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 particularly relates to the technical field of semantic communication. The method comprises the following steps: receiving superposition features from a first user and a second user, and extracting a first user feature and a second user feature from the superposition features according to a pre-stored synchronization header sequence; adopting a pre-stored rearrangement index to recover the original sequence of the first user feature and the original sequence of the second user feature, and obtaining the sorted first user feature and the sorted second user feature; and respectively performing data recovery on the sorted first user characteristics and the sorted second user characteristics, recoding and sorting the recovered data by respectively adopting a semantic encoder and a rearrangement index corresponding to the receiving user, and superposing the sorted data together and sending the superposed data to a receiving end. Therefore, the receiving user of the receiving end recovers to obtain the corresponding original image data, and the communication robustness is enhanced while the communication accuracy is kept.

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] Physical network coding (PNC) is a technology that implements data coding at the physical layer. It allows relay nodes to directly superimpose and decode signals from different users, thereby improving the spectrum efficiency of wireless communication systems. In two-way relay communication (TWRC) scenarios, PNC technology is widely used because it allows relay nodes to receive and transmit information simultaneously, reducing transmission delay and increasing system throughput. Orthogonal mode division multiple access (OMDMA) is a multi-user access technology that allows multiple users to communicate on the same time and frequency resources without interfering with each other. In TWRC systems, PNC technology can be combined with OMDMA to further improve system capacity and efficiency. OMDMA ensures the orthogonality of signals from different users, while PNC technology allows relay nodes to efficiently process these signals.

[0003] However, with the development of 6G wireless communication systems and the rise of semantic communication, existing PNC technology faces new challenges, especially when dealing with phase offset and sign bit offset issues in signal transmission. These issues are particularly prominent in high-density multi-user scenarios and can cause signal distortion and affect communication quality. Therefore, to meet the needs of semantic communication, a new solution is urgently needed to overcome the limitations of existing technologies and realize more efficient and robust wireless communication systems. 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 relay node. The method includes:

[0006] receiving superimposed features from a first user and a second user, the superimposed features including a first semantic feature and a second semantic feature, the first semantic feature being obtained by performing semantic feature extraction on a first input image sent by the first user, reordering the image using a preset first reordering index, and then adding a first synchronization header sequence, and the second semantic feature being obtained by performing semantic feature extraction on a second input image sent by the second user, reordering the image using a preset second reordering index, and then adding a second synchronization header sequence;

[0007] Extracting features corresponding to the first user and the second user from the superimposed features according to a pre-stored synchronization header sequence to obtain first user features and second user features;

[0008] Restore the original order of the first user features and the second user features respectively using a pre-stored rearrangement index to obtain a sorted first user feature and a sorted second user feature;

[0009] Data recovery is performed on the sorted first user features and the sorted second user features respectively, and the recovered data are re-encoded and sorted using a semantic encoder and a reordering index corresponding to the receiving user respectively, and the sorted data are superimposed together and sent to the receiving end, so that the receiving user at the receiving end can recover the corresponding original image data.

[0010] 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:

[0011] Acquire a first input image sent by a first user and a second input image sent by a second user;

[0012] Extracting semantic features from the first input image, and rearranging the extracted features using a preset first rearrangement index and then adding a first synchronization header sequence to obtain a first semantic feature;

[0013] Extracting semantic features from the second input image, and rearranging the extracted features using a preset second rearrangement index and then adding a second synchronization header sequence to obtain a second semantic feature;

[0014] The first semantic feature and the second semantic feature are superimposed to obtain a superimposed feature, which is then sent to a relay node. The relay node extracts features corresponding to the first user and the second user from the superimposed feature according to a pre-stored synchronization header sequence to obtain a first user feature and a second user feature. The pre-stored reordering index is used to restore the original order of the first user feature and the second user feature to obtain a sorted first user feature and a sorted second user feature. Data of the sorted first user feature and the sorted second user feature are restored, and the restored data are re-encoded and sorted using a semantic encoder and a reordering index corresponding to the receiving user. The sorted data are superimposed and sent to a receiving end, so that the receiving user at the receiving end can restore the corresponding original image data.

[0015] According to a third aspect of the present disclosure, a semantic communication method is provided, which is applied to a receiving end, and the method includes:

[0016] Receive aliased data sent from a relay node, the relay node receiving superimposed features from a first user and a second user, the superimposed features including a first semantic feature and a second semantic feature, the first semantic feature being obtained by performing semantic feature extraction on a first input image sent by the first user, reordering the image using a preset first reordering index, and then adding a first synchronization header sequence, and the second semantic feature being obtained by performing semantic feature extraction on a second input image sent by the second user, reordering the image using a preset second reordering index, and then adding a second synchronization header sequence; extracting features corresponding to the first user and the second user from the superimposed features according to a pre-stored synchronization header sequence to obtain first user features and second user features; using a pre-stored reordering index to restore the original order of the first user features and the second user features, respectively, to obtain ordered first user features and ordered second user features; performing data recovery on the ordered first user features and the ordered second user features, respectively, to obtain recovered data, and re-encoding and reordering the recovered data using a semantic encoder and reordering index corresponding to the receiving user, and superimposing the ordered data and sending them to a receiving end;

[0017] Using the first rearrangement index to obtain data corresponding to the second user from the sorted data to obtain second user data;

[0018] Using the second rearrangement index to obtain data corresponding to the first user from the sorted data to obtain first user data;

[0019] Performing data recovery on the second user data using the first decoder to obtain an original image sent by the second user;

[0020] The first user data is recovered using a second decoder to obtain an original image sent by the first user.

[0021] According to a fourth aspect of the present disclosure, a semantic communication method is provided, the method comprising:

[0022] The transmitting end obtains a first input image sent by a first user and a second input image sent by a second user; performs semantic feature extraction on the first input image, rearranges it using a preset first rearrangement index, and then adds a first synchronization header sequence to obtain a first semantic feature; performs semantic feature extraction on the second input image, rearranges it using a preset second rearrangement index, and then adds a second synchronization header sequence to obtain a second semantic feature; superimposes the first semantic feature and the second semantic feature to obtain a superimposed feature, and then sends the superimposed feature to a relay node;

[0023] The relay node receives superimposed features from a first user and a second user, extracts features corresponding to the first user and the second user from the superimposed features according to a pre-stored synchronization header sequence, and obtains first user features and second user features; uses a pre-stored reordering index to restore the original order of the first user features and the second user features, respectively, to obtain sorted first user features and sorted second user features; performs data recovery on the sorted first user features and the sorted second user features, and re-encodes and re-sorts the recovered data using a semantic encoder corresponding to the receiving user and the reordering index, and then superimposes the sorted data and sends it to a receiving end;

[0024] The receiving end receives the sorted data sent from the relay node, uses the first reordering index to obtain data corresponding to the second user from the data, and obtains second user data; uses the second reordering index to obtain data corresponding to the first user from the data, and obtains first user data; uses the first decoder to recover the second user data to obtain the original image sent by the second user; uses the second decoder to recover the first user data to obtain the original image sent by the first user.

[0025] According to a fifth aspect of the present disclosure, a semantic communication device is provided, which is applied to a relay node and includes:

[0026] a first receiving module, configured to receive superimposed features from a first user and a second user, the superimposed features including a first semantic feature and a second semantic feature, the first semantic feature being obtained by performing semantic feature extraction on a first input image sent by the first user, reordering the image using a preset first reordering index, and then adding a first synchronization header sequence; and the second semantic feature being obtained by performing semantic feature extraction on a second input image sent by the second user, reordering the image using a preset second reordering index, and then adding a second synchronization header sequence;

[0027] an extraction module, configured to extract features corresponding to the first user and the second user from the superimposed features according to a pre-stored synchronization header sequence, to obtain first user features and second user features;

[0028] a rearrangement module, configured to restore the original order of the first user features and the second user features respectively using a pre-stored rearrangement index to obtain the sorted first user features and the sorted second user features;

[0029] The recovery module is used to respectively recover the data of the sorted first user features and the sorted second user features, and re-encode and sort the recovered data using the semantic encoder and reordering index corresponding to the receiving user, and then superimpose the sorted data and send them to the receiving end, so that the receiving user at the receiving end can recover the corresponding original image data.

[0030] According to a sixth aspect of the present disclosure, a semantic communication device is provided, applied to a sending end, comprising:

[0031] An acquisition module, configured to acquire a first input image sent by a first user and a second input image sent by a second user;

[0032] A first processing module is configured to extract semantic features from the first input image, and to rearrange the extracted features using a preset first rearrangement index and then add a first synchronization header sequence to obtain a first semantic feature;

[0033] A second processing module is configured to extract semantic features from the second input image, and to rearrange the extracted features using a preset second rearrangement index and then add a second synchronization header sequence to obtain a second semantic feature;

[0034] A sending module is used to superimpose the first semantic feature and the second semantic feature to obtain a superimposed feature and then send it to a relay node, and the relay node extracts the features corresponding to the first user and the second user from the superimposed feature according to a pre-stored synchronization header sequence to obtain the first user feature and the second user feature; uses a pre-stored reordering index to restore the original order of the first user feature and the second user feature to obtain the sorted first user feature and the sorted second user feature; performs data recovery on the sorted first user feature and the sorted second user feature, and re-encodes and sorts the recovered data using the semantic encoder corresponding to the receiving user and the reordering index, and superimposes the sorted data and sends them to the receiving end, so that the receiving user at the receiving end can restore the corresponding original image data.

[0035] According to a seventh aspect of the present disclosure, a semantic communication device is provided, applied to a receiving end, comprising:

[0036] a second receiving module, configured to receive sorted data sent from a relay node, the relay node receiving superimposed features from a first user and a second user, the superimposed features including a first semantic feature and a second semantic feature, the first semantic feature being obtained by performing semantic feature extraction on a first input image sent by the first user, re-arranging the image using a preset first re-arrangement index, and then adding a first synchronization header sequence; and the second semantic feature being obtained by performing semantic feature extraction on a second input image sent by the second user, re-arranging the image using a preset second re-arrangement index, and then adding a second synchronization header sequence; extracting features corresponding to the first user and the second user from the superimposed features according to a pre-stored synchronization header sequence to obtain first user features and second user features; restoring the original sorting of the first user features and the second user features using a pre-stored re-arrangement index to obtain sorted first user features and sorted second user features; performing data recovery on the sorted first user features and the sorted second user features to obtain recovered data, re-encoding and re-sorting the recovered data using a semantic encoder and re-arrangement index corresponding to the receiving user, and superimposing the sorted data and sending them to a receiving end;

[0037] a first separation module, configured to obtain data corresponding to a second user from the data by using the first rearrangement index to obtain second user data;

[0038] a second separation module, configured to obtain data corresponding to the first user from the data by using the second rearrangement index to obtain first user data;

[0039] a first decoding module, configured to recover the second user data using a first decoder to obtain an original image sent by the second user;

[0040] The second decoding module is configured to perform data recovery on the first user data using a second decoder to obtain an original image sent by the first user.

