Semantic communication method and device, electronic equipment and storage medium

Abstract

The invention provides a semantic communication method and device, electronic equipment and a storage medium, and relates to the technical field of computers, in particular to the technical field of semantic communication. The method is applied to a receiving end, and the specific implementation scheme is as follows: target information source data sent by a sending end is received, and the target information source data is obtained by extracting multiple pieces of attribute information from original information source data and then encoding the multiple pieces of attribute information; decoding the target information source data to obtain decoded attribute information; and reconstructing the original information source data through the generative receiver by using the decoded attribute information to obtain reconstructed information source data. According to the invention, only multiple pieces of attribute information of the original information source data are transmitted and processed, so that errors generated in encoding and decoding processes are reduced, efficient compression and reconstruction of data transmission are realized, bandwidth occupation is reduced, and communication reliability and efficiency are improved.

Description

Technical Field

[0001] The present disclosure relates to the field of computer technology, specifically to the field of semantic communication technology, and more particularly to semantic communication methods, devices, electronic devices, and storage media. Background Art

[0002] In traditional communication systems, data transmission typically involves converting original information into a format suitable for transmission (encoding), then sending it through a communication channel to the receiver, which then restores the received signal to the original information (decoding). However, this process is not perfect. During the encoding and decoding process, data may be subject to errors due to various reasons (such as channel noise and interference). These errors can reduce the accuracy of information transmission.

[0003] Semantic communication, a new communication paradigm, aims to improve communication reliability by transmitting semantic information rather than raw bit streams. This approach theoretically enables more robust data transmission under low signal-to-noise ratio conditions. However, semantic communication faces similar challenges in practical applications. While it attempts to reduce errors by extracting and transmitting semantic features of data, errors can still occur during feature extraction, encoding, and decoding, compromising accurate information transmission. Summary of the Invention

[0004] The present disclosure provides a semantic communication method, device, electronic device and storage medium.

[0005] According to one aspect of the present disclosure, a semantic communication method is provided, which is applied to a receiving end, and the method includes:

[0006] Receiving target source data sent from a transmitting end, wherein the target source data is obtained by extracting multiple attribute information from the original source data and encoding the multiple attribute information;

[0007] Decoding the target source data to obtain decoded attribute information;

[0008] The decoded attribute information is used to reconstruct the original source data through a generative receiver to obtain reconstructed source data.

[0009] According to another aspect of the present disclosure, a semantic communication method is provided, which is applied to a sending end, and the method includes:

[0010] Obtain original source data;

[0011] Extracting multiple attribute information from the original source data and encoding the multiple attribute information to obtain target source data;

[0012] The target source data is sent to a receiving end, the target source data is decoded by the receiving end to obtain decoded attribute information, and the decoded attribute information is used to reconstruct the original source data through a generative receiver to obtain reconstructed source data.

[0013] According to a third aspect of the present disclosure, a semantic communication method is provided, the method comprising:

[0014] The sending end obtains original source data, extracts multiple attribute information from the original source data, encodes the multiple attribute information to obtain target source data, and sends the target source data to the receiving end;

[0015] The receiving end decodes the target source data to obtain decoded attribute information, and uses the decoded attribute information to reconstruct the original source data through a generative receiver to obtain reconstructed source data.

[0016] According to a fourth aspect of the present disclosure, a semantic communication device is provided, applied to a receiving end, comprising:

[0017] A receiving module is configured to receive target source data sent from a transmitting end, wherein the target source data is obtained by extracting multiple attribute information from the original source data and encoding the multiple attribute information;

[0018] A decoding module, configured to decode the target source data to obtain decoded attribute information;

[0019] The reconstruction module is used to reconstruct the original source data by using the decoded attribute information through a generative receiver to obtain reconstructed source data.

[0020] According to a fifth aspect of the present disclosure, a semantic communication device is provided, applied to a sending end, comprising:

[0021] An acquisition module is used to obtain original source data;

[0022] An encoding module, configured to extract multiple attribute information from the original source data and encode the multiple attribute information to obtain target source data;

[0023] The sending module is used to send the target source data to the receiving end, decode the target source data through the receiving end to obtain decoded attribute information, and use the decoded attribute information to reconstruct the original source data through the generative receiver to obtain reconstructed source data.

[0024] According to a sixth aspect of the present disclosure, there is provided an electronic device, including:

[0025] at least one processor; and

[0026] a memory communicatively connected to the at least one processor; wherein,

[0027] The memory stores instructions that can be executed by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to execute any one of the methods described in the above technical solutions.

[0028] According to a seventh aspect of the present disclosure, a non-transitory computer-readable storage medium storing computer instructions is provided, wherein the computer instructions are used to enable the computer to execute any one of the methods described above.

[0029] According to an eighth aspect of the present disclosure, a computer program product is provided, comprising a computer program, wherein the computer program implements any one of the methods described above when executed by a processor.

[0030] The present disclosure provides a semantic communication method, apparatus, device, and storage medium. This method receives target source data transmitted by a transmitter. This data is obtained by extracting multiple attribute information from the original source data and encoding the multiple attribute information. The receiver decodes the received data to recover the original attribute information. Subsequently, using this decoded attribute information, a generative receiver can reconstruct reconstructed source data that is highly similar to the original source data. This method reduces the amount of data transmitted by transmitting only the multiple attribute information of the data, which directly reduces the possibility of introducing errors during the transmission process. Furthermore, since this method only transmits multiple attribute information rather than the entire original data, the amount of data required for transmission is significantly reduced, thereby improving communication efficiency. Furthermore, the generative receiver uses the decoded attribute information as a guide to gradually construct new data similar to the original data through a denoising or generative process. This process does not simply copy the original data, but regenerates the data based on the attribute information, thereby reducing reliance on the original data and reducing decoding errors. This enables more accurate and robust data transmission even under low signal-to-noise ratio conditions, enhancing communication reliability. It can be seen that this method reduces the errors generated in the encoding and decoding processes by only transmitting and processing multiple attribute information of the original source data, and realizes efficient compression and reconstruction of data transmission, thereby reducing bandwidth occupancy and improving communication reliability and efficiency.

