Semantic communication method, device, electronic device and storage medium

By separately processing and encoding semantic information and model parameters, using semantic channels and shared channels to process separately, the transmission efficiency and accuracy of semantic communication in a variable channel environment in the prior art is solved, and the semantic model is realized in 6G mobile communication efficient semantic information transmission in.

CN114885370BActive Publication Date: 2024-11-19BEIJING UNIV OF POSTS & TELECOMM
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202210435012.1
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2022-04-24
Publication Date
2024-11-19
Estimated Expiration
2042-04-24

AI Technical Summary

Technical Problem

Existing semantic communication technologies are difficult to cope with variable channel environments. They usually encode parameters in the semantic model and semantic information for shared channel processing, resulting in limited transmission efficiency and accuracy in variable channel environments.

Method used

A semantic communication method is proposed, by separately processing and encoding semantic information and model parameters, using preset semantic channels and shared channels to process separately, generate semantic pilot signals and data signals, and transmit them through wireless channels, and the receiving end is based on the channel Quality feedback adjusts model parameters to improve transmission efficiency.

Benefits of technology

The semantic information transmission efficiency and accuracy are improved in a variable channel environment, and the model parameters can be dynamically adjusted according to channel quality, thereby improving the efficiency of semantic information extraction in semantic model.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN114885370B_ABST
    Figure CN114885370B_ABST
Patent Text Reader

Abstract

The present application provides a semantic communication method, device, electronic device and storage medium; the method includes: a sending method of semantic communication and a receiving method of semantic communication; wherein the sending method of semantic communication includes: determining a semantic model and setting model parameters according to target data, and extracting semantic information from the target data using the semantic model; encoding the model parameters and the semantic information respectively to obtain model parameter bits and semantic information bits, and processing them using a semantic channel and a shared channel respectively to obtain a semantic pilot signal and a data signal, and sending them to a receiving end, and adjusting the semantic model according to feedback from the receiving end; the receiving method of semantic communication includes: demodulating and decoding the acquired semantic pilot signal and data signal to obtain model parameters and semantic information, selecting a suitable semantic model, inputting the semantic information into the semantic model, obtaining the target data, and feeding back the channel quality to the sending end.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] Embodiments of the present application relate to the field of communication technology, and in particular, to a semantic communication method, device, electronic device, and storage medium. Background Art

[0002] In the context of 6G mobile communications, semantic communication is one of the potential key technologies for realizing 6G communications. Semantic communication requires the use of semantic models, that is, neural networks to perform semantic extraction on the transmitted data, obtain semantic information, and transmit the semantic information and the parameters set in the neural network to the receiving end. However, in related semantic communication technologies, the parameters in the neural network are often encoded together with the semantic information, and a shared channel processing method is adopted together. This method is difficult to cope with the changing channel environment.

[0003] Based on this, a solution is needed that can process and encode semantic information and parameters separately. Summary of the invention

[0004] In view of this, the purpose of the present application is to propose a semantic communication method, device, electronic device and storage medium.

[0005] Based on the above purpose, the present application provides a method for sending semantic communication, which is applied to a sending end, and the method includes:

[0006] Get target data;

[0007] Determine a semantic model according to the target data and set model parameters, and extract semantic information from the target data using the semantic model;

[0008] Encoding the model parameters to obtain model parameter bits, and encoding the semantic information to obtain semantic information bits;

[0009] Using a preset semantic channel to process the model parameter bit to obtain a semantic pilot signal, using a preset shared channel to process the semantic information bit to obtain a data signal, framing the semantic pilot signal and the data signal to obtain a target signal about the target data;

[0010] The target signal is sent to a preset receiving end through a preset wireless channel, and the channel quality of the wireless channel fed back by the receiving end is received. In response to determining that the channel quality is less than a preset signal-to-noise ratio threshold, the model parameters in the semantic model are adjusted, and new semantic information is extracted from the target data again, and a new data signal obtained according to the new semantic information is sent. In response to determining that the channel quality is greater than or equal to the signal-to-noise ratio threshold, the sending process of the target data is completed.

[0011] Further, encoding the model parameters to obtain model parameter bits, and encoding the semantic information to obtain semantic information bits, includes:

[0012] Using a preset mapping relationship, the model parameters are encoded into a bit form of 0 and 1 to obtain model parameter bits;

[0013] By using the mapping relationship, the value range of each vector in the semantic information is encoded into a bit form of 0 and 1 to obtain the semantic information bit.

[0014] Further, the using a preset semantic channel to process the model parameter bit to obtain a semantic pilot signal, using a preset shared channel to process the semantic information bit to obtain a data signal, and framing the semantic pilot signal and the data signal to obtain a target signal about the target data, includes:

[0015] Performing semantic channel coding on the model parameter bits by adopting a low code rate coding method, and performing orthogonal phase shift keying modulation on the result of the semantic channel coding by adopting a low-order modulation method to obtain the semantic pilot signal;

[0016] Performing shared channel coding on the semantic information bits by adopting a low-high code rate coding method, and performing orthogonal amplitude modulation on the result of the shared channel coding by adopting a high-order modulation method to obtain the data signal;

[0017] According to preset control information, the semantic pilot signal and the data signal are arranged into a frame structure to obtain a target signal.

[0018] Based on the same inventive concept, the present application also provides a method for receiving semantic communication, which is applied to a receiving end, and the method includes:

[0019] Receive a target signal from a transmitter through a preset wireless channel;

[0020] Identifying the semantic pilot signal and the data signal from the target signal, decoding the semantic pilot signal through a semantic channel to obtain a model parameter bit, and decoding the data signal through a shared channel to obtain a semantic information bit;

[0021] Decoding the model parameter bits to obtain the model parameter, and decoding the semantic information bits to obtain the semantic information;

[0022] Determine a semantic model corresponding to the model parameters according to the model parameters, use the semantic model to restore the semantic information to obtain target data, judge the data quality of the target data, determine the channel quality of the wireless channel according to the data quality, and feed back the channel quality to the sending end.