[0041] According to an eighth aspect of the present disclosure, there is provided an electronic device, including:

[0042] at least one processor; and

[0043] a memory communicatively connected to the at least one processor; wherein,

[0044] 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.

[0045] According to a ninth 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.

[0046] According to a tenth 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.

[0047] The present disclosure provides a semantic communication method, apparatus, device, and storage medium. This method receives superimposed features from a first user and a second user. These features not only include semantic information from their respective input images but are also optimized and sorted using a preset reordering index. A synchronization header sequence is added to the front end of the signal to enhance signal synchronization. This design enables relay nodes to utilize the synchronization header sequence for efficient feature extraction and precise synchronization header detection, accurately isolating each user's characteristic signal. The relay node then uses the pre-stored reordering index to sort and recover the isolated features. This critical step ensures correct signal decoding and, in turn, accurately reconstructs the original image data. Furthermore, to further improve the accuracy and reliability of data transmission, the relay node re-encodes and sorts the recovered data using the semantic encoder and reordering index corresponding to the receiving user. The sorted data is then re-superimposed and transmitted via the downlink. This process not only optimizes the performance of the two-way relay communication system but also significantly improves the quality and efficiency of image data recovery. In this way, the present solution maintains communication accuracy while enhancing communication robustness, enabling stability in changing communication environments and improving communication practicality. This scheme, which combines efficient feature extraction, accurate synchronization header detection, correct signal decoding and high-quality image reconstruction, provides an innovative and effective communication strategy for two-way relay communication.

[0048] 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

[0049] The accompanying drawings are provided to facilitate a better understanding of the present invention and do not constitute a limitation of the present disclosure.

[0050] Figure 1 is a schematic diagram of the steps of a semantic communication method in one embodiment of the present disclosure;

[0051] Figure 2 is a schematic diagram of steps of a semantic communication method in another embodiment of the present disclosure;

[0052] Figure 3 is a schematic diagram of steps of a semantic communication method in yet another embodiment of the present disclosure;

[0053] Figure 4 It is a schematic diagram of the overall process of the semantic communication method in one embodiment of the present disclosure;

[0054] Figure 5 This is a workflow for data reordering and recovery using a semantic encoder and decoder and a reordering index in one embodiment of the present disclosure;

[0055] Figure 6 is a flowchart of a two-way relay communication system in an embodiment of the present disclosure;

[0056] Figure 7 is a structural block diagram of a semantic encoder in an embodiment of the present disclosure;

[0057] Figure 8 A block diagram of the principle of a semantic communication device in an embodiment of the present disclosure;

[0058] Figure 9 A functional block diagram of a semantic communication device in another embodiment of the present disclosure;

[0059] Figure 10 A functional block diagram of a semantic communication device in another embodiment of the present disclosure;

[0060] Figure 11 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

[0061] 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.

[0062] The present disclosure provides a semantic communication method, see Figure 1 As shown, Figure 1 : is a schematic diagram of the steps of a semantic communication method in one embodiment of the present disclosure, which is applied to a relay node and includes:

[0063] Step S101, receiving superimposed features from a first user and a second user, the superimposed features including a first semantic feature and a second semantic feature, the first semantic feature being obtained by extracting semantic features from a first input image sent by the first user, rearranging the image using a preset first rearrangement index, and then adding a first synchronization header sequence, and the second semantic feature being obtained by extracting semantic features from a second input image sent by the second user, rearranging the image using a preset second rearrangement index, and then adding a second synchronization header sequence.

[0064] This communication strategy aims to enable efficient data transmission between a first user and a second user through a relay node. Specifically, in this scheme, "superimposed features" are composite signals containing semantic information from two different users. This information is first extracted from each input image through a "semantic feature extraction" process, a technique that converts image content into transmittable data and aims to capture key information within the image. Subsequently, each user's data is reordered using a "pre-set reordering index." This step alters the data's order to optimize transmission efficiency or meet specific communication protocol requirements. After reordering, each user's data is appended with a unique "synchronization header sequence," such as a Zadoff-Chu sequence. This is a specific signal pattern used for synchronization in communications and helps the receiver accurately identify and distinguish signals from different users. Finally, these two processed semantic feature signals—the first and second semantic features—are superimposed to form a composite signal, which is transmitted over a wireless channel to the relay node. The relay node's task is to receive this superimposed feature and, using pre-stored synchronization header sequence information, extract and separate the characteristic signals belonging to the first and second users, paving the way for subsequent signal processing and data recovery. This process not only improves spectrum utilization, but also enhances communication reliability and efficiency.

[0065] Step S102 : extracting features corresponding to the first user and the second user from the superimposed features according to the pre-stored synchronization header sequence to obtain first user features and second user features.

[0066] Specifically, a "synchronization header sequence" refers to a specific pattern of signals appended before user data transmission. It exhibits excellent autocorrelation properties and helps relay nodes identify and distinguish the signal starting points of different users. The implementation of this solution involves, after receiving the superimposed features from the first and second users, the relay node uses pre-stored synchronization header sequence information—specific synchronization headers pre-installed in each user's signal—to identify and extract the characteristic signals belonging to the first and second users. This extraction process involves analyzing the superimposed signals. By calculating the correlation between the signals and the known synchronization header sequence, the relay node can accurately locate the starting position of each user's signal. Based on this location information, the relay node then separates the characteristic signals of the first and second users from the superimposed signals, namely the "first user signature" and "second user signature." These characteristic signals contain key information about the original user data and serve as the basis for subsequent signal processing and data recovery. In this way, this solution effectively separates and extracts signals in a multi-user environment, providing critical support for ensuring the accuracy and reliability of data transmission.

[0067] Step S103 : using the pre-stored rearrangement index to restore the original order of the first user feature and the second user feature, respectively, to obtain the sorted first user feature and the sorted second user feature.

[0068] Specifically, "reordering index" refers to a predefined data mapping relationship, which is used to reorder user data during the communication process to optimize transmission efficiency or meet specific signal processing requirements. In the specific implementation process of this solution, after the relay node separates the characteristic signals of the first user and the second user from the superimposed signal, the pre-stored reordering index is used to ensure that in a two-way relay communication system, the semantic features from different users are disrupted during encoding and can be accurately restored to the original order at the receiving end, thereby avoiding interference of semantic information between users to improve the quality of the decoded image. That is, this recovery operation is crucial to ensuring the integrity and accuracy of the data, because it allows the relay node to correctly decode and understand the information from different users, and then accurately send the restored characteristic signals to the corresponding receiving users, realizing efficient data transmission and image reconstruction. In this way, this solution not only improves the reliability of data transmission, but also enhances the flexibility and effectiveness of the communication system in managing data from different users.

[0069] In step S104, data of the sorted first user features and the sorted second user features are restored respectively, and the restored data are re-encoded and sorted respectively using the semantic encoder and reordering index corresponding to the receiving user, and the sorted data are superimposed together and sent to the receiving end, so that the receiving user at the receiving end can restore the corresponding original image data.

[0070] Specifically, "data recovery" refers to the process of restoring the sorted user feature signals to the original image data, while the "semantic encoder" is a device used to convert the original image data into a semantic feature signal suitable for transmission. It extracts and encodes key information based on the image content. The "reordering index" is a parameter used to adjust the data order, ensuring that the data is arranged in the predetermined order during transmission.

[0071] The specific implementation process of this solution involves the relay node first recovering the sorted first and second user features. This step involves converting the sorted feature signals back into the original image data. Subsequently, to ensure that the data can be correctly understood by the receiving user, the relay node re-encodes the recovered data using the semantic encoder corresponding to the receiving user. Next, the relay node sorts the re-encoded data using the re-ordering index corresponding to the receiving user to match the receiving user's decoding requirements. Finally, the relay node re-adds the sorted data to the receiving end via the "downlink," the communication link that sends signals from the relay node to the mobile terminal. The receiving user can then recover the corresponding original image data based on the received data and its own semantic decoder, completing the entire communication process. This solution not only improves the accuracy and reliability of data transmission but also optimizes the data recovery and re-encoding processes, making communication more efficient and secure.

[0072] The present disclosure provides a semantic communication method, apparatus, device, and storage medium. This method receives superimposed features from a first user and a second user. These features not only include semantic information from their respective input images but are also optimized and sorted using a preset reordering index. A synchronization header sequence is added to the front end of the signal to enhance signal synchronization. This design enables relay nodes to utilize the synchronization header sequence for efficient feature extraction and precise synchronization header detection, accurately isolating each user's characteristic signal. The relay node then uses the pre-stored reordering index to sort and recover the isolated features. This critical step ensures correct signal decoding and, in turn, accurately reconstructs the original image data. Furthermore, to further improve the accuracy and reliability of data transmission, the relay node re-encodes and sorts the recovered data using the semantic encoder and reordering index corresponding to the receiving user. The sorted data is then re-superimposed and transmitted via the downlink. This process not only optimizes the performance of the two-way relay communication system but also significantly improves the quality and efficiency of image data recovery. In this way, the present solution maintains communication accuracy while enhancing communication robustness, enabling stability in changing communication environments and improving communication practicality. This scheme, which combines efficient feature extraction, accurate synchronization header detection, correct signal decoding and high-quality image reconstruction, provides an innovative and effective communication strategy for two-way relay communication.

[0073] In some optional embodiments, extracting features corresponding to the first user and the second user from the superimposed features according to a pre-stored synchronization header sequence to obtain the first user features and the second user features includes:

[0074] Acquire a pre-stored first synchronization header sequence and a pre-stored second synchronization header sequence;

[0075] Identify the position of the first synchronization header sequence based on the pre-stored correlation between the first synchronization header sequence and the superposition feature, and extract a feature of a preset length from the superposition feature using the position of the first synchronization header sequence as a starting position to obtain a first user feature;

[0076] According to the pre-stored correlation between the second synchronization header sequence and the superposition feature, the position of the second synchronization header sequence is identified, and the position of the second synchronization header sequence is used as the starting position to extract a feature of a preset length from the superposition feature to obtain the second user feature.