[0031] It should be understood that the contents described in this section are not intended to identify the key or important features of the embodiments of the present disclosure, nor are they intended to limit the scope of the present disclosure. Other features of the present disclosure will become readily understood through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0032] The accompanying drawings are provided to facilitate a better understanding of the present invention and do not constitute a limitation of the present disclosure.

[0033] Figure 1 is a schematic diagram of the steps of a semantic communication method in one embodiment of the present disclosure;

[0034] Figure 2 is a schematic diagram of steps of a semantic communication method in another embodiment of the present disclosure;

[0035] Figure 3 It is a schematic diagram of the overall process of the semantic communication method in one embodiment of the present disclosure;

[0036] Figure 4 A block diagram of the principle of a semantic communication device in an embodiment of the present disclosure;

[0037] Figure 5 A functional block diagram of a semantic communication device in another embodiment of the present disclosure;

[0038] Figure 6 It is a block diagram of an electronic device used to implement the semantic communication method according to an embodiment of the present disclosure. DETAILED DESCRIPTION

[0039] The following description of exemplary embodiments of the present disclosure is made in conjunction with the accompanying drawings, including various details of the embodiments of the present disclosure to facilitate understanding. These details should be considered as merely exemplary. Therefore, those skilled in the art will recognize that various changes and modifications may be made to the embodiments described herein without departing from the scope and spirit of the present disclosure. Similarly, for the sake of clarity and conciseness, descriptions of well-known functions and structures are omitted in the following description.

[0040] The present disclosure provides a semantic communication method, see Figure 1 As shown, Figure 1 1 is a schematic diagram of the steps of a semantic communication method in an embodiment of the present disclosure, which is applied to a receiving end and includes:

[0041] Step S101: receiving target source data sent from a transmitting end. The target source data is obtained by extracting multiple attribute information from the original source data and encoding the multiple attribute information.

[0042] Specifically, "target source data" refers to source data that has undergone specific processing, resulting from a series of operations performed on the original source data by the sender. "Original source data" is the initial information to be transmitted, such as an image, an audio clip, or any other type of data. "Multiple attribute information" refers to the different features extracted from the original source data. These features represent the core content of the original data, such as the color, texture, and shape of an image, or the pitch and rhythm of audio.

[0043] The specific implementation process of this solution involves: First, the transmitter uses a technology called a "semantic feature decoupler" to extract these attribute information from the original source data. Next, this attribute information is fed into a "source-channel joint encoder," where it is converted into a format suitable for transmission over the communication channel. This process typically involves data compression to reduce the bandwidth required for transmission. The encoded data, known as the "target source data," contains the key information of the original source data and requires less bandwidth resources for transmission, making it more suitable for transmission.

[0044] In this way, the transmitter can convert the original source data into a more efficient and easier-to-transmit form while preserving the critical information needed to reconstruct the original data. This allows for efficient data transmission even under limited bandwidth or poor channel conditions, and the receiver, through appropriate decoding and reconstruction processes, can recover a close approximation of the original source data. This approach not only improves communication efficiency but also reduces potential errors and data loss during transmission, thereby enhancing communication reliability.

[0045] Step S102: Decode the target source data to obtain decoded attribute information.

[0046] Specifically, "target source data" refers to data that has been encoded and prepared for transmission over a communication channel. It contains attribute information extracted and encoded from the original source data. "Decoding" is the reverse process of encoding, which involves restoring the target source data to its original, more easily processed and understandable form. "Attribute information" refers to the key features of the original source data, such as the texture and color of an image or keywords in a text. These features represent the core content of the original data.

[0047] The specific implementation process of this solution involves the following: Once the target source data is transmitted to the receiver via a communication channel, the receiver's "source-channel joint decoder" begins operation. The decoder uses its internal algorithms and models to decode the encoded data, a process that may include error detection and correction to ensure data integrity and accuracy. The decoder restores the encoded data into attribute information. While this information is not a complete copy of the original source data, it is sufficient to reconstruct the original data or perform further processing.

[0048] Through the decoding process, the receiver can extract useful attribute information from the target source data. This information is the basis for reconstructing the original source data or performing semantic analysis. The decoded attribute information provides the necessary input for the generative receiver, enabling it to reconstruct content that is highly similar to the original source data, thereby achieving high-quality information transmission.

[0049] Step S103 : using the decoded attribute information, reconstructing the original source data through a generative receiver to obtain reconstructed source data.

[0050] Specifically, "decoded attribute information" refers to the key feature data extracted and decoded from the target source data. This information is the essence of the original source data and contains all the necessary information needed to reconstruct the original data, such as the texture and color of an image or the pitch and rhythm of audio. The "generative receiver" is an advanced data processing module that uses artificial intelligence technology, specifically generative models, to reconstruct the original source data based on this attribute information.

[0051] The specific implementation process of this solution involves the following: a generative receiver first receives decoded attribute information, which serves as the basic input for the reconstruction process. Then, a generative model in the generative receiver, such as a diffusion model or a generative adversarial network (GAN), uses its learning and creativity to gradually construct new data similar to the original source data based on the attribute information. This process involves complex algorithms, and the model continuously adjusts and optimizes the generated data until it matches the characteristics of the original data, thereby generating the reconstructed source data.

