Semantic communication method and system based on unified semantic transmission index feedback regulation

Abstract

The embodiment of the application provides a semantic communication method and system based on unified semantic transmission index feedback adjustment, which comprises the following steps: receiving a semantic symbol sent by a semantic transmitter; demodulating the semantic symbol to obtain a received characteristic value; determining a wireless semantic transmission performance index based on the received characteristic value and a characteristic value extracted from source data; and sending a feedback message to the semantic transmitter based on the wireless semantic transmission performance index, so that the semantic transmitter retransmits data by using a combination of high-layer and low-order modulation according to the feedback message. The semantic communication method and system based on unified semantic transmission index feedback adjustment provided by the embodiment of the application propose a unified wireless semantic transmission performance index, which is used for a communication enterprise to evaluate the transmission performance of semantic information, and does not depend on the semantic extraction result of a data source in any mode or task, and has higher applicability.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of communication, and in particular to a semantic communication method and system based on unified semantic transmission index feedback adjustment. BACKGROUND

[0002] Semantic communication is considered a revolutionary communication technology for 6G, which can greatly reduce the amount of data transmission. Most of the research on semantic communication is to design an artificial intelligence-based model to replace the traditional baseband processing module (such as encoder / decoder, modulator / demodulator, etc.). Although some progress has been made in verifying the potential of semantic communication in reducing data transmission, the existing technology has different transmission performance evaluation indexes for information of different modes or different tasks when using semantic communication. For example, the index for image transmission is the peak signal-to-noise ratio (PSNR) or structural similarity (SSIM), and the index for text transmission is the block error ratio (BLER). So far, there is no unified performance index for evaluating the transmission performance of any mode data or task. And the current semantic communication research only considers the joint training of semantic extraction and transmission process, which will lead to cross-industry data training of communication enterprises and service providers in reality, which will hinder the widespread adoption of semantic communication by communication enterprises in real-world applications. SUMMARY

[0003] The embodiments of the present application provide a semantic communication method and system based on unified semantic transmission index feedback adjustment to solve the technical problem of low applicability of transmission performance evaluation indexes in related technologies.

[0004] In a first aspect, the embodiments of the present application provide a semantic communication system based on unified semantic transmission index feedback adjustment, comprising:

[0005] a semantic encoder, a semantic transmitter, a semantic receiver, and a semantic decoder;

[0006] The semantic encoder is configured to determine a feature value according to source data, and send binary data to the semantic transmitter based on the feature value;

[0007] The semantic transmitter is configured to send a semantic symbol to the semantic receiver based on the binary data, the semantic symbol being determined based on a wireless semantic transmission performance index, the wireless semantic transmission performance index being used to represent the semantic difference between the feature value extracted from the source data and the feature value received by the semantic receiver;

[0008] The semantic receiver is configured to convert the received semantic symbol into the binary data and send the binary data to the semantic decoder.

[0009] The semantic decoder is configured to reconstruct the source data based on the binary data.

[0010] In some embodiments, the semantic encoder comprises a semantic extraction module and a binary quantization module.

[0011] The semantic extraction module is configured to extract the feature value from the source data.

[0012] The binary quantization module is configured to convert the feature value into binary data.

[0013] In some embodiments, the semantic transmitter comprises an error-free feature recovery module and a semantic-oriented modulation module.

[0014] The error-free feature recovery module is configured to convert the binary data into the feature value.

[0015] The semantic-oriented modulation module is configured to generate the semantic symbol according to a feedback message and the feature value, wherein the feedback message is determined to be sent based on the wireless semantic transmission performance index.

[0016] In some embodiments, the semantic receiver comprises a semantic-oriented demodulation module and a binary quantization module.

[0017] The semantic-oriented demodulation module is configured to demodulate the semantic symbol into the feature value.

[0018] The binary quantization module is configured to convert the feature value into binary data.

[0019] In some embodiments, the semantic decoder comprises an error-free feature recovery module and a semantic recovery module.

[0020] The error-free feature recovery module is configured to convert the binary data into the feature value.

[0021] The semantic recovery module is configured to reconstruct the source data according to the feature value.