[0077] Specifically, a "synchronization header sequence" is a special signal pattern used to help the receiver identify and distinguish the starting points of different user signals during data transmission. Correlation, a measure of similarity between two signals, is used to determine the location of one signal within the other. The specific process of this solution involves the relay node first obtaining two pre-stored synchronization header sequences, one corresponding to the first user and the other to the second user. The relay node then uses a correlation detection algorithm, such as cross-correlation or autocorrelation, to calculate the correlation between the known synchronization header sequences and the received superimposed feature signals. When the correlation reaches a peak, it indicates that the corresponding synchronization header sequence locations have been found. Using these locations as starting points, the relay node extracts a segment of a predetermined length of feature signal from the superimposed feature. This segment contains key user information, namely the "first user feature" and the "second user feature." In this way, the relay node can accurately separate the feature signals belonging to different users from the superimposed signal, laying the foundation for subsequent signal processing and data recovery. This process not only improves the accuracy of signal separation but also enhances the system's robustness to asynchronous signal transmission, thereby optimizing the performance of the two-way relay communication system.

[0078] In this way, by utilizing the correlation between the synchronization header sequence and the superimposed features, the starting point of each user signal is accurately located, ensuring that the characteristic signal of the preset length is extracted from the correct position. This method not only improves the accuracy of signal extraction, but also enhances the system's anti-interference ability because the design of the synchronization header sequence helps to distinguish the signals of different users in complex communication environments. In addition, accurate signal extraction lays a solid foundation for subsequent signal processing and data recovery, enabling relay nodes to more effectively reorder and decode semantic features and thus restore the original image data. Ultimately, this series of precise operations optimizes the performance of the two-way relay communication system, improves the reliability of data transmission and the robustness of communication, and ensures the efficiency and accuracy of communication between users.

[0079] In some optional embodiments, after obtaining the first user feature and the second user feature, the method further includes:

[0080] The first synchronization header sequence and the second synchronization header sequence are removed.

[0081] Specifically, a "synchronization header sequence" is a specific signal pattern appended before user data to aid signal synchronization and identification at the receiving end. In this solution, the "removal of the first and second synchronization header sequences" step occurs after the relay node successfully identifies and uses these sequences to extract the features of the first and second users from the superimposed signal. The specific implementation process of this solution involves: the relay node first locates the synchronization header sequences by calculating the correlation between the superimposed feature signal and the known synchronization header sequences. Once successfully located, the relay node accurately removes the synchronization header sequences from the extracted user feature signal. This process aims to obtain pure user data, devoid of any additional information used for synchronization. After removing the synchronization header sequences, the relay node can further process the remaining user features, such as reordering, decoding, and data recovery, to ensure accurate reconstruction of the original image data and transmission to the intended recipient. This step is crucial for improving communication efficiency, reducing unnecessary data transmission, and enhancing signal processing accuracy. By accurately removing the synchronization header sequences, efficient data transmission and accurate data decoding at the receiving end are ensured, thereby optimizing the entire communication process.

[0082] In this way, by removing the first synchronization header sequence and the second synchronization header sequence, the purity of the user characteristic signal extracted from the superimposed signal is ensured, and the interference that may be caused by the synchronization header sequence is eliminated, thereby improving the accuracy of signal processing. Secondly, removing the synchronization header sequence reduces the amount of data that needs to be transmitted, which directly improves spectrum utilization and reduces bandwidth consumption, allowing more users to share limited communication resources. In addition, this process simplifies the subsequent signal processing process because the relay node no longer needs to process synchronization-related data, thereby speeding up data processing and improving overall communication efficiency. Ultimately, this series of optimization measures not only enhances the reliability of communication, but also allows users to receive high-quality image data faster and more reliably, improving the user experience.

[0083] In some optional embodiments, the original order of the first user feature and the second user feature is restored using a pre-stored reordering index to obtain the sorted first user feature and the sorted second user feature, including:

[0084] Obtaining a pre-stored first re-arrangement index and a pre-stored second re-arrangement index;

[0085] Reordering the first user features using a pre-stored first reordering index to restore the original order of the first user's corresponding features, thereby obtaining ordered first user features;

[0086] The second user features are reordered using the pre-stored second reordering index to restore the original order of the features corresponding to the second user, thereby obtaining the ordered second user features.

[0087] Specifically, "rearrangement index" refers to a series of parameters or mapping relationships used to guide how to restore processed data to its original order. These indexes are very important in communication systems. The use of pre-stored reordering indexes can ensure that in a two-way relay communication system, the semantic features from different users can be accurately restored to their original order at the receiving end after being disrupted during encoding, thereby avoiding interference of semantic information between users to improve the quality of the decoded image. The specific implementation process of this scheme includes that the relay node first obtains two pre-stored reordering indexes, namely the "first reordering index" and the "second reordering index", which correspond to the first user and the second user respectively. Before users send data, their data (such as the semantic features of the image) will be reordered according to their respective reordering indexes to reduce interference between the data, thereby achieving better image restoration.

[0088] When the relay node receives the superimposed features from two users, it uses these pre-stored reordering indexes to reorder the extracted user features. Specifically, the relay node uses the "first reordering index" to reorder the features of the first user to restore their original order, thereby obtaining the "sorted first user features." Similarly, the relay node uses the "second reordering index" to reorder the features of the second user to restore their original order, thereby obtaining the "sorted second user features." The key to this process is that the reordering index not only helps the relay node correctly identify and separate data from different users, but also ensures the integrity and accuracy of the data during transmission. In this way, efficient data recovery and accurate signal processing can be achieved even in complex multi-user communication environments, thereby improving the overall performance and reliability of the communication system.

[0089] In this way, by obtaining and applying the known first and second reordering indexes, semantic information interference between users is avoided to improve the quality of the decoded image. Specifically, the relay node uses these predefined indexes to reorder the first and second user features extracted from the superimposed signal, thereby accurately restoring the original order of each user's data. This process not only ensures the integrity and accuracy of the data, but also, through precise data reordering, the relay node can more efficiently perform subsequent signal processing and data recovery tasks, including decoding and reconstructing the original image data. The implementation of this feature significantly improves the performance of the two-way relay communication system, optimizes data transmission efficiency, and ultimately enhances the robustness of communication, ensuring that users can receive high-quality image data, thereby improving the overall communication experience.

[0090] In some optional embodiments, data recovery is performed on the sorted first user features and the sorted second user features, and the recovered data are re-encoded and sorted using a semantic encoder and a re-ranking index corresponding to the receiving user, and the sorted data are superimposed and sent to the receiving end, including:

[0091] Decoding the sorted first user features and the sorted second user features respectively to obtain decoded first user features and decoded second user features;

[0092] encoding the decoded first user feature using a preset second encoder, and rearranging the decoded first user feature using a second rearrangement index to obtain a rearranged first user feature;

[0093] encoding the decoded second user feature using a preset first encoder, and rearranging the decoded second user feature using a first rearrangement index to obtain a rearranged second user feature;

[0094] The rearranged first user feature and the rearranged second user feature are superimposed and then sent to a receiving end via a downlink.

[0095] Specifically, "decoding" refers to the process of converting received encoded data back to its original form, while an "encoder" is a hardware or software tool used to convert raw data into a format suitable for transmission. The specific process of this solution involves the relay node first decoding the first and second user features, which have been reordered according to the reordering index, to restore the "decoded first user features" and "decoded second user features." Subsequently, to transmit these features to the correct receiving user, the relay node re-encodes the decoded first user features using a "pre-set second encoder" and reorders them again according to the "second reordering index" to generate the "reordered first user features." Similarly, the decoded second user features are encoded using the "pre-set first encoder" and reordered using the "first reordering index" to generate the "reordered second user features." Finally, these re-encoded and reordered features are superimposed and transmitted to the receiving end via the "downlink." This series of steps ensures data accuracy and security during transmission, while optimizing the data transmission process and improving communication efficiency, enabling the receiving end to correctly understand and process the information from the sending end, completing the end-to-end communication process.

[0096] In this way, by decoding the sorted user features, the original data is accurately recovered from the transmitted signal. Next, the decoded features are re-encoded and rearranged using a preset encoder and rearrangement index before being superimposed. This not only optimizes the data format to meet the requirements of downlink transmission, but also further ensures the correct order of the data through index rearrangement, thereby improving the readability and usability of the data. Finally, the rearranged features are sent to the receiving end via the downlink, allowing the receiving user to receive and parse the data without ambiguity. This feature combination achieves efficient data recovery and accurate signal transmission, reduces the need for retransmission due to data errors or disorder, saves communication resources, and improves communication reliability and user experience. In addition, this method enhances the flexibility of the system, can adapt to the needs of different users and different communication environments, and provides an efficient and reliable solution for multi-user communication.

[0097] In some optional embodiments, before receiving the superimposed features from the first user and the second user, the method further includes:

[0098] receiving a first encoder, a first decoder, a first reordering index, and a first synchronization header sequence sent by a first user;

[0099] A second encoder, a second decoder, a second reordering index, and a second synchronization header sequence sent by a second user are received.

[0100] Specifically, an "encoder" and "decoder" are devices or algorithms used for data conversion, responsible for converting raw data (such as an image) into a format suitable for transmission (encoding) and restoring received data to its original format (decoding), respectively. A "reordering index" is a parameter used to guide data reordering, ensuring that data can be correctly restored even after reordering during transmission. A "synchronization header sequence" is a specific signal pattern appended before data transmission to help the receiving end identify and synchronize signals from different users.

[0101] The specific process of this solution includes: the relay node, as the hub of communication, first receives a series of information from the first user, including a first encoder for data encoding, a first decoder for data decoding, a first reordering index for data reordering, and a first synchronization header sequence for signal synchronization. Similarly, the relay node also receives corresponding information from the second user, namely a second encoder, a second decoder, a second reordering index, and a second synchronization header sequence. This process allows the relay node to prepare and process data for each user, ensuring the accuracy and integrity of the data during transmission. The relay node uses the encoder to encode the user's original data into a format suitable for transmission, uses the synchronization header sequence to identify and distinguish the signals of different users, manages and restores the original order of the data through the reordering index, and finally uses the decoder to restore the received encoded data to its original format. This design not only improves the efficiency and reliability of data transmission, but also enhances the flexibility and scalability of the system, enabling the system to adapt to the needs of different users and different communication scenarios.