[0052] The reconstructed source data is an approximation of the original source data; it may not be identical, but it retains the core semantics and characteristics of the original data. The advantage of this approach is that it significantly reduces the bandwidth required during data transmission while simultaneously restoring high-quality data at the receiving end through intelligent generation techniques, enabling efficient data transmission and storage while maintaining data integrity and availability. This generative reconstruction method based on attribute information not only improves the efficiency of the communication system but also enhances its robustness to transmission errors, enabling reliable data communication even under suboptimal channel conditions.

[0053] The present disclosure provides a semantic communication method, apparatus, device, and storage medium. This method receives target source data transmitted by a transmitter. This data is obtained by extracting multiple attribute information from the original source data and encoding the multiple attribute information. The receiver decodes the received data to recover the original attribute information. Subsequently, using this decoded attribute information, a generative receiver can reconstruct reconstructed source data that is highly similar to the original source data. This method reduces the amount of data transmitted by transmitting only the multiple attribute information of the data, which directly reduces the possibility of introducing errors during the transmission process. Furthermore, since this method only transmits multiple attribute information rather than the entire original data, the amount of data required for transmission is significantly reduced, thereby improving communication efficiency. Furthermore, the generative receiver uses the decoded attribute information as a guide to gradually construct new data similar to the original data through a denoising or generative process. This process does not simply copy the original data, but regenerates the data based on the attribute information, thereby reducing reliance on the original data and reducing decoding errors. This enables more accurate and robust data transmission even under low signal-to-noise ratio conditions, enhancing communication reliability. It can be seen that this method reduces the errors generated in the encoding and decoding processes by only transmitting and processing multiple attribute information of the original source data, and realizes efficient compression and reconstruction of data transmission, thereby reducing bandwidth occupancy and improving communication reliability and efficiency.

[0054] In some optional embodiments, decoding the target source data to obtain decoded attribute information includes:

[0055] The source-channel joint decoder is used to decode the target source data to obtain decoded attribute information.

[0056] Specifically, the "source-channel joint decoder" is a specially designed decoding device that decodes received target source data to recover its original attribute information. "Target source data" is encoded source data. This data is generated by extracting multiple attribute information from the original source data at the transmitter and then encoded and transmitted to the receiver via a communication channel. "Attribute information" refers to the features contained in the original source data, such as the texture and color of an image or the pitch and rhythm of audio. These features serve as the basis for reconstructing the original source data.

[0057] The specific implementation process of this solution includes the following: When the target source data arrives at the receiver, the source-channel joint decoder begins operation. It first identifies the coding pattern in the data and then applies the corresponding decoding algorithm to restore the encoded data to the original attribute information. This process may include error detection and correction steps to ensure the accuracy and integrity of the decoded information. The decoder is designed to adapt to different channel conditions and coding strategies, thereby effectively recovering attribute information in various environments. In this way, the source-channel joint decoder not only extracts useful attribute information from the target source data, but also ensures the integrity and accuracy of this information during transmission, providing a reliable foundation for subsequent source data reconstruction. This process is a key step in realizing an efficient and reliable semantic communication system. It enables the system to transmit more useful information within limited bandwidth resources while reducing errors that may occur during transmission, thereby improving overall communication performance.

[0058] In this way, the process of decoding the target source data using a joint source-channel decoder not only ensures the accurate recovery of the original attribute information from the encoded target source data, but also, through a specially designed decoder, can adapt to various channel conditions, thereby maintaining efficient data recovery capabilities in different communication environments. This decoding method helps reduce data loss and deformation during transmission, improving the reliability of data transmission. At the same time, because the decoder can accurately restore the attribute information, it provides a high-quality foundation for subsequent source data reconstruction, allowing the reconstructed data to more accurately reflect the characteristics of the original data, enhancing the performance of the entire communication system. In addition, accurate attribute information decoding also helps reduce the system's bandwidth requirements, because only key attribute information needs to be transmitted and processed, rather than the entire original data. This is particularly beneficial in bandwidth-constrained situations.

[0059] In some optional embodiments, the generative receiver includes: a diffusion model and a conditional control network;

[0060] The decoded attribute information is used to reconstruct the original source data through a generative receiver to obtain the reconstructed source data, including:

[0061] The decoded attribute information is input into a conditional control network, and the conditional control network generates a control signal corresponding to the decoded attribute information according to the decoded attribute information;

[0062] The decoded attribute information and the control signal are input into the diffusion model, and the control signal is used to guide the diffusion model to reconstruct the original source data according to the decoded attribute information to obtain the reconstructed source data.

[0063] Specifically, the "conditional control network" is an intelligent processing system that generates corresponding control signals based on the decoded attribute information. These signals are used to guide the subsequent data reconstruction process. The "diffusion model" is a generative model that can reconstruct the original source data using the input attribute information and control signals. "Decoded attribute information" refers to the key feature data extracted and decoded from the target source data, which contains the basic information required to reconstruct the original data.

[0064] The implementation process of this solution involves the following: First, the decoded attribute information is input into a conditional control network. This network analyzes the attribute information and generates a set of control signals. These signals correspond to the attribute information and contain instructions for reconstructing the original data based on this attribute information. Next, these control signals, along with the decoded attribute information, are input into a diffusion model. The diffusion model is a complex algorithm that understands the meaning of the control signals and gradually reconstructs the original source data based on these signals and the attribute information.

[0065] Control signals play a crucial role in the reconstruction process, guiding the diffusion model on how to utilize attribute information to reconstruct data, ensuring that the reconstructed data accurately reflects the characteristics of the original data. This allows the diffusion model to generate reconstructed source data that is highly similar to the original source data. Even if there may be some losses or errors during data transmission, these can be compensated for through the intelligent reconstruction process. This approach not only improves the efficiency of data transmission but also enhances its reliability, as it allows the system to reconstruct high-quality data with limited information, thus enabling effective communication even in bandwidth-constrained or poor channel conditions.