[0022] In a second aspect, the embodiments of the present application provide a semantic communication method based on unified semantic transmission index feedback adjustment, applied to a semantic receiver, comprising:

[0023] Receiving a semantic symbol sent by a semantic transmitter;

[0024] Demodulating the semantic symbol to obtain a received feature value;

[0025] determine a wireless semantic transmission performance index based on the received feature value and the feature value extracted from the source data;

[0026] send a feedback message to the semantic transmitter based on the wireless semantic transmission performance index.

[0027] In some embodiments, the formula for determining the wireless semantic transmission performance index based on the received feature value and the feature value extracted from the source data is:

[0028]

[0029] wherein SeER represents the wireless semantic transmission performance index, N represents the total number of semantic features, X i and Y i are the i-th expected vector and received vector respectively, (·) represents the dot product of vectors, |X i | and |Y i | represent the lengths of vectors X i and Y i respectively, and p represents the semantic difference threshold of vector transmission.

[0030] In some embodiments, the sending of the feedback message to the semantic transmitter based on the wireless semantic transmission performance index comprises:

[0031] sending the feedback message to the semantic transmitter when the wireless semantic transmission performance index is greater than the semantic difference threshold.

[0032] In a third aspect, the embodiments of the present application provide a semantic communication method based on unified semantic transmission index feedback adjustment, applied to a semantic transmitter, comprising:

[0033] receiving a feedback message sent by a semantic receiver;

[0034] modulating to generate a semantic symbol based on the feedback message;

[0035] sending the semantic symbol to the semantic receiver.

[0036] In some embodiments, the modulating to generate a semantic symbol based on the feedback message comprises:

[0037] extracting a normalized feature vector from the source data based on normalization processing, wherein each two normalized feature values in the normalized feature vector represent a symbol point of the semantic symbol;

[0038] generating the semantic symbol based on the normalized feature vector.

[0039] The semantic communication method and system based on unified semantic transmission index feedback adjustment provided by the embodiments of the present application propose a unified wireless semantic transmission performance index, which can fill the blank of the performance index that has not been provided yet and does not distinguish data modes or tasks to measure only the semantic transmission performance. The index can be used for a communication enterprise to evaluate the transmission performance of semantic information without relying on the semantic extraction result of a data source in any mode or task, and has higher applicability. BRIEF DESCRIPTION OF DRAWINGS

[0040] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the related art, the drawings needed to be used in the embodiments or the related art description will be briefly introduced. Obviously, the drawings in the following description are some embodiments of the present application, and other drawings can also be obtained by those skilled in the art without creative labor.

[0041] Figure 1 is a structural schematic diagram of the semantic communication system based on unified semantic transmission index feedback adjustment provided by the embodiments of the present application;

[0042] Figure 2 is one of the flow schematic diagrams of the semantic communication method based on unified semantic transmission index feedback adjustment provided by the embodiments of the present application;

[0043] Figure 3 is a double-layer structure schematic diagram of the 16-order mapping scheme provided by the embodiments of the present application;

[0044] Figure 4 is the second flow schematic diagram of the semantic communication method based on unified semantic transmission index feedback adjustment provided by the embodiments of the present application. DETAILED DESCRIPTION

[0045] In order to make the purpose, technical solutions and advantages of the embodiments of the present application more clear, the technical solutions in the embodiments of the present application will be clearly and completely described below with the drawings in the embodiments of the present application. Obviously, the described embodiments are only some of the embodiments of the present application, not all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor are within the scope of protection of the present application.

[0046] Figure 1 is a structural schematic diagram of the semantic communication system based on unified semantic transmission index feedback adjustment provided by the embodiments of the present application, as shown in Figure 1 The embodiments of the present application provide a semantic communication system based on unified semantic transmission index feedback adjustment, which comprises:

[0047] a semantic encoder 101, a semantic transmitter 102, a semantic receiver 103 and a semantic decoder 104;

[0048] The semantic encoder is configured to determine a feature value according to source data, and transmit binary data to the semantic transmitter based on the feature value;

[0049] The semantic transmitter is configured to transmit a semantic symbol to the semantic receiver based on the binary data, the semantic symbol being determined based on a wireless semantic transmission performance index, the wireless semantic transmission performance index being used to represent a semantic difference between the feature value extracted from the source data and a feature value received by the semantic receiver;

[0050] The semantic receiver is configured to convert the received semantic symbol into the binary data, and transmit the binary data to the semantic decoder;

[0051] The semantic decoder is configured to reconstruct the source data based on the binary data.

