Semantic communication method and apparatus, electronic device, and storage medium

Abstract

This disclosure provides a semantic communication method, apparatus, electronic device, and storage medium, relating to the field of computer technology, specifically semantic communication technology. The method is applied at a receiving end, and its specific implementation involves: receiving a semantic data packet sent by a sending end; extracting the semantic checksum contained in the semantic data packet to obtain a received semantic checksum; calculating the semantic distortion degree based on the received semantic checksum and the target semantic checksum to obtain retransmission indication information; and instructing the sending end to retransmit the semantic data packet based on the retransmission indication information. This disclosure's retransmission decision mechanism based on semantic checksum deviation not only improves the accuracy of data transmission but also optimizes communication efficiency.

Description

Technical Field

[0001] This disclosure relates to the field of computer technology, specifically to the field of semantic communication technology, and in particular to semantic communication methods, apparatus, electronic devices, and storage media. Background Technology

[0002] With the rapid development of wireless communication technology, people's requirements for communication systems have shifted from simple data transmission to a higher pursuit of information transmission quality. Especially in the era of intelligent communication, semantic communication, as an emerging communication paradigm, aims not only to maintain data integrity during transmission but also to ensure the semantic accuracy of information. Semantic communication can effectively improve the efficiency and accuracy of information transmission in complex scenarios, and is of great significance for enhancing user experience and meeting the needs of emerging applications.

[0003] However, existing semantic communication systems still face many challenges in practical applications, especially in the detection and handling of semantic errors. Traditional semantic communication systems primarily rely on the receiver's assessment of the final semantic reconstruction performance to identify erroneous semantic features. This a posteriori detection method is not only inefficient but may also lead to a waste of transmission resources, further reducing the overall efficiency of semantic communication. Summary of the Invention

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

[0005] According to one aspect of this disclosure, a semantic communication method is provided, applied at a receiving end, the method comprising:

[0006] Receive semantic data packets sent from the sender;

[0007] The semantic checksum contained in the semantic data packet is extracted to obtain the received semantic checksum;

[0008] The semantic distortion is calculated based on the received semantic check code and the target semantic check code to obtain retransmission indication information;

[0009] The retransmission instruction information indicates to the sending end whether to retransmit the semantic data packet.

[0010] According to another aspect of this disclosure, a semantic communication method is provided, applied at a sending end, the method comprising:

[0011] Obtain target source data and target semantic check code, wherein the target semantic check code is obtained by encoding the input random data using a semantic check code encoder;

[0012] The target source data is semantically encoded to obtain target semantic features;

[0013] The target semantic features are concatenated with the target semantic check code to obtain a semantic data packet, and the semantic data packet is sent to the receiving end. The receiving end extracts the semantic check code contained in the semantic data packet to obtain the received semantic check code, and calculates the distortion based on the received semantic check code and the target semantic check code to obtain retransmission indication information.

[0014] The receiver receives the retransmission indication information sent by the receiving end and determines whether to retransmit the semantic data packet based on the retransmission indication information.

[0015] According to a third aspect of this disclosure, a semantic communication method is provided, the method comprising:

[0016] The sending end acquires the target source data and the target semantic check code, and performs semantic encoding on the target source data to obtain the target semantic features; the target semantic features are concatenated with the target semantic check code to obtain a semantic data packet, and the semantic data packet is sent to the receiving end. The target semantic check code is obtained by encoding the input random data using a semantic check code encoder.

[0017] The receiving end extracts the semantic check code contained in the semantic data packet to obtain the received semantic check code, calculates the distortion based on the received semantic check code and the target semantic check code to obtain retransmission indication information, and sends the retransmission indication information to the sending end.

[0018] The sending end receives the retransmission indication information sent by the receiving end, and determines whether to retransmit the semantic data packet based on the retransmission indication information.

[0019] According to a fourth aspect of this disclosure, a semantic communication apparatus is provided for use at a receiving end, comprising:

[0020] The receiving module is used to receive semantic data packets sent from the sending end;

[0021] The extraction module is used to extract the semantic check code contained in the semantic data packet to obtain the received semantic check code;

[0022] The calculation module is used to calculate the semantic distortion based on the received semantic check code and the target semantic check code in order to obtain retransmission indication information;

[0023] The indication module is used to instruct the sending end whether to retransmit the semantic data packet according to the retransmission indication information.