[0066] In this way, by inputting the decoded attribute information into a conditional control network and using this network to generate control signals corresponding to the attribute information, precise guidance of the diffusion model is achieved, a process that significantly improves the quality and efficiency of data reconstruction. The conditional control network ensures a high correlation between the control signal and the decoded attribute information, allowing the diffusion model to more accurately understand and utilize this information to reconstruct the original source data. This approach not only improves the accuracy of the reconstructed data and reduces decoding errors caused by information loss or transmission errors, but also optimizes the data transmission process because it allows the system to transmit only key attribute information rather than the entire original data, thereby reducing the required bandwidth resources. In addition, by intelligently reconstructing the data, the scheme improves the reliability and robustness of the communication system while maintaining data integrity and original semantics, especially under less than ideal channel conditions.

[0067] The present disclosure provides a semantic communication method, see Figure 2 , Figure 2 : is a schematic diagram of the steps of a semantic communication method in another embodiment of the present disclosure, which is applied to a sending end and includes:

[0068] Step S201: Acquire original source data.

[0069] Specifically, "acquiring raw source data" is an initial step in a communication system. "Raw source data" refers to the original information before any processing or transmission. This can be text files, images, audio, video, or any other form of data. This feature describes the process involved in collecting this raw information from its source for subsequent processing and transmission.

[0070] The specific implementation process of this solution involves: first, identifying the data source, which could be a sensor, a file system, a database, or any other device capable of generating or storing data. Next, using the appropriate interface or protocol to read the data from the data source, which may involve hardware connection, software programming, or network communication. The read data is then preprocessed as necessary, such as format conversion, data cleansing, or secure encryption, to ensure data quality and security. Finally, the processed data is prepared for input into the communication system for encoding, transmission, and final decoding and reconstruction.

[0071] Throughout the entire process, acquiring the original source data is crucial, as it serves as the starting point for the entire communication chain and forms the data foundation for all subsequent processing steps. Only by accurately and completely acquiring the original source data can the subsequent encoding, transmission, and decoding reconstruction processes be carried out efficiently, thereby achieving high-quality data communication.

[0072] Step S202: extract multiple attribute information from the original source data, and encode the multiple attribute information to obtain target source data.

[0073] Specifically, "raw source data" refers to unprocessed, initial data, such as raw images, audio, or text files. "Multiple attribute information" refers to the different features extracted from this raw data. These features represent the core content and semantics of the data. For example, for images, these might be color, texture, and shape; for text, these might be keywords and themes. "Encoding" is the process of converting this attribute information into a format suitable for transmission within a communication system.

[0074] The specific implementation process of this solution includes: First, using specific algorithms or models, such as deep learning networks, to analyze the original source data and identify its attribute information. These algorithms can understand the structure and content of the data and extract the elements that best represent the characteristics of the original data. Next, the extracted attribute information is encoded by an encoder, which converts this information into a more compact format suitable for transmission. This usually involves data compression to reduce the bandwidth required for transmission. The encoding process may also include error detection and correction mechanisms to enhance the robustness of the data during transmission.

[0075] In this way, the original source data is converted into target source data, i.e., encoded attribute information. This target source data not only contains the key information needed to reconstruct the original data, but is also smaller and more suitable for transmission over the communication channel. This approach improves communication efficiency by reducing the amount of data required to be transmitted. It also improves communication reliability by allowing key data characteristics to be transmitted more effectively despite limited bandwidth and potential channel interference. Ultimately, this encoded attribute information can be decoded at the receiving end and used to reconstruct the original source data or for further data processing and analysis.

[0076] In step S203, the target source data is sent to the receiving end, which decodes the target source data to obtain decoded attribute information, and uses the decoded attribute information to reconstruct the original source data through the generative receiver to obtain reconstructed source data.

[0077] Specifically, "target source data" refers to encoded data containing multiple attribute information of the original source data, ready to be sent over a communication network. The "receiving end" is the other end of the communication process, responsible for receiving and processing the data from the sending end. "Decoding" is the process of restoring the encoded data to its original attribute information, which usually involves reversing the algorithm used for encoding. A "generative receiver" is a system that uses artificial intelligence techniques, particularly generative models, to reconstruct the original source data based on the decoded attribute information.

[0078] The implementation process of this solution involves the following: First, the transmitter sends the target source data to the receiver via a communication channel. This data may be affected by noise or other interference during transmission. Upon receiving the data, the receiver uses appropriate decoding techniques to decode the target source data. This step aims to extract the attribute information embedded during encoding. The decoded attribute information contains the key details needed to reconstruct the original source data.

[0079] The decoded attribute information is then fed into a generative receiver. The receiver uses this information to reconstruct the original source data using its internal generative model, such as a diffusion model or a generative adversarial network (GAN). This process involves a complex algorithm that gradually constructs new data similar to the original data based on the attribute information, while continuously adjusting and optimizing the generated data until it matches the characteristics of the original data.

[0080] In this way, the receiver is able to recover an approximate version of the original source data from the target source data, namely the reconstructed source data. This approach not only improves the efficiency of data transmission by reducing the amount of data that needs to be transmitted, but also enhances the reliability of data transmission by allowing key features of the data to be transmitted and reconstructed more effectively despite limited bandwidth and possible channel interference. Ultimately, this ensures high-quality data communication even under less-than-ideal communication conditions.