[0102] In this way, by receiving the encoder, decoder, reordering index and synchronization header sequence sent by the first and second users respectively, it provides the relay node with the necessary information and tools to ensure that it can correctly process and forward signals from different users. Specifically, receiving the first and second encoders enables the relay node to encode the signal in a user-specific manner, thereby adapting to different data characteristics and communication needs. Receiving the corresponding decoder ensures that the relay node can accurately recover the original data from the received signal. Receiving the reordering index allows the relay node to correctly sort and reorganize the data during signal transmission, maintaining the accuracy and integrity of the data. Finally, receiving the synchronization header sequence provides the relay node with a means to synchronize the signals of different users, which helps to achieve accurate signal identification and separation in a multi-user environment. Combining these features, this solution not only improves the reliability and accuracy of communication, but also optimizes the efficiency of data transmission, enhances the flexibility and scalability of the system, and thus provides users with a more stable and efficient communication experience.

[0103] 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:

[0104] Step S201: Acquire a first input image sent by a first user and a second input image sent by a second user.

[0105] Specifically, the "first input image" and "second input image" refer to image data generated or selected by two different users (i.e., the first user and the second user). These images are the content to be transmitted during the communication process. The specific process of this solution includes obtaining the first input image sent by the first user and the second input image sent by the second user at the user's sending end.

[0106] Step S202 : extracting semantic features from the first input image, and rearranging the extracted features using a preset first rearrangement index and then adding a first synchronization header sequence to obtain a first semantic feature.

[0107] Specifically, "semantic feature extraction" refers to the process of analyzing and extracting key visual information from the first input image. This information represents the image's content and meaning, while the "preset first reordering index" is a parameter used to guide the reordering of these extracted feature data. The "first synchronization header sequence" is a specific signal pattern used to help the receiving end identify and synchronize the signal from the first user during data transmission. The implementation of this solution involves first performing semantic feature extraction on the first input image sent by the first user. This step typically involves using advanced image processing and machine learning techniques to identify important elements in the image, such as edges, textures, and objects. The extracted semantic features are then reordered according to the pre-set first reordering index to optimize data transmission efficiency or meet specific communication protocol requirements. The reordered features are then added to the first synchronization header sequence to form a complete "first semantic feature" signal, which is then transmitted to the relay node via the wireless channel. This process not only improves data transmission efficiency and accuracy, but also enhances the system's anti-interference capabilities, as the synchronization header sequence helps distinguish signals from different users in complex communication environments, ensuring correct decoding and reconstruction of data. In this way, this scheme realizes the effective processing and transmission of image data, providing an innovative and efficient communication strategy for the two-way relay communication system.

[0108] Step S203 , extracting semantic features from the second input image, and rearranging the extracted features using a preset second rearrangement index, and then adding a second synchronization header sequence to obtain a second semantic feature.

[0109] Specifically, "semantic feature extraction" is a technical process designed to identify and extract key visual elements and content information from the second input image, which can represent the semantic meaning of the image. The "preset second reordering index" refers to a predefined parameter or mapping used to adjust the extracted semantic features in a specific order to optimize data transmission or meet specific coding requirements. The "second synchronization header sequence" is a special code sequence used for signal synchronization. It is added to the beginning of the user data so that the receiving end can identify and distinguish the signals of different users and perform correct synchronization processing.

[0110] The implementation of this solution involves first extracting semantic features from the second user's input image. This may involve using image recognition algorithms to analyze the image content and extract features such as shape, color, and texture. These features are then reordered using a second reordering index to adjust the data order, which may help improve transmission efficiency or comply with certain transmission protocols. The reordered feature data is then added to a second synchronization header sequence to form a complete "second semantic feature" signal. This signal is then transmitted via a wireless channel to a relay node. The relay node uses the synchronization header sequence to identify the signal's starting point and the reordering index to restore the data's original order, accurately decoding and reconstructing the original image. This feature not only improves the accuracy and reliability of data transmission, but also enhances signal synchronization through the use of the synchronization header sequence, enabling effective differentiation and processing of signals from different users in multi-user environments. Furthermore, by extracting and reordering semantic features, this solution can adapt to diverse communication scenarios and requirements, improving the flexibility and efficiency of the communication system.

[0111] Step S204: superimpose the first semantic feature and the second semantic feature to obtain the superimposed feature and send it to the relay node; the relay node extracts the features corresponding to the first user and the second user from the superimposed feature according to the pre-stored synchronization header sequence to obtain the first user feature and the second user feature; use the pre-stored reordering index to restore the original order of the first user feature and the second user feature respectively to obtain the sorted first user feature and the sorted second user feature; perform data recovery on the sorted first user feature and the sorted second user feature respectively, and re-encode and sort the restored data using the semantic encoder corresponding to the receiving user and the reordering index respectively, and superimpose the sorted data together and send them to the receiving end, so that the receiving user at the receiving end can restore the corresponding original image data.

[0112] Specifically, the "first semantic feature" and the "second semantic feature" refer to signals extracted from the input images of the first and second users respectively and processed after rearrangement and addition of synchronization header sequences. The "superimposed feature" refers to the merging of these two semantic feature signals into a composite signal to achieve shared transmission of multi-user signals on the same time-frequency resources. The specific implementation process includes that the user transmitting end first extracts semantic features from their respective input images, then rearranges these features according to a preset rearrangement index, and then adds their respective synchronization header sequences to form the first and second semantic features. Then, these semantic features are superimposed together to form a superimposed feature, which is sent to the relay node via a wireless channel.

[0113] After receiving the superimposed features, the relay node uses pre-stored synchronization header sequence information to calculate correlations to identify and separate the characteristic signals belonging to the first and second users, namely the first and second user features. The relay node then uses the corresponding reordering index to sort and recover the separated features, obtaining the sorted first and second user features. Next, the relay node performs data recovery on the sorted features to restore image data close to the original. The relay node then uses the semantic encoder corresponding to the receiving user and the reordering index to re-encode and reorder the recovered data to meet the requirements of downlink transmission. Finally, the relay node superimposes the sorted data and transmits it via the downlink to the receiving end. Upon receiving this data, the receiving user device uses its own semantic decoder and reordering index to further recover the corresponding original image data. This entire process not only improves spectrum utilization and communication efficiency, but also ensures accurate and reliable data transmission through precise synchronization header detection, feature separation, sorting recovery, and data recovery, thereby achieving efficient multi-user bidirectional relay communication.

[0114] In this way, by acquiring input images from the first and second users, semantic feature extraction is performed on each user's input image to identify and extract key information from the image. These features are then reordered using a preset reordering index to optimize data transmission efficiency, and a synchronization header sequence is added to assist relay nodes in signal recognition and synchronization. Next, the processed semantic features are superimposed to form a superimposed feature, which is transmitted to the relay node via a wireless channel. Upon receiving the superimposed feature, the relay node uses pre-stored synchronization header sequence information to accurately separate the feature signals belonging to the first and second users from the superimposed signal. The separated features are then sorted and restored using the corresponding reordering index to ensure the original order of the data is preserved. Data recovery is then performed on the sorted features, which involves decoding and reconstructing the original image data. Finally, the relay node uses the semantic encoder and reordering index corresponding to the receiving user to re-encode and reorder the recovered data for downlink transmission, and transmits this data to the receiver via the downlink. This combination of features improves data transmission accuracy and reliability, optimizes spectrum utilization and communication efficiency, enhances the system's anti-interference capabilities, and, through precise signal processing and synchronization mechanisms, ensures that users receive high-quality, original image data. Furthermore, this approach increases the flexibility and scalability of the communication system, enabling it to adapt to diverse communication scenarios and user needs, providing users with a more stable and efficient communication experience.

[0115] In some optional embodiments, before acquiring the first input image sent by the first user and the second input image sent by the second user, the method further includes:

[0116] Obtaining a first random number corresponding to the first user, and sorting the first random number according to a specified rule, and obtaining a first reordering index based on the position of the sorted elements;

[0117] A second random number corresponding to the second user is obtained, and the second random number is arranged according to a specified rule, and a second reordering index is generated based on the position of the arranged elements.

[0118] Specifically, "random numbers" are a set of irregular digital sequences independently generated by each user for data reordering, and the "reordering index" is the index obtained after these random numbers are processed according to specific sorting rules, which is used to guide the subsequent data reordering process. The specific process of this scheme includes first obtaining a set of random numbers for the first user. These numbers have no specific order; then, these random numbers are arranged according to the specified sorting rules (such as ascending or descending order), and the order of the arranged numbers determines the order of data reordering. Based on this sorting result, a "first reordering index" is generated, which reflects the new position of each element of the data after reordering. Similarly, another set of random numbers is obtained for the second user, and the same arrangement and index generation process is performed to obtain a "second reordering index."

[0119] This method, based on random numbers and reordering indices, allows each user's data to be reordered into a unique sequence before transmission. This facilitates data differentiation and processing in multi-user environments, enhancing the flexibility and reliability of data transmission. Furthermore, using random numbers to generate the reordering indices increases the randomness and unpredictability of the data, thereby improving communication security. When a relay node receives the reordered data, it can use the pre-shared reordering indices to restore the original data order, avoiding semantic interference between users and improving decoded image quality. This process not only optimizes the data transmission process but also provides key technical support for efficient two-way relay communication.

[0120] In this way, by obtaining random numbers corresponding to the first and second users and arranging them according to specified rules, the first and second reordering indexes are generated. This method ensures that the data of each user can be reordered in a unique order before transmission, thereby improving the distinguishability of data and the accuracy of processing in a multi-user environment. Secondly, the reordering index generated based on random numbers increases the randomness and unpredictability of the data, which helps to improve the security of communications and prevent potential data leakage or unauthorized access. In addition, the use of this index also optimizes the data transmission process, allowing the relay node to accurately restore the original order of the data based on the pre-shared reordering index, ensuring the correct decoding and processing of the data. Ultimately, the combination of these features not only improves the reliability and efficiency of communications, but also enhances the flexibility and scalability of the system, enabling it to adapt to different communication scenarios and user needs, providing users with a more stable and efficient communication experience.

[0121] In some optional embodiments, after obtaining the first reordering index and the second reordering index, the method further includes:

[0122] Acquire a first synchronization header sequence and a second synchronization header sequence;

[0123] Obtain a first encoder, a first decoder, a second encoder, and a second decoder;

[0124] The first synchronization header sequence, the second synchronization header sequence, the first encoder, the first decoder, the second encoder and the second decoder, the first reordering index and the second reordering index are uploaded to the relay node.

