] The computer instructions stored in the storage medium of the above embodiment are used to enable the computer to execute the method described in any embodiment of the above exemplary method part, and have the beneficial effects of the corresponding method embodiment, which will not be repeated here.

[0176] Based on the same inventive concept, corresponding to the method described in any of the above embodiments, this application also provides a computer program product comprising computer program instructions. In some embodiments, the computer program instructions can be executed by one or more processors of a computer to cause the computer and / or the processor to perform the method. Corresponding to the execution subject corresponding to each step in each embodiment of the method, the processor executing the corresponding step can belong to the corresponding execution subject.

[0177] The computer program product of the above embodiment is used to enable the computer and / or the processor to execute the method described in any of the above embodiments, and has the beneficial effects of the corresponding method embodiments, which will not be described in detail here.

[0178] It is understandable that before using the technical solutions of each embodiment of this application, the type, scope of use, usage scenarios, etc. of the personal information involved will be informed to the user in an appropriate manner, and the user's authorization will be obtained.

[0179] For example, in response to receiving a user's active request, a prompt message is sent to the user to clearly inform the user that the operation requested will require the acquisition and use of the user's personal information. Thus, the user can independently choose whether to provide personal information to the electronic device, application, server, storage medium, or other software or hardware that performs the operation of the technical solution of this application based on the prompt message.

[0180] As an optional but non-limiting implementation, in response to a user's active request, the prompt information may be sent to the user in the form of a pop-up window, in which the prompt information may be presented in text form. Furthermore, the pop-up window may also contain a selection control for the user to select "agree" or "disagree" to provide personal information to the electronic device.

[0181] It is understandable that the above notification and user authorization process are merely illustrative and do not constitute a limitation on the implementation of this application. Other methods that comply with relevant laws and regulations may also be applied to the implementation of this application.

[0182] It will be understood by those skilled in the art that the embodiments of the present application may be implemented as a system, method, or computer program product. Therefore, the present application may be specifically implemented in the following forms, namely: complete hardware, complete software (including firmware, resident software, microcode, etc.), or a combination of hardware and software, generally referred to herein as a "circuit," "module," or "system." In addition, in some embodiments, the present application may also be implemented in the form of a computer program product in one or more computer-readable media, which includes a computer-readable program code.

[0183] Any combination of one or more computer-readable media can be used. A computer-readable medium can be a computer-readable signal medium or a computer-readable storage medium. A computer-readable storage medium can be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or component, or any combination thereof. More specific examples (non-exhaustive examples) of computer-readable storage media can include, for example: an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination thereof. In this document, a computer-readable storage medium can be any tangible medium containing or storing a program that can be used by or in combination with an instruction execution system, device or device.

[0184] A computer-readable signal medium may include a data signal propagated in baseband or as part of a carrier wave, which carries computer-readable program code. Such propagated data signals may take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. A computer-readable signal medium may also be any computer-readable medium other than a computer-readable storage medium that can transmit, propagate, or transport a program for use by or in conjunction with an instruction execution system, apparatus, or device.

[0185] Program code embodied on a computer readable medium may be transmitted using any appropriate medium, including but not limited to wireless, wireline, optical fiber cable, RF, etc., or any suitable combination of the foregoing.

[0186] The computer program code for performing the operations of the present application can be written in one or more programming languages ​​or a combination thereof, including object-oriented programming languages ​​such as Java, Smalltalk, C++, and conventional procedural programming languages ​​such as "C" or similar programming languages. The program code can be executed entirely on the user's computer, partially on the user's computer, as a separate software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving a remote computer, the remote computer can be connected to the user's computer through any type of network, including a local area network (LAN) or a wide area network (WAN), or can be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0187] It should be understood that each block in the flowchart and / or block diagram, and combinations of blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing device to produce a machine. These computer program instructions are executed by the computer or other programmable data processing device to produce a device that implements the functions / operations specified in the blocks in the flowchart and / or block diagram.

[0188] These computer program instructions can also be stored in a computer-readable medium that enables a computer or other programmable data processing device to operate in a specific manner. In this way, the instructions stored in the computer-readable medium produce a product that includes an instruction device that implements the functions / operations specified in the blocks in the flowchart and / or block diagram.

[0189] Computer program instructions may also be loaded onto a computer, other programmable data processing apparatus, or other device so that a series of operational steps are performed on the computer, other programmable data processing apparatus, or other device to produce a computer-implemented process, such that the instructions executed on the computer or other programmable apparatus provide a process that implements the functions / operations specified in the blocks in the flowchart and / or block diagram.