[0081] In this way, by obtaining the original source data, then extracting multiple attribute information from it and encoding it to generate the target source data, this process significantly improves communication efficiency and data compression rate. During the data transmission process, the target source data is sent to the receiving end, which recovers the original attribute information through decoding operations, laying the foundation for data reconstruction. Using the decoded attribute information, the generative receiver can reconstruct reconstructed source data that is highly similar to the original source data. This not only reduces data loss and errors during transmission, but also improves the reliability of data transmission. In addition, since only attribute information is transmitted instead of the entire original data, this solution optimizes bandwidth usage and reduces the demand for communication resources, enabling efficient and reliable data transmission even under limited bandwidth conditions.

[0082] In some optional embodiments, multiple attribute information is extracted from the original source data, including:

[0083] Semantic feature decoupler is used to extract multiple attribute information from the original source data.

[0084] Specifically, a "semantic feature decoupler" is a specially designed algorithm or system that extracts different "attribute information" from "raw source data." "Raw source data" refers to the original data before any processing, such as an unedited photo, a raw recording, or a piece of unprocessed text. "Attribute information" refers to the features contained in the raw data that represent the data's primary semantic content, such as color and texture in an image, pitch and rhythm in audio, or themes and keywords in text.

[0085] The specific implementation process of this solution includes: the semantic feature decoupler first analyzes the original source data and identifies the elements therein. This process may involve complex data analysis techniques, such as machine learning or deep learning, especially when the data type being processed is complex. The decoupler will apply corresponding feature extraction techniques based on the type and structure of the data. For example, in image processing, edge detection and color analysis may be used; in audio processing, frequency component analysis may be used; and in text processing, natural language processing may be performed. Specifically, taking the original source data as an image as an example, the corresponding semantic feature decoupler consists of an image texture extractor and a color extractor. The texture extractor uses the LBP (Local Binary Pattern) algorithm to perform texture extraction and obtain the first attribute of the image, "texture." The color extractor of the image is composed of a downsampling layer, which obtains the second attribute of the image, "color," by downsampling the average pixel to obtain a color mosaic.

[0086] The extracted attribute information is then used for further processing, such as encoding, transmission, or storage. In this way, the semantic feature decoupler not only helps identify and extract key information from the data, but also provides a more compact and efficient data representation for subsequent communication and data processing steps. The advantage of this approach is that it can reduce data redundancy and improve the efficiency of data transmission while retaining enough information to facilitate accurate reconstruction or further analysis of the data. This is particularly beneficial for systems that need to process large amounts of data or communicate in bandwidth-constrained environments.

[0087] In this way, a semantic feature decoupler is used to extract multiple attribute information from the original source data. By accurately extracting multiple attribute information, this method greatly improves the efficiency of data transmission because it allows the system to transmit only the essential part of the data rather than the entire original data, thereby reducing the required bandwidth and storage space. Secondly, this feature extraction process enhances the data compression rate, allowing more information to be transmitted with limited communication resources. In addition, because attribute information usually contains the core semantics of the data, it can maintain high information quality and semantic integrity even when the data volume is reduced. This not only helps to reduce errors and losses during the transmission process, but also improves the accuracy of data reconstruction, allowing the receiver to reconstruct content that is highly similar to the original data based on this key information.

[0088] In some optional embodiments, a semantic feature decoupler is used to extract multiple attribute information from the original source data, including:

[0089] Determine a semantic feature decoupler corresponding to the source type according to the source type of the original source data;

[0090] A semantic feature decoupler corresponding to the source type is used to extract multiple attribute information from the original source data.

[0091] Specifically, "raw source data" refers to the original information before any processing, such as images, audio, and text. "Source type" refers to the type or category of the original source data, such as image, audio, or text. The "semantic feature decoupler" is a specially designed algorithm or system that extracts key semantic features corresponding to different types of source data.

[0092] The specific implementation process of this solution involves: first, identifying and determining the source type of the original source data by analyzing the data's structure, format, or content. For example, if the data is an image file, the source type is image; if it is an audio file, the source type is audio. Next, based on the determined source type, the system selects or configures a corresponding semantic feature decoupler. This decoupler is designed for a specific type of source data and can understand and process that type of data and extract meaningful attribute information from it.

[0093] For example, for image sources, the decoupler might use image processing techniques to extract visual features such as color, texture, and shape; for audio sources, the decoupler might analyze audio features such as frequency, rhythm, and pitch; and for text sources, the decoupler might identify text features such as keywords, themes, and emotions. In this way, the decoupler can extract attribute information from the raw source data that best represents its content and meaning, providing critical data for subsequent data processing and communication.

[0094] The advantage of this approach is that it can adopt the most appropriate feature extraction strategy for different types of data, thereby improving the accuracy and effectiveness of feature extraction. This not only helps preserve the core semantics of the original data but also optimizes data storage and transmission, as only the most representative feature information is extracted and processed. Furthermore, this approach improves the system's flexibility and adaptability, as it can be easily extended to new source types by simply adding or adjusting the corresponding semantic feature decouplers.

[0095] In this way, by determining the corresponding semantic feature decoupler based on the source type of the original source data and using this decoupler to extract multiple attribute information from the original source data, accurate processing and efficient transmission of different types of data are achieved. This approach ensures that the feature extraction process closely matches the characteristics of the data, thereby improving the accuracy and relevance of the extracted attribute information. Secondly, by customizing the decoupler for different source types, data can be compressed more efficiently because only the most representative and informative features are extracted and encoded, which reduces the volume of transmitted data and optimizes bandwidth usage. In addition, this approach enhances the flexibility and scalability of data transmission because it allows the system to easily adapt to new source types by simply adding or adjusting the corresponding decoupler. Finally, this feature extraction for specific source types also helps improve the quality of data reconstruction, because the receiver can use this precise feature information to reconstruct source data that is closer to the original data.

[0096] In some optional embodiments, encoding multiple attribute information to obtain target source data includes:

[0097] A source-channel joint encoder is used to encode multiple attribute information to obtain the target source data.