[0125] Specifically, the "synchronization header sequence" is a signal with a specific pattern used for signal synchronization. It helps relay nodes identify and distinguish signals from different users. The "encoder" and "decoder" are key components in communication systems. The encoder is responsible for converting raw data (such as an image) into a format suitable for transmission, while the decoder is responsible for restoring the received encoded data to its original form. The "reordering index" is a parameter used for data reordering, which ensures that data can be correctly restored even if it has been reordered during transmission.

[0126] The scheme involves first obtaining two sets of synchronization header sequences, one for the first and second users, and two sets of encoders and decoders, each dedicated to processing data from a different user. Next, these synchronization header sequences, encoders, decoders, and pre-generated first and second reordering indices are uploaded to a relay node. The relay node, acting as a communication hub, uses this information to process and forward data from different users.

[0127] In this way, by acquiring and uploading the synchronization header sequences, encoders, decoders and reordering indexes of the first and second users to the relay node, the use of the synchronization header sequence greatly improves the synchronization accuracy of the signal, enabling the relay node to accurately identify and distinguish the signals of different users, thereby reducing confusion and errors in the signal processing process. Secondly, the acquisition and uploading of the encoder and decoder provide the relay node with the necessary tools to ensure the integrity and correctness of the data during transmission, while also enabling the data to be adapted according to the specific needs of the receiving user. In addition, the uploading of the reordering index enables the relay node to correctly sort and reorganize the user data, and even if the order is disordered during the data transmission process, the original data order can be accurately restored. Combining these features, this solution not only optimizes the data transmission process and improves the reliability and efficiency of communication, but also enhances the flexibility and scalability of the system, enabling it to adapt to different communication scenarios and user needs, providing users with a more stable and efficient communication experience.

[0128] The present disclosure provides a semantic communication method, see Figure 3 , Figure 3 : is a schematic diagram of the steps of a semantic communication method in another embodiment of the present disclosure, which is applied to a receiving end and includes:

[0129] Step S301: receiving sorted data sent from a relay node, where the relay node receives superimposed features from a first user and a second user, where the superimposed features include a first semantic feature and a second semantic feature. The first semantic feature is obtained by performing semantic feature extraction on a first input image sent by the first user, re-arranging the image using a preset first re-arrangement index, and then adding a first synchronization header sequence. The second semantic feature is obtained by performing semantic feature extraction on a second input image sent by the second user, re-arranging the image using a preset second re-arrangement index, and then adding a second synchronization header sequence. Features corresponding to the first user and the second user are extracted from the superimposed features according to a pre-stored synchronization header sequence to obtain first user features and second user features. The original sorting of the first user features and the second user features is restored using a pre-stored re-arrangement index to obtain sorted first user features and sorted second user features. Data recovery is performed on the sorted first user features and the sorted second user features to obtain recovered data. The recovered data is re-encoded and re-sorted using a semantic encoder and re-arrangement index corresponding to the receiving user, and the sorted data are superimposed and sent to the receiving end.

[0130] Specifically, relay nodes play a key role in the communication system, responsible for receiving, processing, and forwarding data from different users. "Superimposed features" combine the signals from the first and second users into a composite signal for transmission over the same communication resources. "First semantic features" and "second semantic features" are extracted from the input images of the first and second users, respectively. These features contain key image information such as shape, color, and texture.

[0131] The implementation of this solution involves the first and second users first performing "semantic feature extraction" on their respective input images. This process uses image processing and machine learning techniques to identify and extract important elements from the image. The extracted semantic features are then reordered according to preset "first reordering indices" and "second reordering indices." These indices are used to optimize data transmission or meet specific encoding requirements. The reordered features are then added to corresponding "synchronization header sequences," which help relay nodes identify and synchronize signals from different users.

[0132] After receiving the superimposed features, the relay node uses pre-stored synchronization header sequence information to identify and separate the characteristic signals belonging to the first and second users, namely the first and second user features. The relay node then uses the corresponding reordering index to sort and restore these features to avoid semantic interference between users. Next, the relay node performs "data recovery" on the sorted features, which may include decoding and denoising operations to restore close-to-original image data.

[0133] Finally, the relay node re-encodes and reorders the recovered data using the semantic encoder and reordering index corresponding to the receiving user to meet the requirements of downlink transmission. This re-encoded and reordered data, known as the "sorted data," is then superimposed and transmitted to the receiving end via the downlink. Upon receiving this data, the receiving user device uses its own semantic decoder and reordering index to further recover the corresponding original image data. This entire process not only improves the accuracy and reliability of data transmission, but also optimizes the efficiency and flexibility of the communication system, providing users with a more stable and efficient communication experience.

[0134] Step S302 : Using the first reordering index, obtain data corresponding to the second user from the sorted data to obtain the second user data.

[0135] Specifically, the "first reordering index" is a predefined set of parameters that guides how to accurately extract and identify a specific user's data from the sorted data. This index reflects the new position of each element in the data during the reordering process and is key to ensuring that data is correctly attributed to each user. "Second user data" refers to the original information or content belonging to the second user, which may have undergone encoding, reordering, and other processing steps during the communication process.

[0136] The specific implementation process of this solution includes: After the relay node completes processing of the superimposed features from the first and second users, including removing the synchronization header sequence, separating the features, reordering, and restoring the data, it will obtain sorted data. This data may contain information from multiple users and needs to be further separated to ensure that each user only receives the data that belongs to them.

[0137] Using the first reordering index, the relay node can accurately identify and extract the data belonging to the second user from the sorted data. This is accomplished by searching the reordering index for entries corresponding to the second user. These entries indicate the specific location of the second user's data within the sorted data. In this way, the relay node can ensure the correct attribution of data and avoid confusion or mistransmission of user data.

[0138] Furthermore, this process also involves data security and privacy protection, as correct data extraction and attribution confirmation can prevent unauthorized access or data leakage. By using reordered indexes, this solution not only improves the accuracy and efficiency of data transmission, but also enhances the security and reliability of the communication system, providing users with a more secure and efficient communication environment.

[0139] Step S303: Using the second rearrangement index, obtain data corresponding to the first user from the sorted data to obtain the first user data.

[0140] Specifically, the "second reordering index" is a parameter generated and provided by the first user. It guides the receiving end or relay node on how to accurately extract the first user's data from the sorted data. This index is based on the specific reordering rules that the first user applies to its data before sending it. This ensures that the original data order can be accurately restored even if the order changes during data transmission.

[0141] The specific implementation process of this solution includes the following: After receiving superimposed features from multiple users, the relay node uses a pre-stored synchronization header sequence to separate the features corresponding to the first and second users from these superimposed features. The relay node then uses the first and second reordering indexes to sort and restore these separated features, ensuring that the original order of the data is preserved. After the sorting is restored, the relay node uses the "second reordering index" to extract the data corresponding to the first user from the sorted data. This step is accomplished by searching the index for the entry corresponding to the first user, which indicates the specific location of the first user's data in the sorted data.

[0142] In this way, relay nodes can ensure the correct attribution of data, avoiding confusion or mis-transmission of user data. This approach not only improves the accuracy and efficiency of data transmission, but also enhances the system's flexibility and scalability, enabling it to adapt to diverse communication scenarios and user needs. Furthermore, the use of reordered indexes helps protect data security and privacy, preventing unauthorized access or data leakage, thereby providing users with a more secure and reliable communication experience.

[0143] Step S304: Using the first decoder to recover the second user data, to obtain the original image sent by the second user.

[0144] Specifically, "first decoder" refers to a decoding device or algorithm designed specifically for the first user, used to restore encoded data to its original form. "Second user data" refers to the original information or content from the second user, which may have undergone encoding, reordering, or other processing steps during transmission. "Data recovery" refers to the process of restoring encoded or otherwise processed data to its original form.

[0145] The specific implementation process of this solution includes the following: After the relay node uses the second reordering index to extract the data corresponding to the first user from the sorted data, it treats this data as the second user's data. To restore this data to a form understandable to the first user, the relay node uses the first decoder to decode the data and obtain the original image sent by the second user. This process ensures that even in a multi-user environment, each user receives the correct data that belongs to them. This design not only improves the accuracy and reliability of data transmission but also optimizes the efficiency of the communication system. In addition, this approach enhances the flexibility and scalability of the system, enabling it to adapt to different communication scenarios and user needs, providing users with a more stable and efficient communication experience.

[0146] Step S305: Using the second decoder to recover the first user data, to obtain the original image sent by the first user.

[0147] Specifically, "first user data" refers to the original information generated and sent by the first user, which may have undergone encoding, rearrangement, and the addition of synchronization header sequences to adapt to transmission requirements. A "second decoder" is a decoding tool or algorithm designed specifically for the second user, used to restore the received encoded data to its original, understandable format. "Data recovery" refers to the process of converting encoded or encrypted data back to its original state during the decoding process.

[0148] The specific implementation process of this solution includes: the relay node first uses the first reordering index to extract the data belonging to the first user from the sorted data. This data may have been encoded and reordered, and therefore needs to be restored through decoding. The relay node then uses a second decoder to decode the first user data. This decoder corresponds to the encoder of the second user and can correctly restore the data, thereby obtaining the original image sent by the first user. This process ensures that the data can be correctly transmitted and received between users, even if the data may have undergone complex processing during transmission. In this way, this solution not only improves the accuracy and reliability of data transmission, but also enhances the flexibility and efficiency of the communication system. In addition, this method also helps to protect the security and privacy of the data, because only the corresponding decoder can correctly decode the data, thereby preventing unauthorized access and data leakage. Ultimately, this provides users with a safe, efficient and reliable communication experience.

[0149] In this way, a relay node receives, processes, and forwards the superimposed features from the first and second users, achieving an efficient two-way communication mechanism. The superimposed features consist of first and second semantic features that have undergone semantic feature extraction, reordering, and the addition of a synchronization header sequence. The relay node uses a pre-stored synchronization header sequence and reordering index to accurately extract and recover the features of the first and second users from the superimposed features, then performs data recovery and re-encoding. This process not only ensures data integrity and correctness but also optimizes the data transmission format and improves transmission efficiency by using the semantic encoder and reordering index corresponding to the receiving user. Furthermore, the relay node can retrieve the data corresponding to the second and first users from the sorted data based on the first and second reordering indices, and then use the other party's decoder to recover the data to obtain the original image sent by the other party. This approach not only improves the accuracy and reliability of data transmission, but also enhances the system's flexibility and adaptability, enabling the system to adapt to different user needs and communication scenarios. Ultimately, this design provides users with a more stable, secure, and efficient communication experience, significantly improving the overall performance of the communication system.