[0024] According to a fifth aspect of this disclosure, a semantic communication apparatus is provided for use at a sending end, comprising:

[0025] The acquisition module is used to acquire target source data and target semantic check code. The target semantic check code is obtained by encoding the input random data using a semantic check code encoder.

[0026] The encoding module is used to perform semantic encoding on the target source data to obtain target semantic features;

[0027] The sending module is used to concatenate the target semantic features with the target semantic check code to obtain a semantic data packet, and send the semantic data packet to the receiving end. The receiving end extracts the semantic check code contained in the semantic data packet to obtain a received semantic check code, and calculates the distortion degree based on the received semantic check code and the target semantic check code to obtain retransmission indication information.

[0028] The determining module is used to receive the retransmission indication information sent by the receiving end, and determine whether to retransmit the semantic data packet according to the retransmission indication information.

[0029] According to a sixth aspect of this disclosure, an electronic device is provided, comprising:

[0030] At least one processor; and

[0031] A memory communicatively connected to the at least one processor; wherein,

[0032] The memory stores instructions that can be executed by the at least one processor, which, when executed by the at least one processor, enables the at least one processor to perform the method described in any of the above technical solutions.

[0033] According to a seventh aspect of this disclosure, a non-transitory computer-readable storage medium is provided storing computer instructions, wherein the computer instructions are used to cause the computer to perform any one of the methods described above.

[0034] According to the eighth aspect of this disclosure, a computer program product is provided, comprising a computer program that, when executed by a processor, implements the method described in any one of the above technical solutions.

[0035] This disclosure provides a semantic communication method, apparatus, device, and storage medium. The method involves receiving a semantic data packet containing a semantic checksum, extracting the received semantic checksum, and calculating the deviation between the received and target semantic checksums to obtain retransmission indication information. This retransmission indication information provides the sending end with clear guidance on whether to retransmit the semantic data packet. Therefore, this retransmission decision mechanism based on semantic checksum deviation not only improves the accuracy of data transmission but also optimizes communication efficiency by reducing unnecessary retransmissions, while ensuring the reliable transmission of key semantic information. Furthermore, this method can adapt to different network conditions by dynamically adjusting the transmission strategy, thereby enhancing the adaptability and robustness of the semantic communication system to various transmission environments.

[0036] It should be understood that the description in this section is not intended to identify key or essential features of the embodiments of this disclosure, nor is it intended to limit the scope of this disclosure. Other features of this disclosure will become readily apparent from the following description. Attached Figure Description

[0037] The accompanying drawings are provided to better understand this solution and do not constitute a limitation of this disclosure. Wherein:

[0038] Figure 1 This is a schematic diagram of the steps of a semantic communication method in one embodiment of this disclosure;

[0039] Figure 2 This is a schematic diagram of the steps of a semantic communication method in another embodiment of this disclosure;

[0040] Figure 3 This is a schematic diagram of the processing performed at the sending end in one embodiment of this disclosure;

[0041] Figure 4 This is a flowchart corresponding to the overall semantic communication method in one embodiment of this disclosure;

[0042] Figure 5 This is a flowchart of a multi-node semantic communication method in another embodiment of this disclosure;

[0043] Figure 6 A schematic block diagram of a semantic communication device in one embodiment of this disclosure;

[0044] Figure 7 A schematic block diagram of a semantic communication device according to another embodiment of this disclosure;

[0045] Figure 8 This is a block diagram of an electronic device used to implement the semantic communication method of the embodiments of this disclosure. Detailed Implementation

[0046] The exemplary embodiments of this disclosure are described below with reference to the accompanying drawings, including various details of the embodiments to aid understanding, and should be considered merely exemplary. Therefore, those skilled in the art will recognize that various changes and modifications can be made to the embodiments described herein without departing from the scope and spirit of this disclosure. Similarly, for clarity and brevity, descriptions of well-known functions and structures are omitted in the following description.

[0047] This disclosure provides a semantic communication method, see [link to relevant documentation] Figure 1 As shown, Figure 1 This is a schematic diagram of the steps of a semantic communication method in one embodiment of the present disclosure. The method is applied to a receiving end and includes:

[0048] Step S101: Receive semantic data packets sent from the sending end.