[0023] Further, the identifying the semantic pilot signal and the data signal in the data signal from the target signal, decoding the semantic pilot signal through a semantic channel to obtain a model parameter bit, and decoding the data signal through a shared channel to obtain a semantic information bit includes:

[0024] Determine a frame structure in the target signal, determine an arrangement position of the semantic pilot signal and the data signal in the frame structure by using preset control information, and obtain the semantic pilot signal and the data signal according to the arrangement position;

[0025] Performing orthogonal phase shift keying demodulation on the semantic pilot signal in a low-order demodulation manner to obtain a demodulation result of a semantic channel, and decoding the demodulation result of the semantic channel in a low-code rate decoding manner to obtain the model parameter bit;

[0026] The data signal is orthogonally modulated and demodulated in a high-order demodulation manner to obtain a demodulation result of a shared channel, and the demodulation result of the shared channel is decoded in a high-code rate decoding manner to obtain the semantic information bits.

[0027] Further, the bit decoding of the model parameter to obtain the model parameter and the bit decoding of the semantic information to obtain the semantic information include:

[0028] Decoding the bit forms of 0 and 1 in the model parameter bits by using a preset mapping relationship to obtain the model parameters;

[0029] By using the mapping relationship, the bit forms of 0 and 1 in the semantic information bits are decoded into multiple decimals with fixed value ranges, each of the decimals with fixed value ranges is used as a vector, and all the vectors are used as semantic information about the target data.

[0030] Based on the same inventive concept, the present application also provides a sending device for semantic communication, comprising: a first semantic processing module, a first interface module, a first baseband processing module and a first transmission module;

[0031] The first semantic processing module is configured to obtain target data, determine a semantic model and set model parameters according to the target data, and extract semantic information from the target data using the semantic model;

[0032] The first interface module is configured to encode the model parameters to obtain model parameter bits, and encode the semantic information to obtain semantic information bits;

[0033] The first baseband processing module is configured to encode the model parameters to obtain model parameter bits, encode the semantic information to obtain semantic information bits, process the model parameter bits using a preset semantic channel to obtain a semantic pilot signal, process the semantic information bits using a preset shared channel to obtain a data signal, and frame the semantic pilot signal and the data signal to obtain a target signal regarding the target data;

[0034] The first transmission module is configured to send the target signal to a preset receiving end through a preset wireless channel, receive the channel quality of the wireless channel fed back by the receiving end, and in response to determining that the channel quality is less than a preset signal-to-noise ratio threshold, adjust the model parameters in the semantic model, extract new semantic information from the target data again, and send a new data signal obtained according to the new semantic information, and in response to determining that the channel quality is greater than or equal to the signal-to-noise ratio threshold, complete the target data sending process.

[0035] Based on the same inventive concept, the present application also provides a receiving device for semantic communication, including: a second semantic processing module, a second interface module, a second baseband processing module, and a second transmission module;

[0036] Wherein, the second transmission module is configured to receive a target signal sent from a transmitting end from a preset wireless channel;

[0037] The second baseband processing module is configured to identify a semantic pilot signal and a data signal in the data signal from the target signal, decode the semantic pilot signal through a semantic channel to obtain a model parameter bit, and decode the data signal through a shared channel to obtain a semantic information bit;

[0038] The second interface module is configured to decode the model parameter bits to obtain the model parameters, and decode the semantic information bits to obtain the semantic information;

[0039] The second semantic processing module is configured to determine a semantic model corresponding to the model parameters based on the model parameters, restore the semantic information using the semantic model to obtain target data, determine the data quality of the target data, determine the channel quality of the wireless channel based on the data quality, and use the second transmission module to feed back the channel quality to the transmitting end.

[0040] Based on the same inventive concept, the present application also provides an electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the program, it implements the method for receiving semantic communication and / or the method for receiving semantic communication as described in any one of the above items.

[0041] Based on the same inventive concept, 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 above-mentioned semantic communication receiving method and / or semantic communication receiving method.

[0042] From the above, it can be seen that the semantic communication method, device, electronic device and storage medium provided by the present application are based on the encoding of semantic information and model parameters, comprehensively consider the different processing methods of semantic information and model parameters, and encode the semantic information and model parameters separately, so that the processing of semantic information and model parameters is separated, without the need to incorporate model parameters into semantic information, and the corresponding processing process is specifically completed by the semantic channel and shared channel, so as to be able to cope with more changeable situations.

[0043] Furthermore, based on the channel quality feedback mechanism designed in the embodiments of the present application, the semantic model can be set more accurately and effectively according to the changing environment of the communication channel, thereby effectively improving the efficiency of the semantic model in extracting semantic information. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] In order to more clearly illustrate the technical solutions in the present application or related technologies, the drawings required for use in the embodiments or related technical descriptions are briefly introduced below. Obviously, the drawings described below are only embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0045] Figure 1 A flowchart of sending and receiving semantic communication according to an embodiment of the present application;

[0046] Figure 2A A flow chart of a method for sending semantic communication according to an embodiment of the present application;

[0047] Figure 2B A flowchart of a method for receiving semantic communication according to an embodiment of the present application;

[0048] Figure 3A A schematic diagram of a sending device for semantic communication according to an embodiment of the present application;

[0049] Figure 3B A schematic diagram of a receiving device for semantic communication according to an embodiment of the present application;

[0050] Figure 4 This is a schematic diagram of the structure of an electronic device according to an embodiment of the present application. DETAILED DESCRIPTION

[0051] In order to make the objectives, technical solutions and advantages of the present application more clearly understood, the present application is further described in detail below in combination with specific embodiments and with reference to the accompanying drawings.

[0052] It should be noted that, unless otherwise defined, the technical terms or scientific terms used in the embodiments of the present application should be the usual meanings understood by people with ordinary skills in the field to which the present application belongs. The "first", "second" and similar words used in the embodiments of the present application do not represent any order, quantity or importance, but are only used to distinguish different components. "Including" or "comprising" and similar words mean that the elements or objects appearing in front of the word cover the elements or objects listed after the word and their equivalents, without excluding other elements or objects. "Connecting" or "connected" and similar words are not limited to physical or mechanical connections, but can 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.

[0053] As described in the background technology section, related semantic communication methods are still difficult to meet actual needs.

[0054] In the context of 6G mobile communications, semantic communication is one of the potential key technologies for realizing 6G communications. Semantic communication requires the use of a semantic model, that is, a neural network to perform semantic extraction on the transmitted target data to obtain semantic information, and transmit both the semantic information and the model parameters set in the neural network to the receiving end, so that the receiving end can use the model parameters to set the semantic model, and use the semantic information of the set semantic model to recover and obtain the target data sent by the sending end.