[0150] The above embodiments respectively describe in detail the processing processes at the transmitting end and the receiving end. In order to facilitate the overall understanding of the technical solution of the present application, the embodiments disclosed herein describe the communication process of the solution as a whole. The method includes: the transmitting end obtains a first input image sent by a first user and a second input image sent by a second user; extracts semantic features from the first input image, and adds a first synchronization header sequence after rearranging it using a preset first rearrangement index to obtain a first semantic feature; extracts semantic features from the second input image, and adds a second synchronization header sequence after rearranging it using a preset second rearrangement index to obtain a second semantic feature; superimposes the first semantic feature and the second semantic feature to obtain a superimposed feature and sends it to the relay node;

[0151] The relay node receives superimposed features from the first user and the second user, extracts features corresponding to the first user and the second user from the superimposed features according to a pre-stored synchronization header sequence, and obtains first user features and second user features; uses a pre-stored reordering index to restore the original order of the first user features and the second user features, respectively, to obtain sorted first user features and sorted second user features; performs data recovery on the sorted first user features and the sorted second user features, and re-encodes and re-sorts the recovered data using a semantic encoder corresponding to the receiving user and the reordering index, and then superimposes the sorted data and sends it to the receiving end;

[0152] The receiving end receives the sorted data sent from the relay node, uses the first reordering index to obtain the data corresponding to the second user from the sorted data, and obtains the second user data; uses the second reordering index to obtain the data corresponding to the first user from the sorted data, and obtains the first user data; uses the first decoder to recover the second user data to obtain the original image sent by the second user; and uses the second decoder to recover the first user data to obtain the original image sent by the first user.

[0153] See also Figure 4 , Figure 4 The figure is a schematic diagram of the overall process of the semantic communication method in one embodiment of the present disclosure. Specifically, the process includes: first, before the start of communication, two users (i.e., the first user and the second user) each generate a string of random numbers with a length of L = 1179648, sort the random numbers from large to small, and the resulting sequence numbers are the indexes index1 (also called the first reordering index) and index2 (also called the second reordering index) of their respective semantic information.

[0154] Before the communication begins, the first user and the second user upload their own knowledge base information to the relay node, wherein the first user's knowledge base includes a first semantic encoder, a first semantic decoder, a first reordering index and a first synchronization header sequence; the second user's knowledge base includes a second semantic encoder, a second semantic decoder, a second reordering index and a second synchronization header sequence.

[0155] When two users input a first input image x1 and a second input image x2 of size 3×512×768 respectively, the first user uses the first semantic encoder f e1 Extract the first semantic feature The second user passes the second semantic encoder f e2 Extracting the second semantic feature To reduce data transmission resource consumption, the system compresses semantic features by controlling the compression ratio. In this embodiment, the compression ratio is set to 0.8, which means that the last 20% of the extracted one-dimensional semantic features is reset to zero, thereby reducing the data volume while retaining the main semantic information.

[0156] Next, the first user rearranges the semantic features according to the first rearrangement index (i.e., index1) to obtain The second user rearranges the semantic features according to the second rearrangement index (index2) to obtain Then each user generates its own ZC sequence as a synchronization header sequence with a sequence length of 71. The first user embeds the first synchronization header sequence s1 into Before The second user embeds the second synchronization header sequence s2 into Before The first and second semantic features, which have been rearranged and embedded with synchronization headers, are superimposed via physical layer network coding. The superimposed signal is then transmitted to the relay using the same time-frequency resources. This system supports asynchronous signal superposition, despite the possibility of sign bit shift during transmission.

[0157] The relay node detects the first synchronization header sequence and the second synchronization header sequence of each of the two users, locates the starting position of their semantic information, and extracts the semantic information of length L, then removes the first synchronization header sequence and the second synchronization header sequence to obtain and Then, the relay node restores the original order of the information according to the first reordering index and the second reordering index, and obtains and Based on the OMDMA principle, semantic information generated by different semantic models is rearranged using specific indexes, making it unreadable by other semantic models. Therefore, for one user, the out-of-order information of the other user is treated as Gaussian noise and effectively filtered out during the decoding process, achieving precise separation of the two users' information.

[0158] The relay node uses the preset first semantic decoder and the second semantic decoder to decode the semantic information of the two users respectively and restore the sent image data. and

[0159] The relay node uses the second semantic encoder f e2 Recovering semantic information Extracting semantic features Use the first semantic encoder f e1 Recovering semantic information Extracting semantic features After reordering the semantic features according to the reordering index corresponding to the semantic encoder, the relay node transmits the aliased signal to the downlink. The first user recovers the received signal according to the first reordering index and then uses its own semantic decoder f d1 Decode the received semantic information to obtain image data The second user recovers the received signal according to the second reordering index and then uses its own second semantic decoder f d2 Decode the received semantic information to obtain image data

[0160] Regarding the re-indexing of this solution, it should be noted that Figure 5 , Figure 5This is a workflow for data reordering and recovery using a semantic encoder and decoder and a reordering index in one embodiment of the present disclosure. The process involves: a first user and a second user each generating a set of random numbers. For example, the first user generates a string of random numbers, and the second user generates a string of random numbers. They then sort these random numbers to generate a reordering index, resulting in the first reordering index being [4, 1, 3, 2] and the second reordering index being [1, 4, 3, 2].

[0161] Suppose the first and second users use the first and second semantic encoders, respectively, to extract semantic features from their input images. Assume the extracted features are initially ordered [1, 2, 3, 4]. They then reorder these features according to their respective reordering indices. The first user's features are reordered to [4, 1, 3, 2], and the second user's features are reordered to [1, 4, 3, 2].

[0162] The first user and the second user add their respective synchronization headers (s1 and s2) to the front of the rearranged features, and then send these signals to the relay node through a wireless channel. At the relay node, the two signals are superimposed. After receiving the superimposed signal, the relay node first removes the synchronization headers s1 and s2, and then extracts the feature signals belonging to the first user and the second user based on the position information of the synchronization headers. When the relay node uses the rearrangement indexes of the first user and the second user to restore the extracted features, only the first rearrangement index can restore the correct original order [1,2,3,4], while using the second rearrangement index will result in noise. It can be seen that this scheme can achieve efficient data transmission and accurate data recovery even in a multi-user environment, ensuring the reliability and accuracy of communication.

[0163] See also Figure 6 , Figure 6 This is a flowchart of the working process of a two-way relay communication system in one embodiment of the present disclosure. This diagram illustrates the working principle of a two-way relay communication system, which involves two users (a first user and a second user) and a relay node. The following is a detailed working principle:

[0164] The first and second users each use their own semantic encoders (first encoder and second encoder) to extract semantic features from their input images (first input image and second input image). The extracted semantic feature signals are transmitted to a relay node via a wireless channel. At the relay node, these two signals may be affected by channel effects such as noise and fading. The figure shows the process of superimposing these two signals, where the superimposed signal contains the semantic features of both users.

[0165] Next, the relay node in this solution uses a specific algorithm or technology (such as Orthogonal Mode Division Multiple Access (OMDMA)) to separate the superimposed signals into independent signals belonging to the first and second users. This step ensures that the data of each user can be correctly identified and processed. The separated signals are fed into the corresponding semantic decoders (first semantic decoder and second semantic decoder) for decoding to restore the original image data.

[0166] In this way, the figure demonstrates an efficient two-way relay communication system that is able to process signals from multiple users, ensuring the correct transmission and recovery of data, and maintaining communication reliability and accuracy even under complex wireless channel conditions.

[0167] See also Figure 7 , Figure 7 This is a block diagram of the semantic encoder in one embodiment of the present disclosure. The implementation structure of the semantic encoder in this solution is: a neural network architecture consisting of an attention feature module (AF module) and a basic feature processing module (Basic block module); the AF module realizes dynamic perception and resource allocation of wireless channel conditions through the channel soft attention mechanism, and the Basic block module realizes feature dimension compression and multi-level feature extraction through multi-layer convolution. The two work in cascade to form a distributed representation of semantic features, improve the distinguishability of the feature space and reduce transmission redundancy.

[0168] Specifically, the basic feature processing module (Basic Block) is the basic building block in the network, which contains a series of convolutional layers (Conv), normalization layers (such as GDN, Group Normalization) and activation functions (such as Prelu).

[0169] The AF module (short for Attention Feature Module) is an attention feature module used to enhance the network's ability to capture important features in an image. This module helps the network focus on the most relevant parts of the image, thereby improving feature extraction.

[0170] Conv 3*3: This represents a 3x3 convolutional layer, a common layer type in convolutional neural networks used to extract local features. Convolutional layers apply filters to the input data using a sliding window to identify patterns in the image.

[0171] GDN: Group Normalization is a normalization technique that can accelerate training and improve model stability.

[0172] Prelu: This is a parameterized ReLU (Rectified Linear Unit) activation function, which allows non-zero mapping of negative values, thereby retaining some gradient information and helping to solve the "dead ReLU" problem.

[0173] Trans Conv: refers to the transposed convolution layer, which is used for upsampling operations, that is, increasing the spatial dimension of the data. It is usually used in the decoder to reconstruct the image.

[0174] Addition operation: The plus sign (⊕) in the figure represents a skip connection, a feature of the ResNet. Skip connections add the input directly to the output of the subsequent layer, helping to alleviate the vanishing gradient problem in deep networks and promoting the flow of information.

[0175] Prelu (Sigmoid): In the last layer of the network, the Prelu activation function with Sigmoid is used. This may be used in the output layer to generate output values ​​between 0 and 1, which is suitable for tasks that require probabilistic output, such as image segmentation.

[0176] The entire network structure, through the combination of these modules and layers, achieves deep feature extraction and processing of the input image. In the encoder, these features can then be used for compression or encoding; in the decoder, these features are used to reconstruct or decode the image. The introduction of the attention module further improves the model's ability to identify important features, thereby improving the efficiency and accuracy of encoding and decoding.

[0177] 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.

[0178] The present disclosure also provides a semantic communication device 800, such as Figure 8 As shown, it is applied to relay nodes and includes:

[0179] A first receiving module 801 is configured to receive superimposed features from a first user and a second user, the superimposed features including a first semantic feature and a second semantic feature, the first semantic feature being obtained by performing semantic feature extraction on a first input image sent by the first user, rearranging the image using a preset first rearrangement index, and then adding a first synchronization header sequence; and the second semantic feature being obtained by performing semantic feature extraction on a second input image sent by the second user, rearranging the image using a preset second rearrangement index, and then adding a second synchronization header sequence.