[0052] Specifically, since the unified index of the conventional communication system, such as the bit error rate (BER), its calculation almost does not involve the semantic related coding and decoding process, and cannot be directly reused in the semantic communication field. In addition, in real application, there are cases where the communication enterprise and the service provider are different companies, and it is difficult for cross-industry companies to jointly train or process data in reality. The physical separation between the source encoder / decoder and the transceiver is of great significance to the service provider and the communication enterprise.

[0053] In view of the above technical problems, the embodiment of the present application considers the need to make different system designs according to actual conditions, and proposes a semantic communication system in which the semantic extraction module and the coding and decoding module are separated. In addition, it is necessary to develop a semantic communication system with a unified semantic communication performance index adjustment to realize the transmission quality feedback of semantic communication and generate an adjustment strategy for semantic communication.

[0054] The semantic communication system based on the unified semantic transmission index feedback adjustment provided by the embodiment of the present application includes a semantic encoder, a semantic transmitter, a semantic receiver and a semantic decoder. The semantic encoder and the semantic decoder can be the same or different service providers, and the semantic transmitter and the semantic receiver can be the same or different communication enterprises.

[0055] The semantic encoder can extract a feature value from source data, and then convert the feature value into binary data and send it to the semantic transmitter. The binary data can be sent through a traditional wired transmission method.

[0056] The semantic transmitter converts a binary sequence, i.e., binary data, into feature values in an error-free manner, and maps the continuous feature values to discrete constellation symbols, so that the modulated semantic symbols are more suitable for wireless transmission.

[0057] The semantic receiver maps the received semantic symbols into feature values, converts the feature values into binary data, and sends the binary data to a semantic decoder.

[0058] The semantic decoder reconstructs the source data according to the binary data.

[0059] The semantic communication system based on the unified semantic transmission index feedback adjustment provided in the embodiments of the present application proposes a unified wireless semantic transmission performance index, which can fill the blank of the performance index that has not been provided for measuring the semantic transmission performance regardless of data modes or tasks. The index can be used for a communication enterprise to evaluate the transmission performance of semantic information without relying on the semantic extraction result of a data source in any mode or task, and has higher applicability.

[0060] In some embodiments, the semantic encoder comprises a semantic extraction module and a binary quantization module.

[0061] The semantic extraction module is configured to extract the feature values from the source data.

[0062] The binary quantization module is configured to convert the feature values into binary data.

[0063] Specifically, the semantic encoder comprises a semantic extraction module and a binary quantization module.

[0064] The semantic extraction module usually relies on a deep neural network (DNN), and the output of the deep neural network is the feature values extracted from the source content.

[0065] The binary quantization module converts the feature values into a binary sequence.

[0066] The semantic communication system based on the unified semantic transmission index feedback adjustment provided in the embodiments of the present application proposes a unified wireless semantic transmission performance index, which can fill the blank of the performance index that has not been provided for measuring the semantic transmission performance regardless of data modes or tasks. The index can be used for a communication enterprise to evaluate the transmission performance of semantic information without relying on the semantic extraction result of a data source in any mode or task, and has higher applicability.

[0067] In some embodiments, the semantic transmitter comprises an error-free feature recovery module and a semantic-oriented modulation module.

[0068] The error-free feature recovery module is configured to convert the binary data into the feature values.

[0069] The semantic-oriented modulation module is configured to generate the semantic symbol according to a hybrid automatic repeat request (HARQ) feedback message and the feature value, wherein the HARQ feedback message is determined to be sent based on the wireless semantic transmission performance index.

[0070] Specifically, the semantic transmitter comprises an error-free feature recovery module and a semantic-oriented modulation module.

[0071] The error-free feature recovery module converts the binary sequence into the feature value in an error-free manner.

[0072] The semantic-oriented modulation module needs to map the continuous feature value to the discrete constellation symbol, so that the modulated semantic symbol is more suitable for wireless transmission.

[0073] Further, the semantic transmitter aims to accurately transmit the semantic feature to the receiver through the wireless channel, and thus, a wireless semantic transmission performance index (Semantic Error Ratio, SeER) can be used as an index for evaluating the semantic transmission performance, i.e., measuring the difference between the feature vector F and F'.