[0055] In the process of implementing the present application, the applicant discovered that the main problem with the related semantic communication methods is that in the related semantic communication technologies, the model parameters in the semantic model are often encoded together with the semantic information. For example, during encoding, the model parameters are added to the front or back of the semantic information, and a shared channel processing method is used to transmit the semantic information and model parameters to the receiving end. However, this method is often difficult to cope with a changing channel environment.

[0056] Below, through specific embodiments, and specifically combined with Figure 1 The sending and receiving flow chart of semantic communication is shown to illustrate the technical method of the present application in detail.

[0057] like Figure 1 As shown, the process of semantic communication is based on a preset sender and a preset receiver. After the sender sends relevant data to the receiver, the receiver will estimate the channel quality based on the received data and feed back the channel quality to the sender. After receiving the feedback channel quality, the sender will adjust the related data sent according to the channel quality and send it to the receiver again. The receiver will repeat the above operations until the estimated channel quality reaches the preset standard.

[0058] In an embodiment of the present application, 6G mobile communication is taken as a specific example, wherein a sending end and a receiving end transmit information through wireless communication, and a knowledge base storing a variety of semantic models is configured at both the sending end and the receiving end.

[0059] refer to Figure 2A , a method for sending semantic communication according to an embodiment of the present application comprises the following steps:

[0060] Step S201: Acquire target data, determine a semantic model based on the target data and set model parameters, and extract semantic information from the target data using the semantic model.

[0061] In this step, first, the sending end determines the data type of the obtained target data, wherein the data type of the target data can be one or more of text data, image data, audio data and multi-source data.

[0062] Furthermore, according to the data type of the target data, a suitable semantic model is obtained from the knowledge base.

[0063] The semantic model may be one or more neural networks, such as a convolutional neural network. In this embodiment, the knowledge base is configured at the sending end and is used to store various types of semantic models.

[0064] In this embodiment, the semantic model in the knowledge base can be applicable to different data types. For example, when the target data is text data, a semantic model applicable to text data can be selected.

[0065] Furthermore, when the sending end processes the target data for the first time, the model parameters of the selected semantic model need to be initialized.

[0066] The model parameters may include the type of the semantic model, that is, the type of the neural network, the number of convolutional layers in the neural network, the step size, the connection method of each neural layer in the neural network, etc.

[0067] In this embodiment, when the model parameters of the semantic model are initialized, they may be set according to expert experience, or according to a pre-acquired channel state, that is, channel quality.

[0068] In this embodiment, the signal-to-noise ratio is used to measure the channel quality. In some embodiments, other indicators may also be used to measure the channel quality.

[0069] Furthermore, when the pre-acquired signal-to-noise ratio is smaller, it can be determined that the channel quality is worse. The worse the channel quality is, the more convolutional layers can be set in the model parameters of the semantic model.

[0070] Furthermore, the semantic model that has completed the model parameter initialization is used to perform semantic extraction on the target data, and the extracted semantic information is used as the feature vector of the target data. That is to say, the extracted semantic information can be in the form of a vector. In some embodiments, the extracted semantic information can also be a subgraph of the knowledge graph.

[0071] Step S202: Encode the model parameters to obtain model parameter bits, and encode the semantic information to obtain semantic information bits.

[0072] In this step, based on the model parameters set for the semantic model in the previous step and the extracted semantic information, the model parameters and the semantic information can be encoded respectively to obtain bit forms represented by 0 and 1. The encoding process can also be regarded as a process of encoding the model parameters and semantic information into bit forms according to a preset mapping relationship, where the mapping relationship can be defined according to the specific situation.

[0073] Specifically, in this embodiment, the encoding of model parameters mainly involves the type of semantic model and the number of convolutional layers within it, wherein the type of semantic model may be, for example, a convolutional neural network, a recurrent neural network, and the like.

[0074] In this embodiment, each type of semantic model is pre-numbered in the knowledge base so that each type of semantic model has its own unique number. Further, according to a preset mapping relationship, the number representing the selected semantic model type is encoded into a bit form.

[0075] In a specific example, for example, the selected semantic model number is 1-1, and according to a predetermined binary mapping relationship, it is encoded into binary bits: 00010001, wherein the first 4 bits and the last 4 bits represent the first and last two digits of the number respectively.

[0076] Furthermore, the number of convolutional layers in the selected semantic model is determined, and the number of convolutional layers set according to the channel quality is encoded into a bit form according to a preset mapping relationship.

[0077] In a specific example, for example, the number of convolutional layers is set to 10, which is encoded into binary bits: 1010 according to a predetermined binary mapping relationship.

[0078] Furthermore, all the above-encoded model parameters are merged, and the merged result is defined as the model parameter bit.

[0079] Furthermore, the extracted semantic information is encoded according to the same mapping relationship as the encoding model parameters.

[0080] Specifically, in this embodiment, a vector can be used as a form of existence of semantic information. Based on this, the semantic information can include multiple vectors. For each of the vectors, the value range of the vector is determined, and the value range is evenly divided into multiple sub-regions in a uniformly quantized manner. Further, all sub-regions are encoded into bit forms of 0 and 1 according to the aforementioned mapping relationship.

[0081] In a specific example, the size of the vector can be expressed as a decimal, and the value range of the decimal is fixed. Therefore, its value range can be evenly divided into multiple sub-regions, each sub-region covering a part of the range of the value range, wherein the number of divided sub-regions can be defined according to specific needs.

[0082] Furthermore, the value range covered by each sub-region is determined, and according to the above preset mapping relationship, the value range covered by each sub-region is encoded into a bit form.

[0083] In the aforementioned specific example, each sub-region of the fractional value domain may be encoded into binary bits according to a predetermined binary mapping relationship.

[0084] Furthermore, all vector-encoded bits in the semantic information are obtained and merged, and the merged result is defined as the semantic information bit.

[0085] Step S203: Use a preset semantic channel to process the model parameter bits to obtain a semantic pilot signal, use a preset shared channel to process the semantic information bits to obtain a data signal, and frame the semantic pilot signal and the data signal to obtain a target signal about the target data.

[0086] In this step, based on the model parameter bits and semantic information bits generated above, they are processed using different channels respectively, so as to obtain semantic pilot signals related to the model parameters and data signals related to the semantic information bits respectively.