[0180] An extraction module 802 is configured to extract features corresponding to the first user and the second user from the superimposed features according to a pre-stored synchronization header sequence, thereby obtaining first user features and second user features;

[0181] A rearrangement module 803 is configured to restore the original order of the first user feature and the second user feature using a pre-stored rearrangement index to obtain the sorted first user feature and the sorted second user feature;

[0182] The recovery module 804 is used to recover the data of the sorted first user features and the sorted second user features respectively, and re-encode and sort the recovered data using the semantic encoder and reordering index corresponding to the receiving user respectively, and then superimpose the sorted data and send them to the receiving end, so that the receiving user at the receiving end can recover the corresponding original image data.

[0183] In some optional embodiments, the extraction module 802 extracts features corresponding to the first user and the second user from the superimposed features according to the pre-stored synchronization header sequence, to obtain the first user features and the second user features, including:

[0184] Acquire a pre-stored first synchronization header sequence and a pre-stored second synchronization header sequence;

[0185] Identify the position of the first synchronization header sequence based on the pre-stored correlation between the first synchronization header sequence and the superposition feature, and extract a feature of a preset length from the superposition feature using the position of the first synchronization header sequence as a starting position to obtain a first user feature;

[0186] According to the pre-stored correlation between the second synchronization header sequence and the superposition feature, the position of the second synchronization header sequence is identified, and the position of the second synchronization header sequence is used as the starting position to extract a feature of a preset length from the superposition feature to obtain the second user feature.

[0187] In some optional embodiments, after obtaining the first user feature and the second user feature, the extraction module 802 is further configured to:

[0188] The first synchronization header sequence and the second synchronization header sequence are removed.

[0189] In some optional embodiments, the rearrangement module 803 uses a pre-stored rearrangement index to restore the original order of the first user feature and the second user feature, respectively, to obtain the sorted first user feature and the sorted second user feature, including:

[0190] Obtaining a pre-stored first re-arrangement index and a pre-stored second re-arrangement index;

[0191] Reordering the first user features using a pre-stored first reordering index to restore the original order of the first user's corresponding features, thereby obtaining ordered first user features;

[0192] The second user features are reordered using the pre-stored second reordering index to restore the original order of the features corresponding to the second user, thereby obtaining the ordered second user features.

[0193] In some optional embodiments, the recovery module 804 recovers the sorted first user features and the sorted second user features, respectively, re-encodes and re-sorts the recovered data using a semantic encoder and a re-ranking index corresponding to the receiving user, and superimposes the sorted data and sends them to the receiving end, including:

[0194] Decoding the sorted first user features and the sorted second user features respectively to obtain decoded first user features and decoded second user features;

[0195] encoding the decoded first user feature using a preset second encoder, and rearranging the decoded first user feature using a second rearrangement index to obtain a rearranged first user feature;

[0196] encoding the decoded second user feature using a preset first encoder, and rearranging the decoded second user feature using a first rearrangement index to obtain a rearranged second user feature;

[0197] The rearranged first user feature and the rearranged second user feature are superimposed and then sent to a receiving end via a downlink.

[0198] In some optional embodiments, before receiving the superimposed features from the first user and the second user, the first receiving module 801 is further configured to:

[0199] receiving a first encoder, a first decoder, a first reordering index, and a first synchronization header sequence sent by a first user;

[0200] A second encoder, a second decoder, a second reordering index, and a second synchronization header sequence sent by a second user are received.

[0201] The present disclosure also provides a semantic communication device 900, such as Figure 9 As shown, it is applied to the sending end and includes:

[0202] An acquisition module 901 is configured to acquire a first input image sent by a first user and a second input image sent by a second user;

[0203] A first processing module 902 is configured to extract semantic features from a first input image, rearrange the extracted features using a preset first rearrangement index, and then add a first synchronization header sequence to obtain a first semantic feature;

[0204] The second processing module 903 is configured to extract semantic features from the second input image, and to rearrange the extracted features using a preset second rearrangement index and then add a second synchronization header sequence to obtain a second semantic feature;

[0205] The sending module 904 is used to superimpose the first semantic feature and the second semantic feature to obtain the superimposed feature and then send it to the relay node, and the relay node extracts the features corresponding to the first user and the second user from the superimposed feature according to the pre-stored synchronization header sequence to obtain the first user feature and the second user feature; uses the pre-stored reordering index to restore the original order of the first user feature and the second user feature respectively to obtain the sorted first user feature and the sorted second user feature; performs data recovery on the sorted first user feature and the sorted second user feature respectively, and re-encodes and sorts the recovered data using the semantic encoder corresponding to the receiving user and the reordering index respectively, and superimposes the sorted data together and sends it to the receiving end, so that the receiving user at the receiving end can restore the corresponding original image data.

[0206] In some optional embodiments, before acquiring the first input image sent by the first user and the second input image sent by the second user, the acquisition module 901 is further configured to:

[0207] Obtaining a first random number corresponding to the first user, and sorting the first random number according to a specified rule, and obtaining a first reordering index based on the position of the sorted elements;

[0208] A second random number corresponding to the second user is obtained, and the second random number is arranged according to a specified rule, and a second reordering index is generated based on the position of the arranged elements.

[0209] In some optional embodiments, after obtaining the first reordering index and the second reordering index, the obtaining module 901 is further configured to:

[0210] Acquire a first synchronization header sequence and a second synchronization header sequence;

[0211] Obtain a first encoder, a first decoder, a second encoder, and a second decoder;

[0212] The first synchronization header sequence, the second synchronization header sequence, the first encoder, the first decoder, the second encoder and the second decoder, the first reordering index and the second reordering index are uploaded to the relay node.

[0213] The present disclosure also provides a semantic communication device 1000, such as Figure 10 As shown, it is applied to the receiving end and includes:

[0214] The second receiving module 1001 is configured to receive sorted data sent from a relay node, where the relay node receives superimposed features from a first user and a second user, the superimposed features including a first semantic feature and a second semantic feature. The first semantic feature is obtained by performing semantic feature extraction on a first input image sent by the first user, re-arranging the image using a preset first re-arrangement index, and then adding a first synchronization header sequence. The second semantic feature is obtained by performing semantic feature extraction on a second input image sent by the second user, re-arranging the image using a preset second re-arrangement index, and then adding a second synchronization header sequence. Features corresponding to the first user and the second user are extracted from the superimposed features according to the pre-stored synchronization header sequence to obtain first user features and second user features. The pre-stored re-arrangement index is used to restore the original sorting of the first user features and the second user features to obtain sorted first user features and sorted second user features. The sorted first user features and the sorted second user features are respectively restored to obtain recovered data, and the recovered data are re-encoded and re-sorted using the semantic encoder and re-arrangement index corresponding to the receiving user, and the sorted data are superimposed and sent to the receiving end.

[0215] A first separation module 1002 is configured to obtain data corresponding to a second user from the sorted data using a first reordering index to obtain second user data;

[0216] A second separation module 1003 is configured to obtain data corresponding to the first user from the sorted data using a second reordering index to obtain first user data;

[0217] A first decoding module 1004 is configured to recover the second user data using the first decoder to obtain the original image sent by the second user;

[0218] The second decoding module 1005 is configured to recover the first user data using a second decoder to obtain the original image sent by the first user.

[0219] 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.

[0220] 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.

[0221] Figure 11A schematic block diagram of an example electronic device 1100 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.

[0222] like Figure 11 As shown, the electronic device 1100 includes a computing unit 1101, which can perform various appropriate actions and processes according to a computer program stored in a read-only memory (ROM) 1102 or a computer program loaded from a storage unit 1108 into a random access memory (RAM) 1103. Various programs and data required for the operation of the device 1100 can also be stored in the RAM 1103. The computing unit 1101, the ROM 1102, and the RAM 1103 are connected to each other via a bus 1104. An input / output (I / O) interface 1105 is also connected to the bus 1104.

[0223] Various components in device 1100 are connected to I / O interface 1105, including an input unit 1106, such as a keyboard and mouse; an output unit 1107, such as various types of displays and speakers; a storage unit 1108, such as a magnetic disk and optical disk; and a communication unit 1109, such as a network card, a modem, a wireless communication transceiver, etc. Communication unit 1109 allows device 1100 to exchange information / data with other devices via a computer network such as the Internet and / or various telecommunication networks.

[0224] The computing unit 1101 can be a variety of general and / or special processing components with processing and computing capabilities. Some examples of the computing unit 1101 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 1101 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 1108. In some embodiments, part or all of the computer program can be loaded and / or installed on the device 1100 via the ROM 1102 and / or the communication unit 1109. When the computer program is loaded into the RAM 1103 and executed by the computing unit 1101, one or more steps of the applet distribution described above can be performed. Alternatively, in other embodiments, the computing unit 1101 can be configured to perform the semantic communication method by any other appropriate means (e.g., by means of firmware).

[0225] 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.

[0226] 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.

[0227] 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.

[0228] 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).

[0229] 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.

[0230] 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.

[0231] 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.

[0232] 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 relay node, wherein: The method comprises: receiving superimposed features from a first user and a second user, the superimposed features including a first semantic feature and a second semantic feature, the first semantic feature being obtained by performing semantic feature extraction on a first input image sent by the first user, reordering the image using a preset first reordering index, and then adding a first synchronization header sequence, and the second semantic feature being obtained by performing semantic feature extraction on a second input image sent by the second user, reordering the image using a preset second reordering index, and then adding a second synchronization header sequence; Extracting features corresponding to the first user and the second user from the superimposed features according to a pre-stored synchronization header sequence to obtain first user features and second user features; Restore the original order of the first user features and the second user features respectively using a pre-stored rearrangement index to obtain a sorted first user feature and a sorted second user feature; Data recovery is performed on the sorted first user features and the sorted second user features respectively, and the recovered data are re-encoded and sorted using a semantic encoder and a reordering index corresponding to the receiving user respectively, and the sorted data are superimposed together and sent to the receiving end, so that the receiving user at the receiving end can recover the corresponding original image data.

2. The method according to claim 1, wherein The extracting the features corresponding to the first user and the second user from the superimposed features according to the pre-stored synchronization header sequence to obtain the first user features and the second user features includes: Acquire a pre-stored first synchronization header sequence and a pre-stored second synchronization header sequence; identifying a position of the first synchronization header sequence based on a correlation between the pre-stored first synchronization header sequence and the superimposed feature, and extracting a feature of a preset length from the superimposed feature using the position of the first synchronization header sequence as a starting position to obtain a first user feature; According to the correlation between the pre-stored second synchronization header sequence and the superposition feature, the position of the second synchronization header sequence is identified, and the position of the second synchronization header sequence is used as the starting position to extract a feature of a preset length from the superposition feature to obtain a second user feature.