[0074] The semantic communication system based on the unified semantic transmission index feedback adjustment provided by the embodiments of the present application proposes a unified wireless semantic transmission performance index, which can fill the gap of the performance index that has not been provided for measuring the semantic transmission performance regardless of the data mode or task. The index can be used by a communication enterprise to evaluate the transmission performance of semantic information, without relying on the semantic extraction result of the data source in any mode or task, and has higher applicability.

[0075] In some embodiments, the semantic receiver comprises a semantic-oriented demodulation module and a binary quantization module.

[0076] The semantic-oriented demodulation module is configured to demodulate the semantic symbol into the feature value.

[0077] The binary quantization module is configured to convert the feature value into a binary sequence.

[0078] Specifically, the semantic receiver comprises a semantic-oriented demodulation module and a binary quantization module.

[0079] The semantic-oriented demodulation module maps the received semantic symbol to the feature.

[0080] The binary quantization module maps the feature to the binary sequence.

[0081] The semantic communication system based on unified semantic transmission index feedback adjustment provided in the embodiments of the present application proposes a unified wireless semantic transmission performance index, which can fill the blank of the performance index that has not been provided to measure the semantic transmission performance regardless of data mode or task. The index can be used for a communication enterprise to evaluate the transmission performance of semantic information without relying on the semantic extraction result of a data source in any mode or task, and has higher applicability.

[0082] In some embodiments, the semantic decoder comprises an error-free feature recovery module and a semantic recovery module.

[0083] The error-free feature recovery module is configured to convert the binary data into the feature value.

[0084] The semantic recovery module is configured to reconstruct the source data according to the feature value.

[0085] Specifically, the semantic decoder comprises an error-free feature recovery module and a semantic recovery module.

[0086] The error-free feature recovery module converts the binary sequence into the feature value in an error-free manner.

[0087] The semantic recovery module reconstructs the source data according to the feature value.

[0088] The semantic communication system based on unified semantic transmission index feedback adjustment provided in the embodiments of the present application proposes a unified wireless semantic transmission performance index, which can fill the blank of the performance index that has not been provided to measure the semantic transmission performance regardless of data mode or task. The index can be used for a communication enterprise to evaluate the transmission performance of semantic information without relying on the semantic extraction result of a data source in any mode or task, and has higher applicability.

[0089] Figure 2 is one of the flowcharts of the semantic communication method based on unified semantic transmission index feedback adjustment provided in the embodiments of the present application, as shown in Figure 2 The present application provides a semantic communication method based on unified semantic transmission index feedback adjustment, applied to a semantic receiver, comprising:

[0090] Step 201, receiving the semantic symbol sent by a semantic transmitter;

[0091] Step 202, demodulating the semantic symbol to obtain a received feature value;

[0092] Step 203, determining a wireless semantic transmission performance index based on the received feature value and a feature value extracted from source data;

[0093] Step 204, sending a feedback message to the semantic transmitter based on the wireless semantic transmission performance index.

[0094] Specifically, the unified semantic transmission index SeER proposed in the embodiments of the present application can adjust the modulation mode of the transmission of semantic communication in order to realize the transmission quality feedback of semantic communication.

[0095] The SeER infers the semantic difference from the original semantics through transmission, and the difference can be measured by the probability that the difference between the feature value extracted from the source and the feature value received after transmission exceeds the predefined semantic difference threshold for different features.

[0096] The semantic receiver can obtain the SeER according to the difference between the feature value of the semantic symbol sent by the semantic transmitter received and the feature value of the source data stored by itself.

[0097] When the SeER is greater than the semantic difference threshold p, the feedback mechanism will be started to retransmit the data, and at the same time the modulation mode adjustment strategy is performed at the semantic transmitter.

[0098] Further, the feedback mechanism can be realized by sending the HARQ feedback from the semantic receiver to the semantic transmitter.

[0099] The semantic communication method based on the feedback adjustment of the unified semantic transmission index provided by the embodiments of the present application proposes a unified wireless semantic transmission performance index, which can fill the gap of the performance index that has not been provided to measure the semantic transmission performance regardless of the data mode or task. According to the index SeER, the modulation mode is adjusted by using the combination of high layer and low order modulation at the sending end through the HARQ feedback, the data is retransmitted, and more accurate feature description, higher transmission efficiency and higher transmission reliability can be realized.