[0087] Specifically, the model parameter bits are processed using a semantic channel. In the processing method of the semantic channel, a low-code rate coding method is first used for encoding, and then the coding result is modulated. A low-order modulation method can be adopted in the modulation process to improve the reliability of the model parameter bits during transmission and reduce the probability of bit errors during transmission.

[0088] In a specific example, the model parameter bits may be first CRC (cyclic redundancy check) encoded to obtain a first CRC encoding result, and then the first CRC encoding result may be encoded using, for example, a low-rate Turbo code or a low-rate LDPC code (low-density parity check code diagram) to obtain a low-rate encoding result, and the low-rate encoding result may be further subjected to regular randomization processing to obtain a first scrambling code, and then the first scrambling code may be further modulated using a low-order modulation method such as BPSK (binary phase shift keying) or QPSK (quadrature phase shift keying), and the modulated first scrambling code may be layer mapped to map the codewords therein to the corresponding layers of the predetermined time-frequency resources of the transmitting end, and the obtained layer data may be used as the first layer data, and then the first layer data may be further matched to the antenna port of the first baseband processing module, and carrier modulation may be performed using OFDM (orthogonal frequency division multiplexing) technology, and the result after carrier modulation may be defined as a semantic pilot signal.

[0089] Furthermore, the semantic information bits are processed using a shared channel. In the shared channel processing method, a high-code rate encoding method is first used for encoding, and then the encoding result is modulated. In the modulation process, a high-order modulation method can be adopted to increase the semantic information bits that can be accommodated per unit time during transmission, thereby improving the transmission efficiency of the semantic information bits.

[0090] In a specific example, the semantic information bits may be first CRC (cyclic redundancy check) encoded to obtain a second CRC encoding result, and then the second CRC encoding result may be encoded using, for example, a high-rate LDPC code (low-density parity check code diagram) to obtain a high-rate encoding result, and the high-rate encoding result may be further regularly randomized to obtain a second scrambling code, and then the second scrambling code may be further modulated using a high-order modulation method such as 256QAM modulation (quadrature amplitude modulation), and the modulated second scrambling code may be layer mapped to map the codewords therein to the corresponding layers of the predetermined time-frequency resources of the transmitting end, and the obtained layer data may be used as the second layer data, and then the second layer data may be further matched to the antenna port of the first baseband processing module, and carrier modulation may be performed using OFDM (orthogonal frequency division multiplexing) technology, and the result after carrier modulation may be defined as a data signal.

[0091] In this embodiment, the above-mentioned LDPC code (Low Density Parity Check Code) can select a low code rate or a high code rate encoding method according to specific needs.

[0092] Based on the above-mentioned separate processing of model parameter bits and semantic information bits through different channels, the relevant data of model parameters and the relevant data of semantic information can be separated in encoding without adding the relevant data of model parameters to the relevant data of semantic information.

[0093] It should be noted that the content processed by the above-mentioned semantic channel may also be the encoding result in other bit forms. In this embodiment, the content processed by the semantic channel is not limited.

[0094] Furthermore, the semantic pilot signal and the data signal obtained above are framed to obtain a target signal.

[0095] Specifically, first, control information can be pre-set in the high layer of the transmitting end, and the position allocation rules of semantic pilot signals and data signals in the physical layer are defined in the control information, wherein the position allocation in the control information can be set according to the specific situation of the physical layer.

[0096] Furthermore, according to the control information, the semantic pilot signal and the data signal may be arranged according to a predetermined position allocation rule, that is, framed, and the frame structure after framing is used as the target signal.

[0097] Step S204: sending the target signal to a preset receiving end through a preset wireless channel, receiving the channel quality of the wireless channel fed back by the receiving end, and in response to determining that the channel quality is less than a preset signal-to-noise ratio threshold, adjusting the model parameters in the semantic model, extracting new semantic information from the target data again, and sending a new data signal obtained according to the new semantic information; in response to determining that the channel quality is greater than or equal to the signal-to-noise ratio threshold, completing the target data sending process.

[0098] In this step, firstly, based on the target signal obtained in the above steps, it is sent to the receiving end by wireless transmission. Specifically, the target is transmitted by using a wireless channel.

[0099] Furthermore, feedback on the channel quality of the wireless channel sent by the transmitting end is received, and the fed-back channel quality is determined.

[0100] Specifically, based on the setting in the aforementioned step of using the signal-to-noise ratio as a measure of channel quality, a signal-to-noise ratio threshold for channel quality is pre-set, and a channel quality with a signal-to-noise ratio less than the signal-to-noise ratio threshold is defined as unqualified, and a channel quality with a signal-to-noise ratio greater than the signal-to-noise ratio threshold is defined as qualified.

[0101] Furthermore, when the received feedback indicates that the channel quality of the wireless channel is unqualified, the parameters of the semantic model are adjusted according to the channel quality. For example, when the signal-to-noise ratio is too low, multiple convolutional layers can be added to the semantic model, and the adjusted semantic model can be used to extract semantic information from the target data again to obtain new semantic information, and the above steps S202 and S203 are repeated based on the new semantic information to obtain a new target signal.

[0102] Furthermore, the obtained new target signal is sent to a receiving end, and channel quality feedback about the new target signal is received.

[0103] Furthermore, when the received feedback indicates that the channel quality of the wireless channel is qualified, the sending of the semantic information may be terminated.

[0104] refer to Figure 2B , a method for receiving semantic communication according to another embodiment of the present application comprises the following steps:

[0105] Step S205: Receive a target signal sent from the transmitting end through a preset wireless channel.

[0106] In this step, first, a wireless channel for wireless communication is agreed upon with the transmitting end.

[0107] Furthermore, a target signal sent by the sending end is received by utilizing a wireless channel pre-agreed with the sending end.

[0108] Step S206: Identify the semantic pilot signal and the data signal from the target signal, decode the semantic pilot signal through the semantic channel to obtain model parameter bits, and decode the data signal through the shared channel to obtain semantic information bits.

[0109] In this step, the semantic pilot signal and the data set signal in the target signal may be distinguished according to preset control information.

[0110] Specifically, first, control information may be pre-set in the upper layer of the transmitting end, in which the position allocation rules of the semantic pilot signal and the data signal in the physical layer are defined, that is, the arrangement positions of the semantic pilot signal and the data signal.

[0111] In this embodiment, all information such as the position allocation rule in the control information is consistent with the control information set in the sending end that sends the target signal.