[0098] Specifically, the "joint source-channel encoder" is an advanced data processing unit that encodes "attribute information" extracted from the original source data to generate "target source data" suitable for transmission over a communication channel. "Attribute information" refers to the distinct characteristics of the original data, such as the texture and color of an image or the pitch and rhythm of audio, which represent the core content of the original data. The "target source data" is the encoded attribute information, specifically formatted for efficient transmission during communication.

[0099] The specific implementation process of this solution involves the following: First, attribute information is input into a joint source-channel encoder. The encoder processes this information using specific algorithms, which may include compression, encryption, or other forms of data transformation to reduce the data size and improve its robustness during transmission. The encoding process aims to optimize the data to adapt to specific channel conditions. For example, it may adjust the encoding strategy based on the channel's bandwidth limitations and noise characteristics. In this way, the encoder ensures the efficiency and reliability of attribute information during transmission while minimizing the risk of data loss or corruption.

[0100] After encoding, the generated target source data is transmitted to the receiver. This data not only contains the critical information needed to reconstruct the original source data, but also, thanks to the optimized encoding process, offers improved transmission efficiency and interference immunity. At the receiver, a corresponding decoder decodes the received target source data, recovering the original attribute information and enabling accurate data reconstruction. This approach not only improves data transmission efficiency but also enhances the reliability of the communication system, enabling it to maintain high-quality data transmission under various channel conditions.

[0101] In this way, a joint source-channel encoder is used to encode the attribute information to obtain the target source data. This encoding method reduces the amount of data required for transmission by compressing and optimizing the attribute information, thereby improving bandwidth efficiency and data transmission speed. Secondly, the joint source-channel encoder takes into account channel characteristics, such as bandwidth limitations and noise interference, during the encoding process, making the encoded target source data more adaptable to the channel environment and enhancing the reliability and robustness of data transmission. In addition, the encoding process may also include error detection and correction mechanisms, which help reduce errors and losses during data transmission and ensure data integrity. Finally, because the target source data contains the core content of the original source data, the receiver can use this data for effective decoding and reconstruction, thereby achieving efficient data communication while maintaining data quality. In summary, this feature provides a solution for modern communication systems that both saves resources and ensures quality through intelligent encoding and optimized data transmission.

[0102] The above embodiments describe in detail the processing at the transmitting end and the receiving end, respectively. To facilitate an overall understanding of the technical solution of this application, the present embodiment describes the communication process of the solution as a whole. The method includes: the transmitting end obtains original source data, extracts multiple attribute information from the original source data, encodes the multiple attribute information to obtain target source data, and sends the target source data to the receiving end;

[0103] The receiving end decodes the target source data to obtain decoded attribute information, and uses the decoded attribute information to reconstruct the original source data through a generative receiver to obtain reconstructed source data.

[0104] Specifically, the "sender" refers to the party responsible for sending information during the communication process, while the "receiver" refers to the party receiving the information. "Raw source data" refers to the unprocessed data initially acquired by the sender. It can be any form of information, such as images, audio, or text. "Attribute information" refers to the different features extracted from the raw source data. These features represent the core content of the original data, such as the color and texture of an image, the pitch and rhythm of audio, and the theme and keywords of a text.

[0105] For the specific implementation process of this solution, see Figure 3 , Figure 3 The figure below is a schematic diagram of the overall process of the semantic communication method in one embodiment of the present disclosure. The process includes: the transmitter first obtains the original source data, then uses a semantic feature decoupler to extract multiple attribute information from the original source data. This attribute information is then encoded using a source-channel joint encoder, that is, converted into a format suitable for transmission over a communication channel. This step includes compressing the data to reduce the bandwidth required for transmission. After encoding, the "target source data" is obtained, which is then sent to the receiver via the channel.

[0106] After receiving the target source data, the receiver uses a source-channel joint decoder to decode the target source data to obtain decoded attribute information, and inputs the decoded attribute information into a conditional control network. The conditional control network generates a control signal corresponding to the decoded attribute information based on the decoded attribute information; finally, the decoded attribute information and the control signal are input into the diffusion model, and the control signal guides the diffusion model to reconstruct the original source data according to the decoded attribute information to obtain the reconstructed source data.

[0107] In this way, the transmitter first obtains the original source data, then extracts and encodes multiple attribute information to generate the target source data, which is then sent to the receiver. The receiver decodes the received target source data, recovers the attribute information, and uses this information to reconstruct the original source data through a generative receiver. This method significantly reduces the amount of data transmitted by transmitting only the attribute information rather than the entire original data, thereby improving transmission efficiency and reducing bandwidth consumption. Secondly, the encoding process enhances the data's anti-interference ability, making the data more reliable when transmitted over noisy channels. Thirdly, the generative receiver reconstructs the data using the decoded attribute information, which not only preserves the core semantics of the original data but also tolerates errors during transmission to a certain extent, improving the accuracy of data reconstruction. Finally, this method also enhances the flexibility and scalability of the communication system because it can adapt to different types of original source data and provide an effective data transmission solution for various communication scenarios. In summary, this feature significantly improves the efficiency, reliability, and flexibility of the communication system by optimizing the data extraction, encoding, transmission, and reconstruction processes.

[0108] The following describes an embodiment of the device of the present application, which can be used to execute the semantic communication method in the above embodiment of the present application. For details not disclosed in the embodiment of the device of the present application, please refer to the embodiment of the semantic communication method in the above embodiment of the present application.