[0100] In some embodiments, the calculation formula for determining the wireless semantic transmission performance index based on the received feature value and the feature value extracted from the source data is:

[0101]

[0102] wherein SeER represents the wireless semantic transmission performance index, N represents the total number of semantic features, X i and Y i are the expected vector and the received vector of the i-th group respectively, (·) represents the dot product of the vector, |X i | and |Y i | represent the length of the vector X i and Y i respectively, and p represents the semantic difference threshold of the vector transmission.

[0103] ​Specifically, SeER infers semantic difference from transmission with the original semantics, which can be measured by the probability that the difference between the feature values extracted from the source and the received feature values after transmission exceeds the predefined semantic difference threshold for different features. There are many ways to realize the difference between features, not limited to the definition of SeER by cosine similarity mentioned in this proposal.

[0104] For the above formula, the predefined semantic difference threshold p depends on the importance of the feature, that is, the higher the importance, the lower the semantic difference threshold, and vice versa. The importance or weight of the feature usually comes from the semantic extraction model together with the semantic feature. Obviously, the larger p, the more stringent the requirement for semantic transmission. Because the cosine similarity of X and Y is negatively correlated with the Euclidean distance of the two, when SeER is greater than the threshold p, it is considered that the vector transmission is wrong.

[0105] The semantic communication method based on unified semantic transmission index feedback adjustment provided by the embodiments of the present application proposes a unified wireless semantic transmission performance index, which can fill the gap of not providing performance indexes that do not distinguish between data modes or tasks and only measure semantic transmission performance. According to the index SeER, feedback is provided through HARQ, and the modulation mode is adjusted at the sending end by combining high-level and low-level modulation, and the data is retransmitted, which can achieve more accurate feature description, higher transmission efficiency and higher transmission reliability.

[0106] In some embodiments, the sending of the feedback message to the semantic transmitter based on the wireless semantic transmission performance index comprises:

[0107] In the case where the wireless semantic transmission performance index is greater than the semantic difference threshold, a feedback message is sent to the semantic transmitter.

[0108] Specifically, in semantic communication, the importance of semantic features is not the same, which means that the requirements of multiple features in the vector for transmission accuracy are different. The more important the feature, the higher the requirement for transmission accuracy, and vice versa. High-precision feature data transmission often requires higher data transmission rate or larger information capacity. High-order modulation can meet this requirement, but high-order modulation signals are more sensitive to noise and interference, so they can only be effectively used in good channel quality, otherwise the error rate will increase rapidly, resulting in a decline in communication performance. In theory, the number of points of l-layer 2 n order constellation is l·2 n However, the number of points of classical 2 m order modulation is 2 m . m≤n for l≥1 means that at the same transmission efficiency, high-level is beneficial to improve transmission reliability (i.e. l·2 n = 2 m ). To achieve the feature accuracy described by l-layer 2 n order constellation, it is necessary to use (2m ) l Classical modulation, which will result in the loss of transmission reliability.

[0109] To solve the above technical problems, the embodiment of the present application proposes that when SeER is greater than the semantic difference threshold ρ, the HARQ mechanism is started to retransmit data, and the modulation mode adjustment strategy of the semantic transmitter is performed.

[0110] In modulation, in order to limit the range of constellation, normalization processing is added in the previous semantic extraction to provide a normalized feature vector. Each two normalized feature values in the vector are called a symbol point in the constellation. The i-th layer constellation represents the smallest unit in the (i-1)-th layer, but its length is the same as that of the (i-1)-th layer, that is, the same length presents a smaller actual range.

[0111] For example, Figure 3 is a two-layer structure diagram of a 16-order mapping scheme provided by the embodiment of the present application, as shown in Figure 3 The first layer constellation plane is divided into 16 coarse intervals, and the second layer constellation plane further divides the smallest unit interval to which the feature value belongs into 16 finer intervals. In hierarchical modulation, the accuracy requirement of the feature value described by 32 constellation points adopts 256QAM. In this case, the transmission reliability of hierarchical modulation is much higher than that of the traditional 256QAM.

[0112] The semantic communication method based on unified semantic transmission index feedback adjustment provided by the embodiment of the present application proposes a unified wireless semantic transmission performance index, which can fill the blank of the performance index that has not yet provided the performance index of measuring semantic transmission performance regardless of data mode or task. According to the index SeER, the modulation mode is adjusted by using the combination of high-layer and low-order modulation at the sending end through HARQ feedback, and the data is retransmitted, which can realize more accurate feature description, higher transmission efficiency and higher transmission reliability.