[0112] Furthermore, the frame structure of the target signal can be separated according to the position allocation rule in the control information. By separating the control information, the semantic pilot signal and the data signal before framing can be obtained.

[0113] Furthermore, the semantic pilot signal and the data signal are demodulated and decoded respectively to obtain model parameter bits and semantic information bits.

[0114] Specifically, the semantic pilot signal is processed by a semantic channel. In the processing method of the semantic channel, the semantic pilot signal is first demodulated. In the demodulation process, a low-order demodulation method can be adopted. Then, the demodulation result is decoded. In the decoding process, a low-code rate encoding method can be adopted. After decoding, the model parameter bits represented in the bit form of 0 and 1 can be obtained.

[0115] In a specific example, the semantic pilot signal may be first demodulated by OFDM (orthogonal frequency division multiplexing) to obtain a first demodulation result, and deprecoding may be performed according to the first demodulation result, and each first layer of data may be determined to obtain the modulated first scrambling code before layer mapping in the aforementioned embodiment, and then further demodulated. In this embodiment, the demodulation mechanism selected in the demodulation process is consistent with the modulation mechanism of the first scrambling code in the aforementioned embodiment, that is, a low-order demodulation method such as BPSK (binary phase shift keying) or QPSK (quadrature phase shift keying) is used for demodulation to obtain the first scrambling code. Further, the obtained first scrambling code is descrambled to obtain a low code rate coding result, and the low code rate coding result is low-code decoded. In this embodiment, the decoding mechanism selected in the low code rate decoding process is consistent with the coding mechanism in the aforementioned embodiment, that is, a low code rate decoding method such as Turbo code or low code rate LDPC is used for decoding. After low code rate decoding, a first CRC coding result may be obtained, and the first CRC coding result is CRC decoded to obtain the model parameter bit.

[0116] Furthermore, the data number is processed using a shared channel. In the shared channel processing method, the data signal is first demodulated. In the modulation process, a high-order demodulation method can be adopted. The demodulation result is then decoded. In the decoding process, a high-code rate encoding method can be adopted for decoding. After decoding, semantic information bits represented by bits of 0 and 1 can be obtained.

[0117] In a specific example, the data signal may be first demodulated by OFDM (orthogonal frequency division multiplexing) to obtain a second demodulation result, and deprecoding may be performed according to the second demodulation result, and each second layer of data may be determined to obtain the modulated second scrambling code before layer mapping in the aforementioned embodiment, and then further demodulated. In this embodiment, the demodulation mechanism selected in the demodulation process is consistent with the modulation mechanism of the second scrambling code in the aforementioned embodiment, that is, a high-order demodulation method such as 256QAM is used for demodulation to obtain the second scrambling code. Further, the obtained second scrambling code is descrambled to obtain a high code rate coding result, and the high code rate coding result is decoded at a high code rate. In this embodiment, the decoding mechanism selected in the high code rate decoding process is consistent with the coding mechanism in the aforementioned embodiment, that is, a high code rate decoding method such as LDPC is used for decoding. After high code rate decoding, a second CRC coding result may be obtained, and the second CRC coding result is CRC decoded to obtain semantic information bits.

[0118] Step S207: bit-decode the model parameters to obtain model parameters, and bit-decode the semantic information to obtain semantic information.

[0119] In this step, based on the model parameter bits and semantic information bits obtained above, they are decoded respectively to obtain model parameters and semantic information.

[0120] Specifically, in this embodiment, the decoding of model parameters mainly involves the type of semantic model and the number of convolutional layers within it, where the type of semantic model can be, for example, a convolutional neural network, a recurrent neural network, etc.

[0121] In this embodiment, the knowledge base pre-configured in the receiving end includes a semantic model consistent with the sending end, wherein the semantic model can be one or more neural networks, such as a convolutional neural network.

[0122] Furthermore, the bits of the model parameters in the model parameter bits are separated and decoded respectively, and the bits of the model parameters are decoded into model parameters that can be directly input into the semantic model.

[0123] In this embodiment, each model parameter mainly includes: a semantic model type and the number of convolutional layers within it, wherein the semantic model type may be, for example, a convolutional neural network, a recurrent neural network, or the like.

[0124] Wherein, based on the pre-numbers configured in advance in the knowledge base for each type of semantic model, each pre-number corresponds to a semantic model type.

[0125] Further, the binary bits of the semantic model type are determined, and decoded into the corresponding pre-number in the knowledge base according to the binary mapping relationship consistent with the sending end. In a specific example, for example, the binary bits of the semantic model type are: 00010001, and the first 4 bits of the binary bits of the semantic model type can be decoded as 1 and the last 4 bits can be decoded as 1 according to the binary mapping relationship consistent with the sending end, and the pre-number of the semantic model type in the knowledge base can be obtained as 1-1.

[0126] Furthermore, the number of convolutional layers in the selected semantic model is determined, and the number of convolutional layers set according to the channel quality is encoded into a bit form according to a preset mapping relationship.

[0127] In a specific example, for example, the binary bits of the number of convolutional layers are: 1010. The binary bits of the number of convolutional layers can be decoded as 10 according to the binary mapping relationship consistent with the sending end, which means that the number of convolutional layers in the semantic model is 10.

[0128] Furthermore, the decoded semantic model type and the number of convolutional layers are used together as model parameters of the semantic model.

[0129] Furthermore, the semantic information bits are decoded according to a mapping relationship consistent with the sending end to obtain semantic information that can be recognized by the semantic model.

[0130] Specifically, in this embodiment, there may be multiple groups of semantic information bits, each group of semantic information bits represents the value domain of a vector, that is, the value range, wherein each bit represents a sub-region in the value domain, and each bit is decoded according to a mapping relationship consistent with the transmitting end to obtain each sub-region of the value domain, and each sub-region is merged to determine the entire value domain of the vector in terms of size, wherein the vector represents a semantic information.

[0131] Furthermore, each group of semantic information bits is decoded in the above manner to obtain the entire semantic information.

[0132] Step S208, determine the semantic model corresponding to the model parameters according to the model parameters, use the semantic model to restore the semantic information to obtain target data, judge the data quality of the target data, determine the channel quality of the wireless channel according to the data quality, and feed back the channel quality to the sending end.

[0133] In this step, based on the model parameters and semantic information obtained above, the semantic model can be set, and the semantic information can be restored to the target data using the set preset model.