[0049] Specifically, a "semantic data packet" refers to a special data packet generated and sent by the sender. It not only contains the semantic features of the target source data (such as text, images, or videos) after semantic encoding, but also embeds a code called a "Semantic Check Code" (SCC). The semantic check code is a digital digest that represents the semantic content of the data. It is obtained by extracting key semantic features from the source data and encoding them through a semantic check code encoder. This encoder uses advanced technologies such as deep learning to capture and encode high-level semantic information of the data, thereby generating a check code closely related to the corresponding data semantics.

[0050] The specific implementation process of this scheme includes: First, the sending end performs semantic encoding on the target source data, extracting semantic features that represent its core meaning. Then, a semantic checksum encoder further processes these features to generate a fixed-length semantic checksum, which uniquely identifies the semantic content of the data. Next, this semantic checksum is embedded into the semantic data packet, forming a complete data packet together with the semantic features. This semantic data packet is then sent to the receiving end through the communication channel.

[0051] Step S102: Extract the semantic check code contained in the semantic data packet to obtain the received semantic check code.

[0052] Specifically, "received semantic checksum" refers to the semantic checksum extracted by the receiving end from the received semantic data packet. The specific implementation process of this scheme includes: when the semantic data packet arrives at the receiving end, the receiving end uses a semantic checksum extractor to process the received semantic data packet and extract the contained semantic checksum, i.e., "received semantic checksum".

[0053] Step S103: Calculate the semantic distortion degree based on the received semantic check code and the target semantic check code to obtain retransmission indication information.

[0054] Specifically, the "received semantic check code" refers to the semantic check code extracted by the receiving end from the received semantic data packet, while the "target semantic check code" is the check code obtained by encoding the input random data using a semantic check code encoder and embedded in the semantic data packet, used to maintain the semantic consistency of the data during transmission.

[0055] The specific implementation process of this scheme includes: when a semantic data packet is transmitted in the communication channel and arrives at the receiving end, the receiving end uses a semantic checksum extractor to extract the received semantic checksum from the semantic data packet. Then, the receiving end compares the extracted received semantic checksum with the target semantic checksum pre-shared by the sending end. By calculating the distortion between the two, such as using Hamming distance, Euclidean distance, or correlation coefficient, the semantic changes that may occur during data transmission are quantified. The result of this distortion calculation is used to generate a "retransmission indication message," which is a decision output indicating whether the data packet needs to be retransmitted. If the distortion exceeds a preset threshold, it indicates that the semantic information of the data packet has been significantly affected during transmission. In this case, the receiving end generates a negative acknowledgment (NACK) signal, requesting the sending end to retransmit the data packet. Conversely, if the distortion is within an acceptable range, it indicates that the data packet has successfully preserved the semantic information of the original data. The receiving end generates an acknowledgment (ACK) signal, indicating that the data packet has been correctly received and does not need to be retransmitted. This process not only improves the reliability of data transmission, but also optimizes communication efficiency and resource utilization through precise semantic distortion measurement and dynamic retransmission control.

[0056] Step S104: Instruct the sending end, based on the retransmission instruction information, whether to retransmit the semantic data packet.

[0057] Specifically, the "retransmission indication" is a control signal generated based on the receiver's assessment of the data packet's integrity and accuracy. It guides the sender on whether to retransmit the sent "semantic data packet." If the retransmission indication is a negative acknowledgment (NACK) signal, it explicitly instructs the sender to retransmit the semantic data packet to ensure data integrity and accuracy. Conversely, if the retransmission indication is an acknowledgment (ACK) signal, it informs the sender that the data packet has been correctly received and does not require retransmission. This mechanism not only improves the reliability of data transmission but also optimizes communication efficiency and resource utilization by dynamically adjusting transmission strategies. Especially in complex and ever-changing communication environments, it significantly enhances the performance and user experience of semantic communication systems.

[0058] This disclosure provides a semantic communication method, apparatus, device, and storage medium. The method involves receiving a semantic data packet containing a semantic checksum, extracting the received semantic checksum, and calculating the deviation between the received and target semantic checksums to obtain retransmission indication information. This retransmission indication information provides the sending end with clear guidance on whether to retransmit the semantic data packet. Therefore, this retransmission decision mechanism based on semantic checksum deviation not only improves the accuracy of data transmission but also optimizes communication efficiency by reducing unnecessary retransmissions, while ensuring the reliable transmission of key semantic information. Furthermore, this method can adapt to different network conditions by dynamically adjusting the transmission strategy, thereby enhancing the adaptability and robustness of the semantic communication system to various transmission environments.