[0190] Those skilled in the art should understand that the discussion of any of the above embodiments is merely illustrative and is not intended to imply that the scope of the present application (including the claims) is limited to these examples. Within the scope of the present application, the technical features in the above embodiments or different embodiments may be combined, the steps may be implemented in any order, and there are many other variations of the different aspects of the embodiments of the present application as described above, which are not provided in detail for the sake of simplicity.

[0191] In addition, for simplicity of description and discussion, and in order not to make the embodiment of the application difficult to understand, the known power supply / ground connection with integrated circuit (IC) chip and other components may or may not be shown in the accompanying drawings provided. In addition, the device can be shown in the form of a block diagram to avoid making the embodiment of the application difficult to understand, and this also takes into account the following fact, that is, the details of the embodiment of these block diagram devices are highly dependent on the platform to be implemented in the embodiment of the application (that is, these details should be fully within the scope of understanding of those skilled in the art). When specific details (for example, circuit) are set forth to describe exemplary embodiments of the application, it will be apparent to those skilled in the art that the embodiment of the application can be implemented without these specific details or when these specific details are changed. Therefore, these descriptions should be considered to be illustrative rather than restrictive.

[0192] Although the present invention has been described in conjunction with specific embodiments thereof, many alternatives, modifications, and variations of these embodiments will be apparent to those skilled in the art based on the foregoing description. For example, other memory architectures (e.g., dynamic RAM (DRAM)) may utilize the embodiments discussed.

[0193] The embodiments of the present application are intended to cover all such substitutions, modifications, and variations that fall within the broad scope of the appended claims. Therefore, any omissions, modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the embodiments of the present application should be included in the scope of protection of this application.

Claims

1. A semantic communication method, comprising: Acquire first input information and second input information of a sending end; Inputting the first input information into a pre-trained semantic encoder, extracting semantic features of the first input information, and obtaining a first semantic signal; Inputting the second input information into a pre-trained semantic encoder, extracting semantic features of the second input information, and obtaining a second semantic signal; Sending the first semantic signal and the second semantic signal to a first channel, and performing signal superposition through the first channel to obtain a third semantic signal; Performing transmitter signal processing on the first semantic signal and the second semantic signal according to a predetermined channel coefficient of a second channel to obtain a processed first semantic signal and a processed second semantic signal; Sending the processed first semantic signal and the processed second semantic signal to the second channel for signal superposition to obtain a fourth semantic signal; Performing receiving-end signal processing on the fourth semantic signal according to the channel coefficient to obtain a fifth semantic signal; optimizing the fifth semantic signal with the goal of minimizing the error between the third semantic signal and the fifth semantic signal to obtain an optimized fifth semantic signal; The optimized fifth semantic signal is input into a pre-trained semantic decoder for semantic decoding to obtain first decoding information and second decoding information; wherein the first decoding information corresponds to the semantics of the first input information; and the second decoding information corresponds to the semantics of the second input information.

2. The method according to claim 1, wherein: Inputting the first input information into a pre-trained semantic encoder, extracting semantic features of the first input information, and obtaining a first semantic signal includes: Converting the semantic feature of the first input information into a first complex number corresponding to a carrier wave; The first complex number is normalized according to a preset power constraint condition to obtain the first semantic signal.

3. The method according to claim 1, wherein: Inputting the second input information into a pre-trained semantic encoder, extracting semantic features of the second input information, and obtaining a second semantic signal includes: Converting the semantic feature of the second input information into a second complex number corresponding to a carrier wave; The second complex number is normalized according to a preset power constraint condition to obtain the second semantic signal.

4. The method according to claim 1, wherein: The first channel is an additive white Gaussian noise channel.

5. The method according to claim 1, wherein: The second channel is a Rayleigh fading channel.

6. The method according to claim 1, wherein: The second channel is a Rice fading channel.

7. The method according to claim 1, further comprising: Acquire historical input messages of the sending end as a training set, extract semantic features of the training set through an initial semantic encoder, and obtain a first training signal; Sending the first training signal to a receiving end through the first channel, parsing the first training signal through an initial semantic decoder to obtain a second training signal; A loss function is constructed according to the first training signal and the second training signal, and the initial semantic encoder and the initial semantic decoder are trained by minimizing the loss function to obtain the pre-trained semantic encoder and the pre-trained semantic decoder.