[0134] Specifically, a suitable semantic model is selected from the knowledge base according to the model parameters. As mentioned above, the semantic model type can be one or more neural networks, wherein different semantic model types can be applicable to one or more of text data, image data, audio data and multi-source data.

[0135] Furthermore, the model parameters are input into the selected semantic model to complete the setting of the semantic model, and further, the semantic information is input into the set semantic model for restoration, and the restored target data is obtained using the semantic model.

[0136] Further, based on the obtained target data, the channel quality of the wireless channel for receiving the target signal is determined.

[0137] Specifically, the target data may be one or more of text data, image data, audio data, and multi-source data. Therefore, the data quality may be judged by different indicators according to different data types.

[0138] In a specific example, for example, for image data, the same indicator can be used to judge the data quality of text data, image data, and audio data, for example, Block Error Rate (BLE) and the like.

[0139] It should be noted that in some other embodiments, data quality may also be judged by different indicators. For example, the quality of video data may be judged by the clarity of the video image; the quality of audio data may be judged by the degree of voice distortion, etc.

[0140] Based on the judgment of data quality, the channel quality can be inferred. It can be seen that the better the data quality, the better the channel quality. Based on the setting of measuring channel quality with signal-to-noise ratio in the above embodiment, it can be concluded that the better the data quality, the higher the signal-to-noise ratio, and the worse the data quality, the worse the signal-to-noise ratio. The proportional relationship between data quality and signal-to-noise ratio can be set according to specific circumstances or expert experience.

[0141] Furthermore, the obtained signal-to-noise ratio is sent to the transmitting end to feed back the channel quality thereto.

[0142] It can be seen that the semantic communication method of the embodiment of the present application, based on the encoding of semantic information and model parameters, comprehensively considers the different processing methods of semantic information and model parameters, and encodes the semantic information and model parameters respectively, so that the processing of semantic information and model parameters is separated, without having to incorporate the model parameters into the semantic information, and the corresponding processing process is specifically completed by the semantic channel and the shared channel, so as to be able to cope with more changeable situations.

[0143] Furthermore, based on the channel quality feedback mechanism designed in the embodiments of the present application, the semantic model can be set more accurately and effectively according to the changing environment of the communication channel, thereby effectively improving the efficiency of the semantic model in extracting semantic information.

[0144] 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 to a distributed scenario and completed by multiple devices cooperating with each other. In the case of such a distributed scenario, one of the multiple devices can only perform one or more steps in the method of the embodiment of the present application, and the multiple devices will interact with each other to complete the described method.

[0145] It should be noted that the above describes some embodiments of the present application. Other embodiments are within the scope of the appended claims. In some cases, the actions or steps recorded in the claims can be performed in an order different from that in the above embodiments and still achieve the desired results. In addition, the processes depicted in the accompanying drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0146] Based on the same inventive concept, corresponding to any of the above-mentioned embodiment methods, an embodiment of the present application also provides a sending device for semantic communication.

[0147] refer to Figure 3A , the semantic communication sending device comprises: a first semantic processing module, a first interface module, a first baseband processing module and a first transmission module;

[0148] The first semantic processing module 301 is configured to obtain target data, determine a semantic model and set model parameters according to the target data, and extract semantic information from the target data using the semantic model;

[0149] The first interface module 302 is configured to encode the model parameters to obtain model parameter bits, and encode the semantic information to obtain semantic information bits;

[0150] The first baseband processing module 303 is configured to encode the model parameters to obtain model parameter bits, encode the semantic information to obtain semantic information bits, process the model parameter bits using a preset semantic channel to obtain a semantic pilot signal, process the semantic information bits using a preset shared channel to obtain a data signal, and frame the semantic pilot signal and the data signal to obtain a target signal regarding the target data;

[0151] The first transmission module 304 is configured to send the target signal to a preset receiving end through a preset wireless channel, receive the channel quality of the wireless channel fed back by the receiving end, and in response to determining that the channel quality is less than a preset signal-to-noise ratio threshold, adjust the model parameters in the semantic model, extract new semantic information from the target data again, and send a new data signal obtained according to the new semantic information. In response to determining that the channel quality is greater than or equal to the signal-to-noise ratio threshold, complete the target data sending process.

[0152] As an optional embodiment, the first interface module 302 is specifically configured as follows:

[0153] Using a preset mapping relationship, the model parameters are encoded into a bit form of 0 and 1 to obtain model parameter bits;

[0154] By using the mapping relationship, the value range of each vector in the semantic information is encoded into a bit form of 0 and 1 to obtain the semantic information bit.

[0155] As an optional embodiment, the first baseband processing module 303 is specifically configured as follows:

[0156] Performing semantic channel coding on the model parameter bits by adopting a low code rate coding method, and performing orthogonal phase shift keying modulation on the result of the semantic channel coding by adopting a low-order modulation method to obtain the semantic pilot signal;

[0157] Performing shared channel coding on the semantic information bits by adopting a low-high code rate coding method, and performing orthogonal amplitude modulation on the result of the shared channel coding by adopting a high-order modulation method to obtain the data signal;

[0158] According to preset control information, the semantic pilot signal and the data signal are arranged into a frame structure to obtain a target signal.

[0159] Based on the same inventive concept, corresponding to any of the above-mentioned embodiment methods, an embodiment of the present application also provides a receiving device for semantic communication.

[0160] refer to Figure 3B , the semantic communication receiving device comprises: a second semantic processing module, a second interface module, a second baseband processing module and a second transmission module;

[0161] The second transmission module 305 is configured to receive a target signal sent from a transmitting end through a preset wireless channel;

[0162] The second baseband processing module 306 is configured to identify the semantic pilot signal and the data signal in the data signal from the target signal, decode the semantic pilot signal through the semantic channel to obtain the model parameter bit, and decode the data signal through the shared channel to obtain the semantic information bit;

[0163] The second interface module 307 is configured to decode the model parameter bits to obtain the model parameters, and decode the semantic information bits to obtain the semantic information;

[0164] The second semantic processing module 308 is configured to determine a semantic model corresponding to the model parameters based on the model parameters, restore the semantic information using the semantic model to obtain target data, determine the data quality of the target data, determine the channel quality of the wireless channel based on the data quality, and use the second transmission module to feed back the channel quality to the transmitting end.