[0109] The present disclosure also provides a semantic communication device 400, such as Figure 4As shown, it is applied to the receiving end and includes:

[0110] The receiving module 401 is configured to receive target source data sent from a transmitting end, wherein the target source data is obtained by extracting multiple attribute information from the original source data and encoding the multiple attribute information;

[0111] The decoding module 402 is used to decode the target source data to obtain decoded attribute information;

[0112] The reconstruction module 403 is configured to utilize the decoded attribute information to reconstruct the original source data through a generative receiver to obtain reconstructed source data.

[0113] In some optional embodiments, the decoding module 402 decodes the target source data to obtain decoded attribute information, including:

[0114] The source-channel joint decoder is used to decode the target source data to obtain decoded attribute information.

[0115] In some optional embodiments, the reconstruction module 403 uses the decoded attribute information to reconstruct the original source data through a generative receiver to obtain the reconstructed source data, including:

[0116] The decoded attribute information is input into a conditional control network, and the conditional control network generates a control signal corresponding to the decoded attribute information according to the decoded attribute information;

[0117] The decoded attribute information and the control signal are input into the diffusion model, and the control signal is used to guide the diffusion model to reconstruct the original source data according to the decoded attribute information to obtain the reconstructed source data.

[0118] The present disclosure also provides a semantic communication device 500, such as Figure 5 As shown, it is applied to the sending end and includes:

[0119] Acquisition module 501, used to acquire original source data;

[0120] The encoding module 502 is used to extract multiple attribute information from the original source data and encode the multiple attribute information to obtain the target source data;

[0121] The sending module 503 is used to send the target source data to the receiving end, which decodes the target source data to obtain decoded attribute information, and uses the decoded attribute information to reconstruct the original source data through the generative receiver to obtain reconstructed source data.

[0122] In some optional embodiments, the encoding module 502 extracts multiple attribute information from the original source data, including:

[0123] Semantic feature decoupler is used to extract multiple attribute information from the original source data.

[0124] In some optional embodiments, the encoding module 502 uses a semantic feature decoupler to extract multiple attribute information from the original source data, including:

[0125] Determine a semantic feature decoupler corresponding to the source type according to the source type of the original source data;

[0126] A semantic feature decoupler corresponding to the source type is used to extract multiple attribute information from the original source data.

[0127] In some optional embodiments, the encoding module 502 encodes multiple attribute information to obtain target source data, including:

[0128] A source-channel joint encoder is used to encode multiple attribute information to obtain the target source data.

[0129] In the technical solutions disclosed herein, the acquisition, storage, and application of user personal information involved comply with the provisions of relevant laws and regulations and do not violate public order and good morals.

[0130] According to an embodiment of the present disclosure, the present disclosure also provides an electronic device, a readable storage medium, and a computer program product.

[0131] Figure 6 A schematic block diagram of an example electronic device 600 that can be used to implement embodiments of the present disclosure is shown. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device can also represent various forms of mobile devices, such as personal digital assistants, cellular phones, smartphones, wearable devices, and other similar computing devices. The components shown herein, their connections and relationships, and their functions are provided as examples only and are not intended to limit the implementation of the present disclosure described and / or claimed herein.

[0132] like Figure 6As shown, electronic device 600 includes a computing unit 601, which can perform various appropriate actions and processes according to a computer program stored in a read-only memory (ROM) 602 or a computer program loaded from a storage unit 608 into a random access memory (RAM) 603. Various programs and data required for the operation of device 600 can also be stored in RAM 603. Computing unit 601, ROM 602, and RAM 603 are connected to each other via a bus 604. An input / output (I / O) interface 605 is also connected to bus 604.

[0133] Various components in device 600 are connected to I / O interface 605, including an input unit 606, such as a keyboard, mouse, etc.; an output unit 607, such as various types of displays, speakers, etc.; a storage unit 608, such as a magnetic disk, optical disk, etc.; and a communication unit 609, such as a network card, modem, wireless communication transceiver, etc. The communication unit 609 allows device 600 to exchange information / data with other devices via a computer network such as the Internet and / or various telecommunication networks.

[0134] The computing unit 601 can be a variety of general and / or special processing components with processing and computing capabilities. Some examples of the computing unit 601 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various dedicated artificial intelligence (AI) computing chips, various computing units that run machine learning model algorithms, digital signal processors (DSPs), and any appropriate processors, controllers, microcontrollers, etc. The computing unit 601 performs the various methods and processes described above, such as the semantic communication method. For example, in some embodiments, the semantic communication method can be implemented as a computer software program that is tangibly contained in a machine-readable medium, such as a storage unit 608. In some embodiments, part or all of the computer program can be loaded and / or installed on the device 600 via the ROM 602 and / or the communication unit 609. When the computer program is loaded into the RAM 603 and executed by the computing unit 601, one or more steps of the applet distribution described above can be performed. Alternatively, in other embodiments, the computing unit 601 can be configured to perform the semantic communication method in any other appropriate manner (e.g., by means of firmware).

[0135] Various embodiments of the systems and techniques described herein can be implemented in digital electronic circuit systems, integrated circuit systems, field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), application specific standard products (ASSPs), system-on-chip systems (SOCs), programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various embodiments can include being implemented in one or more computer programs that are executable and / or interpreted on a programmable system comprising at least one programmable processor, which can be a special purpose or general purpose programmable processor that can receive data and instructions from a storage system, at least one input device, and at least one output device, and transmit data and instructions to the storage system, the at least one input device, and the at least one output device.

[0136] The program code for implementing the method of the present disclosure can be written in any combination of one or more programming languages. These program codes can be provided to a processor or controller of a general-purpose computer, a special-purpose computer, or other programmable data processing device so that when the program code is executed by the processor or controller, the functions / operations specified in the flow chart and / or block diagram are implemented. The program code can be executed entirely on the machine, partially on the machine, as a stand-alone software package, partially on the machine and partially on a remote machine, or entirely on a remote machine or server.