8. The method according to claim 1, wherein: The performing transmitting-end signal processing on the first semantic signal and the second semantic signal according to the predetermined channel coefficient of the second channel to obtain the processed first semantic signal and the processed second semantic signal comprises: constructing a transmitting end signal processing function according to a predetermined channel coefficient of a second channel; Obtaining the processed first semantic signal according to the first semantic signal and the sending end signal processing function; The processed second semantic signal is obtained according to the second semantic signal and the sending end signal processing function.

9. The method according to claim 1, wherein: The performing receiving-end signal processing on the fourth semantic signal according to the channel coefficient to obtain a fifth semantic signal comprises: constructing a receiving-end signal processing function according to a predetermined channel coefficient of a second channel; The fifth semantic signal is obtained according to the fourth semantic signal and the receiving end signal processing function.

10. The method according to claim 1, wherein: The optimizing the fifth semantic signal with the goal of minimizing the error between the third semantic signal and the fifth semantic signal to obtain the optimized fifth semantic signal comprises: Optimizing the signal processing function of the transmitting end and the signal processing function of the receiving end with the goal of minimizing the error between the third semantic signal and the fifth semantic signal; The fifth semantic signal is adjusted according to the optimized sending-end processing function and the optimized receiving-end processing function to minimize the difference between the third semantic signal and the fifth semantic signal, thereby obtaining the optimized fifth semantic signal.

11. The method according to any one of claims 8 to 10, wherein: The signal processing function at the transmitting end is a carrier power allocation function; the signal processing function at the receiving end is a signal amplitude operation function.

12. The method according to claim 11, further comprising: Determine the channel migration power optimization problem according to the multi-carrier power allocation function, the signal amplitude operation function and the goal of minimizing the error between the third semantic signal and the fifth semantic signal; Performing a semantically similar dual transformation on the channel migration power optimization problem to obtain a channel migration power optimization model; The analytical solution of the channel migration power optimization model is determined by the Lagrange dual method, and the carrier power allocation result of the transmitting end and the amplitude scaling parameter of the receiving end are obtained according to the analytical solution and the preset power constraint condition.

13. The method according to claim 1, wherein: Inputting the optimized fifth semantic signal into a pre-trained semantic decoder for semantic decoding, and obtaining first decoding information and second decoding information includes: Inputting the optimized fifth semantic signal into the pre-trained semantic decoder, extracting the semantic features for the first semantic signal and the semantic features for the second semantic signal contained in the optimized fifth semantic signal; Semantic recovery and information reconstruction are performed according to the semantic features of the first semantic signal and the semantic features of the second semantic signal to obtain the first decoding information and the second decoding information.

14. A semantic communication device with channel migration, comprising: The encoding module is configured to obtain first input information and second input information of the sending end, and convert the first input information The information is input into a pre-trained semantic encoder, and the semantic features of the first input information are extracted to obtain a first semantic signal; Inputting the second input information into a pre-trained semantic encoder, extracting semantic features of the second input information, and obtaining a second semantic signal; A first superposition module is configured to send the first semantic signal and the second semantic signal to a first channel, and perform signal superposition through the first channel to obtain a third semantic signal; an optimization module configured to perform a transmitting end signal processing on the first semantic signal and the second semantic signal according to a predetermined channel coefficient of a second channel to obtain a processed first semantic signal and a processed second semantic signal; A second superposition module is configured to send the processed first semantic signal and the processed second semantic signal to the second channel, perform signal superposition to obtain a fourth semantic signal, and perform receiving end signal processing on the fourth semantic signal according to the channel coefficient to obtain a fifth semantic signal; an optimization module, configured to optimize the fifth semantic signal with the goal of minimizing an error between the third semantic signal and the fifth semantic signal to obtain an optimized fifth semantic signal; The decoding module is configured to input the optimized fifth semantic signal into a pre-trained semantic decoder for semantic decoding to obtain first decoding information and second decoding information; wherein the first decoding information corresponds to the semantics of the first input information; and the second decoding information corresponds to the semantics of the second input information.

15. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the program, the method according to any one of claims 1 to 13 is implemented.

16. A non-transitory computer-readable storage medium, characterized in that: The non-transitory computer-readable storage medium stores computer instructions, and the computer instructions are used to cause the computer to execute any one of claims 1 to 13.

17. A computer program product, comprising computer program instructions, which, when executed on a computer, cause the computer to execute the method according to any one of claims 1 to 13.

Images