[0165] As an optional embodiment, the second baseband processing module 306 is specifically configured as follows:

[0166] Determine a frame structure in the target signal, determine an arrangement position of the semantic pilot signal and the data signal in the frame structure by using preset control information, and obtain the semantic pilot signal and the data signal according to the arrangement position;

[0167] Performing orthogonal phase shift keying demodulation on the semantic pilot signal in a low-order demodulation manner to obtain a demodulation result of a semantic channel, and decoding the demodulation result of the semantic channel in a low-code rate decoding manner to obtain the model parameter bit;

[0168] The data signal is orthogonally modulated and demodulated in a high-order demodulation manner to obtain a demodulation result of a shared channel, and the demodulation result of the shared channel is decoded in a high-code rate decoding manner to obtain the semantic information bits.

[0169] As an optional embodiment, the second interface module 307 is specifically configured as follows:

[0170] Decoding the bit forms of 0 and 1 in the model parameter bits by using a preset mapping relationship to obtain the model parameters;

[0171] By using the mapping relationship, the bit forms of 0 and 1 in the semantic information bits are decoded into multiple decimals with fixed value ranges, each of the decimals with fixed value ranges is used as a vector, and all the vectors are used as semantic information about the target data.

[0172] For the convenience of description, the above devices are described in terms of functions and are divided into various modules. Of course, when implementing the embodiments of the present application, the functions of each module can be implemented in the same or multiple software and / or hardware.

[0173] The device of the above embodiment is used to implement the corresponding semantic communication sending method and / or semantic communication receiving method in any of the above embodiments, and has the beneficial effects of the corresponding method embodiments, which will not be repeated here.

[0174] Based on the same inventive concept, corresponding to any of the above-mentioned embodiments and methods, an embodiment of 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 when the processor executes the program, it implements the method for sending semantic communication and / or the method for receiving semantic communication as described in any of the above embodiments.

[0175] Figure 4 A more specific schematic diagram of the hardware structure of an electronic device provided in this embodiment is shown, and the device may include: a processor 1010, a memory 1020, an input / output interface 1030, a communication interface 1040, and a bus 1050. The processor 1010, the memory 1020, the input / output interface 1030, and the communication interface 1040 are connected to each other through the bus 1050 in the device.

[0176] The processor 1010 can be implemented by 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 the present application.

[0177] The memory 1020 may be implemented in the form of ROM (Read Only Memory), RAM (Random Access Memory), static storage device, dynamic storage device, etc. The memory 1020 may store an operating system and other application programs. When the technical solution provided in the embodiment of the present application is implemented by software or firmware, the relevant program code is stored in the memory 1020 and is called and executed by the processor 1010.

[0178] The input / output interface 1030 is used to connect the input / output module to realize information input and output. The input / output module can be configured in the device as a component (not shown in the figure), or it can be externally connected to the device to provide corresponding functions. The input device may include a keyboard, a mouse, a touch screen, a microphone, various sensors, etc., and the output device may include a display, a speaker, a vibrator, an indicator light, etc.

[0179] The communication interface 1040 is used to connect a communication module (not shown) to realize communication interaction between the device and other devices. The communication module can realize communication through a wired mode (such as USB, network cable, etc.) or a wireless mode (such as mobile network, WIFI, Bluetooth, etc.).

[0180] The bus 1050 includes a path that transmits information between the various components of the device (eg, the processor 1010, the memory 1020, the input / output interface 1030, and the communication interface 1040).

[0181] It should be noted that, although the above device only shows the processor 1010, the memory 1020, the input / output interface 1030, the communication interface 1040 and the bus 1050, in the specific implementation process, the device may also include other components necessary for normal operation. In addition, it can be understood by those skilled in the art that the above device may also only include the components necessary for implementing the embodiment of the present application, and does not necessarily include all the components shown in the figure.

[0182] The device of the above embodiment is used to implement the corresponding semantic communication sending method and / or semantic communication receiving method in any of the above embodiments, and has the beneficial effects of the corresponding method embodiments, which will not be repeated here.

[0183] Based on the same inventive concept, corresponding to any of the above-mentioned embodiment 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 semantic communication sending method and / or the semantic communication receiving method as described in any of the above embodiments.

[0184] The computer-readable medium of this embodiment includes permanent and non-permanent, removable and non-removable media, and information storage can be implemented by any method or technology. Information can be computer-readable instructions, data structures, modules of programs, or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, read-only compact disk read-only memory (CD-ROM), digital versatile disk (DVD) or other optical storage, magnetic cassettes, tape disk storage or other magnetic storage devices or any other non-transmission media that can be used to store information that can be accessed by a computing device.

[0185] The computer instructions stored in the storage medium of the above embodiments are used to enable the computer to execute the semantic communication sending method and / or semantic communication receiving method described in any of the above embodiments, and have the beneficial effects of the corresponding method embodiments, which will not be repeated here.

[0186] 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. In line with the concept 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 different aspects of the embodiments of the present application as described above, which are not provided in detail for the sake of simplicity.

[0187] In addition, to simplify the description and discussion, and in order not to make the embodiments of the present application difficult to understand, the known power supply / ground connection with the integrated circuit (IC) chip and other components may or may not be shown in the provided drawings. In addition, the device can be shown in the form of a block diagram to avoid making the embodiments of the present application difficult to understand, and this also takes into account the fact that the details of the implementation of these block diagram devices are highly dependent on the platform on which the embodiments of the present application will be implemented (that is, these details should be fully within the scope of understanding of those skilled in the art). In the case of elaborating specific details (e.g., circuits) to describe exemplary embodiments of the present application, it is obvious to those skilled in the art that the embodiments of the present application can be implemented without these specific details or when these specific details are changed. Therefore, these descriptions should be considered illustrative rather than restrictive.

[0188] Although the present application has been described in conjunction with specific embodiments of the present application, many replacements, modifications and variations of these embodiments will be apparent to those skilled in the art from the foregoing description. For example, other memory architectures (e.g., dynamic RAM (DRAM)) may use the embodiments discussed.

[0189] 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 protection scope of the present application.