3. The method according to claim 2, wherein: After obtaining the first user feature and the second user feature, the method further includes: The first synchronization header sequence and the second synchronization header sequence are removed.

4. The method according to claim 1, wherein The using of the pre-stored rearrangement index to restore the original order of the first user features and the second user features to obtain the sorted first user features and the sorted second user features includes: Obtaining a pre-stored first re-arrangement index and a pre-stored second re-arrangement index; Reordering the first user features using the pre-stored first reordering index to restore the original order of the features corresponding to the first user, thereby obtaining ordered first user features; The second user features are reordered using the pre-stored second reordering index to restore the original order of the features corresponding to the second user, thereby obtaining ordered second user features.

5. The method according to any one of claims 1 to 4, wherein: The method of respectively recovering the sorted first user features and the sorted second user features, re-encoding and sorting the recovered data using a semantic encoder and a reordering index corresponding to the receiving user, and superimposing the sorted data and sending them to the receiving end includes: Decoding the sorted first user features and the sorted second user features respectively to obtain decoded first user features and decoded second user features; encoding the decoded first user feature using a preset second encoder, and rearranging the decoded first user feature using the second rearrangement index to obtain a rearranged first user feature; encoding the decoded second user feature using a preset first encoder, and rearranging the decoded second user feature using the first rearrangement index to obtain a rearranged second user feature; The rearranged first user feature and the rearranged second user feature are superimposed and then sent to a receiving end via a downlink.

6. The method according to any one of claims 1 to 4, wherein: Before receiving the overlay features from the first user and the second user, the method further includes: receiving a first encoder, a first decoder, a first reordering index, and a first synchronization header sequence sent by the first user; Receive a second encoder, a second decoder, a second reordering index, and a second synchronization header sequence sent by the second user.

7. A semantic communication method, applied to a sending end, wherein: The method comprises: Acquire a first input image sent by a first user and a second input image sent by a second user; Extracting semantic features from the first input image, and rearranging the extracted features using a preset first rearrangement index and then adding a first synchronization header sequence to obtain a first semantic feature; Extracting semantic features from the second input image, and rearranging the extracted features using a preset second rearrangement index and then adding a second synchronization header sequence to obtain a second semantic feature; The first semantic feature and the second semantic feature are superimposed to obtain a superimposed feature, which is then sent to a relay node. The relay node extracts features corresponding to the first user and the second user from the superimposed feature according to a pre-stored synchronization header sequence to obtain a first user feature and a second user feature. The pre-stored reordering index is used to restore the original order of the first user feature and the second user feature to obtain a sorted first user feature and a sorted second user feature. Data of the sorted first user feature and the sorted second user feature are restored, and the restored data are re-encoded and sorted using a semantic encoder and a reordering index corresponding to the receiving user. The sorted data are superimposed and sent to a receiving end, so that the receiving user at the receiving end can restore the corresponding original image data.

8. The method according to claim 7, wherein: Before acquiring the first input image sent by the first user and the second input image sent by the second user, the method further includes: Obtaining a first random number corresponding to the first user, and sorting the first random number according to a specified rule, and obtaining a first reordering index based on the position of the sorted elements; A second random number corresponding to the second user is obtained, and the second random number is arranged according to a specified rule, and a second reordering index is generated based on the position of the arranged elements.

9. The method according to claim 8, wherein After obtaining the first reordering index and the second reordering index, the method further includes: Acquire a first synchronization header sequence and a second synchronization header sequence; Obtain a first encoder, a first decoder, a second encoder, and a second decoder; The first synchronization header sequence, the second synchronization header sequence, the first encoder, the first decoder, the second encoder and the second decoder, the first reordering index, and the second reordering index are uploaded to the relay node.

10. A semantic communication method, applied to a receiving end, wherein: The method comprises: Receive sorted data sent from a relay node, the relay node receiving superimposed features from a first user and a second user, the superimposed features including a first semantic feature and a second semantic feature, the first semantic feature being obtained by performing semantic feature extraction on a first input image sent by the first user, re-arranging the image using a preset first re-arrangement index, and then adding a first synchronization header sequence, and the second semantic feature being obtained by performing semantic feature extraction on a second input image sent by the second user, re-arranging the image using a preset second re-arrangement index, and then adding a second synchronization header sequence; extract features corresponding to the first user and the second user from the superimposed features according to a pre-stored synchronization header sequence to obtain first user features and second user features; restore the original sorting of the first user features and the second user features using a pre-stored re-arrangement index to obtain sorted first user features and sorted second user features; perform data recovery on the sorted first user features and the sorted second user features to obtain recovered data, re-encode and re-sort the recovered data using a semantic encoder and re-arrangement index corresponding to the receiving user, and superimpose the sorted data and send them to a receiving end; Using the first rearrangement index to obtain data corresponding to the second user from the sorted data to obtain second user data; Using the second rearrangement index to obtain data corresponding to the first user from the sorted data to obtain first user data; Performing data recovery on the second user data using the first decoder to obtain an original image sent by the second user; The first user data is recovered using a second decoder to obtain an original image sent by the first user.

11. A semantic communication method, wherein: The method comprises: The transmitting end obtains a first input image sent by a first user and a second input image sent by a second user; performs semantic feature extraction on the first input image, rearranges it using a preset first rearrangement index, and then adds a first synchronization header sequence to obtain a first semantic feature; performs semantic feature extraction on the second input image, rearranges it using a preset second rearrangement index, and then adds a second synchronization header sequence to obtain a second semantic feature; superimposes the first semantic feature and the second semantic feature to obtain a superimposed feature, and then sends the superimposed feature to a relay node; The relay node receives superimposed features from a first user and a second user, extracts features corresponding to the first user and the second user from the superimposed features according to a pre-stored synchronization header sequence, and obtains first user features and second user features; uses a pre-stored reordering index to restore the original order of the first user features and the second user features, respectively, to obtain sorted first user features and sorted second user features; performs data recovery on the sorted first user features and the sorted second user features, and re-encodes and re-sorts the recovered data using a semantic encoder corresponding to the receiving user and the reordering index, and then superimposes the sorted data and sends it to a receiving end; The receiving end receives the sorted data sent from the relay node, uses the first reordering index to obtain data corresponding to the second user from the sorted data, and obtains second user data; uses the second reordering index to obtain data corresponding to the first user from the sorted data, and obtains first user data; uses the first decoder to recover the second user data to obtain the original image sent by the second user; and uses the second decoder to recover the first user data to obtain the original image sent by the first user.

12. A semantic communication device, applied to a relay node, wherein: include: a first receiving module, configured to receive superimposed features from a first user and a second user, the superimposed features including a first semantic feature and a second semantic feature, the first semantic feature being obtained by performing semantic feature extraction on a first input image sent by the first user, reordering the image using a preset first reordering index, and then adding a first synchronization header sequence; and the second semantic feature being obtained by performing semantic feature extraction on a second input image sent by the second user, reordering the image using a preset second reordering index, and then adding a second synchronization header sequence; an extraction module, configured to extract features corresponding to the first user and the second user from the superimposed features according to a pre-stored synchronization header sequence, to obtain first user features and second user features; a rearrangement module, configured to restore the original order of the first user features and the second user features respectively using a pre-stored rearrangement index to obtain the sorted first user features and the sorted second user features; The recovery module is used to respectively recover the data of the sorted first user features and the sorted second user features, and re-encode and sort the recovered data using the semantic encoder and reordering index corresponding to the receiving user, and then superimpose the sorted data and send them to the receiving end, so that the receiving user at the receiving end can recover the corresponding original image data.

13. A semantic communication device, applied to a sending end, wherein: include: An acquisition module, configured to acquire a first input image sent by a first user and a second input image sent by a second user; A first processing module is configured to extract semantic features from the first input image, and to rearrange the extracted features using a preset first rearrangement index and then add a first synchronization header sequence to obtain a first semantic feature; A second processing module is configured to extract semantic features from the second input image, and to rearrange the extracted features using a preset second rearrangement index and then add a second synchronization header sequence to obtain a second semantic feature; A sending module is used to superimpose the first semantic feature and the second semantic feature to obtain a superimposed feature and then send it to a relay node, and the relay node extracts the features corresponding to the first user and the second user from the superimposed feature according to a pre-stored synchronization header sequence to obtain the first user feature and the second user feature; uses a pre-stored reordering index to restore the original order of the first user feature and the second user feature to obtain the sorted first user feature and the sorted second user feature; performs data recovery on the sorted first user feature and the sorted second user feature, and re-encodes and sorts the recovered data using the semantic encoder corresponding to the receiving user and the reordering index, and superimposes the sorted data and sends them to the receiving end, so that the receiving user at the receiving end can restore the corresponding original image data.

14. A semantic communication device, applied to a receiving end, wherein: include: a second receiving module, configured to receive sorted data sent from a relay node, the relay node receiving superimposed features from a first user and a second user, the superimposed features including a first semantic feature and a second semantic feature, the first semantic feature being obtained by performing semantic feature extraction on a first input image sent by the first user, re-arranging the image using a preset first re-arrangement index, and then adding a first synchronization header sequence; and the second semantic feature being obtained by performing semantic feature extraction on a second input image sent by the second user, re-arranging the image using a preset second re-arrangement index, and then adding a second synchronization header sequence; extracting features corresponding to the first user and the second user from the superimposed features according to a pre-stored synchronization header sequence to obtain first user features and second user features; restoring the original sorting of the first user features and the second user features using a pre-stored re-arrangement index to obtain sorted first user features and sorted second user features; performing data recovery on the sorted first user features and the sorted second user features to obtain recovered data, re-encoding and re-sorting the recovered data using a semantic encoder and re-arrangement index corresponding to the receiving user, and superimposing the sorted data and sending them to a receiving end; a first separation module, configured to obtain data corresponding to a second user from the sorted data using the first rearrangement index to obtain second user data; a second separation module, configured to obtain data corresponding to the first user from the sorted data using the second rearrangement index to obtain first user data; a first decoding module, configured to recover the second user data using a first decoder to obtain an original image sent by the second user; The second decoding module is configured to perform data recovery on the first user data using a second decoder to obtain an original image sent by the first user.

15. 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 11.

16. 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 1-11.

17. 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 11.

Images