[0137] In the context of the present disclosure, a machine-readable medium can be a tangible medium that can contain or store a program for use by or in conjunction with an instruction execution system, device or equipment. A machine-readable medium can be a machine-readable signal medium or a machine-readable storage medium. A machine-readable medium can include, but is not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or equipment, or any suitable combination of the foregoing. A more specific example of a machine-readable storage medium can include an electrical connection based on one or more lines, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

[0138] To provide interaction with a user, the systems and techniques described herein can be implemented on a computer having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user; and a keyboard and pointing device (e.g., a mouse or trackball) through which the user can provide input to the computer. Other types of devices can also be used to provide interaction with the user; for example, the feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including acoustic input, voice input, or tactile input).

[0139] The systems and techniques described herein can be implemented in a computing system that includes back-end components (e.g., as a data server), or a computing system that includes middleware components (e.g., an application server), or a computing system that includes front-end components (e.g., a user computer having a graphical user interface or a web browser through which a user can interact with implementations of the systems and techniques described herein), or a computing system that includes any combination of such back-end components, middleware components, or front-end components. The components of the system can be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include a local area network (LAN), a wide area network (WAN), and the Internet.

[0140] A computer system may include a client and a server. The client and server are generally remote from each other and typically interact through a communication network. The client-server relationship arises through computer programs running on the respective computers and having a client-server relationship with each other. The server may be a cloud server, a server in a distributed system, or a server integrated with a blockchain.

[0141] It should be understood that the various forms of processes shown above can be used to reorder, add, or delete steps. For example, the steps described in this disclosure can be performed in parallel, sequentially, or in a different order, as long as the desired results of the technical solutions of this disclosure can be achieved, and this document is not limited here.

[0142] The above specific embodiments do not constitute a limitation on the scope of protection of this disclosure. Those skilled in the art will appreciate that various modifications, combinations, sub-combinations, and substitutions may be made based on design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this disclosure shall be included within the scope of protection of this disclosure.

Claims

1. A semantic communication method, applied to a receiving end, wherein: The method comprises: Receiving target source data sent from a transmitting end, wherein the target source data is obtained by extracting multiple attribute information from the original source data and encoding the multiple attribute information; Decoding the target source data to obtain decoded attribute information; The decoded attribute information is used to reconstruct the original source data through a generative receiver to obtain reconstructed source data.

2. The method according to claim 1, wherein The decoding of the target source data to obtain decoded attribute information includes: The target source data is decoded using a source-channel joint decoder to obtain decoded attribute information.

3. The method according to claim 1 or 2, wherein: The generative receiver includes: a diffusion model and a conditional control network; The method of reconstructing the original source data by using the decoded attribute information through a generative receiver to obtain the reconstructed source data includes: Inputting the decoded attribute information into the conditional control network, and generating a control signal corresponding to the decoded attribute information through the conditional control network according to the decoded attribute information; The decoded attribute information and the control signal are input into the diffusion model, and the control signal is used to guide the diffusion model to reconstruct the original source data according to the decoded attribute information to obtain reconstructed source data.

4. A semantic communication method, applied to a sending end, wherein: The method comprises: Obtain original source data; Extracting multiple attribute information from the original source data and encoding the multiple attribute information to obtain target source data; The target source data is sent to a receiving end, the target source data is decoded by the receiving end to obtain decoded attribute information, and the decoded attribute information is used to reconstruct the original source data through a generative receiver to obtain reconstructed source data.

5. The method according to claim 4, wherein The extracting multiple attribute information from the original source data includes: A semantic feature decoupler is used to extract multiple attribute information from the original source data.

6. The method according to claim 5, wherein: The extracting of multiple attribute information from the original source data using a semantic feature decoupler includes: Determining a semantic feature decoupler corresponding to the source type according to the source type of the original source data; A plurality of attribute information is extracted from the original source data using a semantic feature decoupler corresponding to the source type.

7. The method according to claim 4, wherein: The encoding of the plurality of attribute information to obtain target source data includes: A source-channel joint encoder is used to encode the multiple attribute information to obtain target source data.

8. A semantic communication method, wherein: The method comprises: The sending end obtains original source data, extracts multiple attribute information from the original source data, encodes the multiple attribute information to obtain target source data, and sends the target source data to the receiving end; The receiving end decodes the target source data to obtain decoded attribute information, and uses the decoded attribute information to reconstruct the original source data through a generative receiver to obtain reconstructed source data.

9. A semantic communication device, applied to a receiving end, wherein: include: A receiving module is configured to receive target source data sent from a transmitting end, wherein the target source data is obtained by extracting multiple attribute information from the original source data and encoding the multiple attribute information; A decoding module, configured to decode the target source data to obtain decoded attribute information; The reconstruction module is used to reconstruct the original source data by using the decoded attribute information through a generative receiver to obtain reconstructed source data.

10. A semantic communication device, applied to a sending end, wherein: include: An acquisition module is used to obtain original source data; An encoding module, configured to extract multiple attribute information from the original source data and encode the multiple attribute information to obtain target source data; The sending module is used to send the target source data to the receiving end, decode the target source data through the receiving end to obtain decoded attribute information, and use the decoded attribute information to reconstruct the original source data through the generative receiver to obtain reconstructed source data.

11. An electronic device, wherein: include: at least one processor; as well as a memory communicatively connected to the at least one processor; wherein, The memory stores instructions that can be executed by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to perform the method according to any one of claims 1 to 8.

12. A non-transitory computer-readable storage medium storing computer instructions, wherein: The computer instructions are used to cause the computer to execute the method according to any one of claims 1-8.

13. A computer program product comprising a computer program, which, when executed by a processor, implements the method according to any one of claims 1 to 8.

Images