[0059] In some optional embodiments, semantic distortion is calculated based on the received semantic check code and the target semantic check code to obtain retransmission indication information, including:

[0060] Receive the target semantic check code, which is obtained by encoding the input random data using a semantic check code encoder;

[0061] The degree of deviation between the received semantic check code and the target semantic check code is calculated to obtain the semantic distortion.

[0062] The semantic distortion is compared with a preset threshold, and the retransmission instruction information is obtained based on the comparison result.

[0063] Specifically, a "target semantic check code" refers to a check code generated at the sending end using a semantic check code encoder to analyze the semantic content of random input data (such as text, images, or videos). It can uniquely identify and represent the semantic information of the random input data.

[0064] The specific implementation process of this scheme is as follows: First, the sending end uses a semantic checksum encoder to extract key semantic features from the input random data and generate a target semantic checksum. This checksum is then embedded into the semantic data packet and sent to the receiving end via the communication channel. Upon receiving the semantic data packet, the receiving end uses a semantic checksum extractor to extract the "received semantic checksum." Next, the receiving end compares the extracted received semantic checksum with a pre-obtained target semantic checksum or one transmitted via other means, calculating the degree of deviation between the two. This degree of deviation is called "semantic distortion," which reflects the change or loss of semantic information during data transmission. Then, the receiving end compares the calculated semantic distortion with a pre-set threshold and generates a "retransmission indication message" based on the comparison result. If the semantic distortion exceeds a preset threshold, it indicates that the data packet has undergone significant semantic distortion during transmission. In this case, the receiver will generate a negative acknowledgment (NACK) signal, indicating that the sender needs to retransmit the data packet. Conversely, if the semantic distortion is within an acceptable range, it indicates that the data packet has successfully preserved the semantic information of the original data. The receiver will generate an acknowledgment (ACK) signal, indicating that the data packet has been correctly received and does not need to be retransmitted.

[0065] This approach first receives the target semantic checksum and calculates the deviation between the received and target semantic checksums to obtain the semantic distortion, effectively assessing the semantic integrity of data during transmission. Second, the obtained semantic distortion is compared with a preset threshold, generating a retransmission indication. This ensures that retransmission is only triggered when the data's semantic quality fails to meet requirements. This process not only optimizes the retransmission strategy and reduces network resource waste caused by unnecessary retransmissions but also improves the reliability and transmission efficiency of semantic communication, thereby enhancing the overall performance and user experience of the communication system. Furthermore, by precisely controlling the semantic distortion, this method can adapt to different network conditions and data types, demonstrating good flexibility and robustness.

[0066] In some optional embodiments, the semantic distortion is compared with a preset threshold, and retransmission indication information is obtained based on the comparison result, including:

[0067] If the semantic distortion exceeds a preset threshold, the retransmission indication information is used to indicate that the data packet should be retransmitted.

[0068] Specifically, "semantic distortion" is a quantitative indicator that measures the difference between the semantic information in the received data packet and the semantic information in the original data packet, while the "preset threshold" is an upper limit for semantic distortion set to ensure data transmission quality. The specific implementation process of this scheme is as follows: First, the receiving end extracts the "received semantic checksum" from the received data packet and compares it with the "target semantic checksum" provided by the sending end or previously shared. The semantic distortion is obtained by calculating the degree of deviation between the two. This measurement process may involve complex algorithms, such as Hamming distance, Euclidean distance, or correlation coefficients, to accurately reflect the semantic changes that occur in the data packet during transmission. Next, the calculated semantic distortion is compared with the preset threshold. If the semantic distortion exceeds this preset threshold, it indicates that the data packet has suffered significant semantic loss during transmission, which may prevent the receiving end from accurately recovering the semantic content of the original data. In this case, the system generates a "retransmission indication message." This retransmission indication message is usually a negative acknowledgment (NACK) signal, used to explicitly instruct the sending end to retransmit the damaged data packet to ensure data integrity and accuracy.

[0069] Thus, when the calculated semantic distortion exceeds a system-preset threshold, a retransmission indication is triggered, instructing the sender to retransmit the data packet. This mechanism ensures that retransmission only occurs when the semantic information of the data packet is significantly affected during transmission, thereby avoiding unnecessary bandwidth waste and transmission delays. Furthermore, by precisely controlling retransmission conditions, this scheme improves the utilization efficiency of communication resources, reduces network congestion caused by frequent retransmissions, enhances the reliability of data transmission, and ensures the accurate and error-free delivery of critical semantic information.