Claims

1. A method for sending semantic communication, characterized in that: Applied to the sending end, the method includes: Acquire target data, determine a semantic model and set model parameters according to the target data, and extract semantic information from the target data using the semantic model; Encoding the model parameters to obtain model parameter bits, and encoding the semantic information to obtain semantic information bits; Using a preset semantic channel to process the model parameter bit to obtain a semantic pilot signal, using a preset shared channel to process the semantic information bit to obtain a data signal, framing the semantic pilot signal and the data signal to obtain a target signal about the target data; The target signal is sent to a preset receiving end through a preset wireless channel, and the channel quality of the wireless channel fed back by the receiving end is received. In response to determining that the channel quality is less than a preset signal-to-noise ratio threshold, adjusting the model parameters in the semantic model, extracting new semantic information from the target data again, and sending a new target signal obtained according to the new semantic information, In response to determining that the channel quality is greater than or equal to the signal-to-noise ratio threshold, the target data sending process is completed.

2. The method according to claim 1, characterized in that The encoding of the model parameters to obtain model parameter bits, and the encoding of the semantic information to obtain semantic information bits, comprises: Using a preset mapping relationship, the model parameters are encoded into a bit form of 0 and 1 to obtain model parameter bits; By using the mapping relationship, the value range of each vector in the semantic information is encoded into a bit form of 0 and 1 to obtain the semantic information bit.

3. The method according to claim 1, characterized in that The method of using a preset semantic channel to process the model parameter bits to obtain a semantic pilot signal, using a preset shared channel to process the semantic information bits to obtain a data signal, and framing the semantic pilot signal and the data signal to obtain a target signal about the target data includes: Performing semantic channel coding on the model parameter bits by adopting a low code rate coding method, and performing orthogonal phase shift keying modulation on the result of the semantic channel coding by adopting a low-order modulation method to obtain the semantic pilot signal; Performing shared channel coding on the semantic information bits by adopting a low-high code rate coding method, and performing orthogonal amplitude modulation on the result of the shared channel coding by adopting a high-order modulation method to obtain the data signal; According to preset control information, the semantic pilot signal and the data signal are arranged into a frame structure to obtain the target signal.

4. A method for receiving semantic communication, characterized in that: Applied to the receiving end, the method comprises: Receive a target signal from a transmitter through a preset wireless channel; Identifying a semantic pilot signal and a data signal from the target signal, decoding the semantic pilot signal through a semantic channel to obtain a model parameter bit, and decoding the data signal through a shared channel to obtain a semantic information bit; Decoding the model parameter bits to obtain the model parameter, and decoding the semantic information bits to obtain the semantic information; Determine a semantic model corresponding to the model parameters according to the model parameters, use the semantic model to restore the semantic information to obtain target data, judge the data quality of the target data, determine the channel quality of the wireless channel according to the data quality, and feed back the channel quality to the sending end.

5. The method according to claim 4, characterized in that The step of identifying the semantic pilot signal and the data signal in the data signal from the target signal, decoding the semantic pilot signal through a semantic channel to obtain a model parameter bit, and decoding the data signal through a shared channel to obtain a semantic information bit includes: Determine a frame structure in the target signal, determine an arrangement position of the semantic pilot signal and the data signal in the frame structure by using preset control information, and obtain the semantic pilot signal and the data signal according to the arrangement position; Performing orthogonal phase shift keying demodulation on the semantic pilot signal in a low-order demodulation manner to obtain a demodulation result of a semantic channel, and decoding the demodulation result of the semantic channel in a low-code rate decoding manner to obtain the model parameter bit; The data signal is orthogonally modulated and demodulated in a high-order demodulation manner to obtain a demodulation result of a shared channel, and the demodulation result of the shared channel is decoded in a high-code rate decoding manner to obtain the semantic information bits.

6. The method according to claim 4, characterized in that The bit decoding of the model parameter to obtain the model parameter and the bit decoding of the semantic information to obtain the semantic information include: Decoding the bit forms of 0 and 1 in the model parameter bits by using a preset mapping relationship to obtain the model parameters; By using the mapping relationship, the bit forms of 0 and 1 in the semantic information bits are decoded into multiple decimals with fixed value ranges, each of the decimals with fixed value ranges is used as a vector, and all the vectors are used as semantic information about the target data.

7. A sending device for semantic communication, comprising: A first semantic processing module, a first interface module, a first baseband processing module and a first transmission module; The first semantic processing module is configured to obtain target data, determine a semantic model and set model parameters according to the target data, and extract semantic information from the target data using the semantic model; The first interface module is configured to encode the model parameters to obtain model parameter bits, and encode the semantic information to obtain semantic information bits; The first baseband processing module is configured to encode the model parameters to obtain model parameter bits, encode the semantic information to obtain semantic information bits, process the model parameter bits using a preset semantic channel to obtain a semantic pilot signal, process the semantic information bits using a preset shared channel to obtain a data signal, and frame the semantic pilot signal and the data signal to obtain a target signal regarding the target data; The first transmission module is configured to send the target signal to a preset receiving end through a preset wireless channel, receive the channel quality of the wireless channel fed back by the receiving end, and in response to determining that the channel quality is less than a preset signal-to-noise ratio threshold, adjust the model parameters in the semantic model, extract new semantic information from the target data again, and send a new target signal obtained according to the new semantic information. In response to determining that the channel quality is greater than or equal to the signal-to-noise ratio threshold, complete the target data sending process.

8. A receiving device for semantic communication, comprising: A second semantic processing module, a second interface module, a second baseband processing module and a second transmission module; Wherein, the second transmission module is configured to receive a target signal sent from a transmitting end from a preset wireless channel; The second baseband processing module is configured to identify a semantic pilot signal and a data signal in the target signal from the target signal, decode the semantic pilot signal through a semantic channel to obtain a model parameter bit, and decode the data signal through a shared channel to obtain a semantic information bit; The second interface module is configured to decode the model parameter bits to obtain the model parameters, and decode the semantic information bits to obtain the semantic information; The second semantic processing module is configured to determine a semantic model corresponding to the model parameters based on the model parameters, restore the semantic information using the semantic model to obtain target data, determine the data quality of the target data, determine the channel quality of the wireless channel based on the data quality, and use the second transmission module to feed back the channel quality to the transmitting end.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable by the processor, characterized in that: When the processor executes the computer program, the method according to any one of claims 1 to 3 or any one of claims 5 to 7 is implemented.

10. A non-transitory computer-readable storage medium, characterized in that: The non-transitory computer-readable storage medium stores computer instructions for causing a computer to execute the method according to any one of claims 1 to 3 or any one of 5 to 7.