[0113] Figure 4 is a flowchart of the semantic communication method based on unified semantic transmission index feedback adjustment provided by the embodiment of the present application, as shown in Figure 4 The embodiment of the present application provides a semantic communication method based on unified semantic transmission index feedback adjustment, applied to a semantic transmitter, comprising:

[0114] Step 401, receiving a feedback message sent by a semantic receiver;

[0115] Step 402, generating a semantic symbol based on the feedback message modulation;

[0116] Step 403, sending the semantic symbol to the semantic receiver.

[0117] In some embodiments, the generating semantic symbols based on the feedback message comprises:

[0118] extracting a normalized feature vector from the source data based on the normalization processing, each two normalized feature values in the normalized feature vector being a symbol point of the semantic symbol;

[0119] generating the semantic symbol based on the normalized feature vector.

[0120] Specifically, the above-mentioned semantic communication method based on the unified semantic transmission index feedback adjustment provided by the embodiments of the present application can realize all the method steps implemented by the semantic communication method embodiment based on the unified semantic transmission index feedback adjustment of the execution subject as the semantic receiver, and can achieve the same technical effects. Here, the same parts and beneficial effects in the method embodiment will not be described in detail.

[0121] From the above description of the embodiments, those skilled in the art can clearly understand that the embodiments can be realized by means of software plus the necessary general hardware platforms, and of course, can also be realized by hardware. Based on such understanding, the above technical solutions can be embodied in the form of a software product, which can be stored in a computer readable storage medium, such as a ROM / RAM, a magnetic disk, an optical disk, etc., and includes a number of instructions to make a computer device (which can be a personal computer, a server, or a network device, etc.) execute the methods described in each embodiment or some parts of the embodiments.

[0122] In addition, it should be noted that the terms "first", "second" and the like in the embodiments of the present application are used to distinguish similar objects, and are not used to describe a specific order or sequence. It should be understood that the terms used in this way can be interchanged under appropriate circumstances, so that the embodiments of the present application can be implemented in an order other than that illustrated or described herein, and the objects distinguished by "first", "second" are usually a class, not limited to the number of objects, for example, the first object can be one or more.

[0123] In the embodiments of the present application, the term "and / or" describes the association relationship between the associated objects, which means that there can be three relationships, for example, A and / or B can mean that there are three cases of A alone, A and B together, and B alone. The character " / " generally represents an "or" relationship between the associated objects before and after it.

[0124] In the embodiments of the present application, the term "a plurality of" means two or more, and other quantifiers are similar.

[0125] In the present application, "determining B based on A" means that A is considered as a factor when determining B. It is not limited to "determining B based on A only", but also includes "determining B based on A and C", "determining B based on A, C and E", "determining C based on A, and determining B based on C further", and the like. In addition, it can also include A as a condition for determining B, for example, "when A meets the first condition, determining B using the first method"; for example, "when A meets the second condition, determining B"; for example, "when A meets the third condition, determining B based on the first parameter"; and the like. Of course, A can also be a condition for determining B as a factor, for example, "when A meets the first condition, determining C using the first method, and determining B further based on C"; and the like.

[0126] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present application, and are not limited thereto. Although the present application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that the technical solutions recorded in the foregoing embodiments can still be modified, or some technical features can be replaced by equivalents. The modification or replacement does not make the essence of the corresponding technical solution deviate from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. A semantic communication system based on unified semantic transmission metric feedback regulation, characterized by, The application relates to a semantic encoder, a semantic transmitter, a semantic receiver and a semantic decoder. The semantic encoder is used for determining characteristic values according to source data and sending binary data to the semantic transmitter based on the characteristic values. The semantic transmitter is used for sending semantic symbols to the semantic receiver based on the binary data, wherein the semantic symbols are determined based on a wireless semantic transmission performance index, and the wireless semantic transmission performance index is used for representing a semantic difference between the characteristic values extracted from the source data and the characteristic values received by the semantic receiver. The semantic receiver is used for converting the received semantic symbols into the binary data and sending the binary data to the semantic decoder. The semantic decoder is used for reconstructing the source data based on the binary data. The wireless semantic transmission performance is determined based on the characteristic values received by the semantic receiver and the characteristic values extracted from the source data, and a calculation formula of the wireless semantic transmission performance index is as follows: The semantic receiver is further used for sending a feedback message to the semantic transmitter when the wireless semantic transmission performance index is greater than a semantic difference threshold value, wherein the semantic difference threshold value is determined based on the importance of the characteristic. ； wherein, denotes a wireless semantic transmission performance indicator, N denotes a total number of semantic features, and denote a vector of expected and received, respectively, a first denotes a dot product of vectors, and denote a length of vectors and respectively, denotes a semantic difference threshold for vector transmission; The semantic encoder comprises a semantic extraction module and a binary quantization module.

2. The unified semantics-based transmission metrics feedback adjusted semantic communication system of claim 1, wherein, The semantic extraction module is used for extracting the characteristic values from the source data. The binary quantization module is used for converting the characteristic values into binary data. The semantic transmitter comprises an error-free characteristic recovery module and a semantic-oriented modulation module.

3. The unified semantics-based transmission metrics feedback adjusted semantic communication system of claim 1, wherein, The error-free characteristic recovery module is used for converting the binary data into the characteristic values. The semantic-oriented modulation module is used for generating the semantic symbols according to the feedback message and the characteristic values. The semantic receiver comprises a semantic-oriented demodulation module and a binary quantization module.

4. The unified semantics-based transmission metrics feedback adjusted semantic communication system of claim 1, wherein, The semantic-oriented demodulation module is used for demodulating the semantic symbols into the characteristic values. The binary quantization module is used for converting the characteristic values into binary data. The semantic decoder comprises an error-free characteristic recovery module and a semantic recovery module.

5. The unified semantics-based transmission metrics feedback adjusted semantic communication system of claim 1, wherein, The error-free characteristic recovery module is used for converting the binary data into the characteristic values. The semantic recovery module is used for reconstructing the source data according to the characteristic values. The application is applied to a semantic receiver and comprises the following steps:

6. A semantic communication method based on uniform semantic transmission index feedback adjustment, characterized in that, receiving semantic symbols sent by a semantic transmitter; demodulating the semantic symbols to obtain received characteristic values; determining a wireless semantic transmission performance index based on the received characteristic values and characteristic values extracted from source data; sending a feedback message to the semantic transmitter based on the wireless semantic transmission performance index; a calculation formula for determining the wireless semantic transmission performance index based on the received characteristic values and the characteristic values extracted from the source data is as follows: the step of sending the feedback message to the semantic transmitter based on the wireless semantic transmission performance index comprises the following steps: ； wherein, denotes a wireless semantic transmission performance indicator, N denotes the total number of semantic features, and are the expected and received vectors, respectively, of the first denotes the dot product of vectors, and denote the length of the vectors and , respectively, denotes a semantic difference threshold for vector transmission; sending the feedback message to the semantic transmitter when the wireless semantic transmission performance index is greater than a semantic difference threshold value, wherein the semantic difference threshold value is determined based on the importance of the characteristic. ​ 7. A semantic communication method based on uniform semantic transmission index feedback adjustment, characterized by, The application is applied to a semantic transmitter, comprising: receiving a feedback message sent by a semantic receiver; generating a semantic symbol based on the feedback message; sending the semantic symbol to the semantic receiver; the feedback message is sent by the semantic receiver to the semantic transmitter when a wireless semantic transmission performance index is greater than a semantic difference threshold value, the semantic difference threshold value is determined based on the importance of features, the wireless semantic transmission performance is determined based on feature values received by the semantic receiver and feature values extracted from source data, and the calculation formula of the wireless semantic transmission performance index is: ； wherein, denotes a wireless semantic transmission performance indicator, N denotes a total number of semantic features, and are the expected and received vectors, respectively, of the first denotes a dot product of vectors, and denote the lengths of the vectors and , respectively, denotes a semantic difference threshold for vector transmission.

8. The unified semantics-based transmission metrics feedback adjusted semantic communication method of claim 7, wherein, the generation of the semantic symbol based on the feedback message comprises: extracting a normalized feature vector from source data based on normalization processing, and each two normalized feature values in the normalized feature vector are a symbol point of the semantic symbol; generating the semantic symbol based on the normalized feature vector.

Images