[0070] In some optional embodiments, the semantic distortion is compared with a preset threshold, and retransmission indication information is obtained based on the comparison result. The method further includes:

[0071] If the semantic distortion does not exceed the preset threshold, the retransmission indication information is used to indicate that the data packet does not need to be retransmitted.

[0072] Specifically, after the receiving end successfully receives the data packet, it extracts the "receive semantic checksum" from the data packet and compares it with the "target semantic checksum" provided in advance by the sending end. The semantic distortion is obtained by calculating the degree of deviation between the two. If the calculated semantic distortion does not exceed the preset threshold, it indicates that the loss of semantic information in the data packet during transmission is within an acceptable range. At this time, a "retransmission indication message" is generated, which is usually an acknowledgment (ACK) signal, to indicate that the sending end does not need to retransmit the data packet.

[0073] Thus, when the semantic distortion does not exceed a preset threshold, the retransmission indication message indicates that the data packet does not need to be retransmitted. This mechanism can significantly reduce unnecessary data retransmissions, thereby saving communication resources, reducing network congestion and transmission latency, and improving overall communication efficiency. Secondly, by avoiding redundant retransmission processes, this scheme optimizes bandwidth usage, allowing more data to be transmitted with limited network resources, thus increasing network throughput. Furthermore, this method also helps reduce the processing burden on the sending and receiving ends, lowers energy consumption, and extends equipment lifespan. In summary, this scheme, by intelligently determining whether retransmission is necessary, not only ensures the reliability and accuracy of data transmission but also improves the utilization efficiency of network resources, enhances the performance of the communication system and user experience, and achieves efficient and reliable semantic communication.

[0074] This disclosure provides a semantic communication method, see [link to relevant documentation] Figure 2 , Figure 2 This is a schematic diagram of the steps of a semantic communication method according to another embodiment of this disclosure. The method is applied at the sending end and includes:

[0075] Step S201: Obtain the target source data and the target semantic check code. The target semantic check code is obtained by encoding the input random data using a semantic check code encoder.

[0076] Specifically, "target source data" refers to data that needs to be transmitted through a communication system, such as text, images, and videos. "Target semantic check code" (SCC) is a special code that is a check code generated by a semantic check code encoder from the semantic content of random input data (such as text, images, or videos). It can uniquely identify and represent the semantic information of random input data.

[0077] The specific implementation process of this scheme is as follows: First, the sending end uses a semantic checksum encoder to analyze the input random data. This semantic checksum encoder can be based on deep learning or other advanced data processing technologies to understand the semantic meaning of the data. Next, the semantic checksum encoder extracts features from the data that represent its core semantics, such as the topic in text, objects in images, or actions in videos. Then, based on these features, the semantic checksum encoder generates a fixed-length code, namely the target semantic checksum. This checksum captures the semantic essence of the data and is used to detect and correct possible semantic distortions during data transmission. In this way, the target semantic checksum provides a semantic-level "fingerprint" for the data, enabling the receiving end to verify whether the received data retains the semantic integrity of the original data, thereby ensuring the semantic accuracy of data transmission and the effectiveness of the communication system.

[0078] Step S202: Semantically encode the target source data to obtain target semantic features.

[0079] Specifically, after obtaining the target source data, the debtor performs semantic encoding on the target source data. Semantic encoding is a process of converting data into semantic features that can represent its deeper meaning.

[0080] The specific implementation process of this scheme includes: First, using advanced data processing techniques, such as deep learning models, to analyze the target source data to identify and extract its core semantic information. These models, such as Convolutional Neural Networks (CNNs) for images, Recurrent Neural Networks (RNNs) or Transformers for text and sequence data, are capable of learning and understanding complex patterns and relationships in the data. Then, the model transforms this learned semantic information into a compact, typically vector-like, "target semantic feature." These features not only preserve the semantic content of the original data but also remove redundant information, making data transmission more efficient. The target semantic feature can then be used to generate semantic check codes or directly for data compression and transmission, thereby improving the performance of the communication system, including reducing required bandwidth, increasing transmission speed, and enhancing the semantic accuracy of the data. In this way, semantic encoding not only optimizes the efficiency of data transmission but also ensures that critical semantic information is accurately delivered to the receiving end even under poor network conditions.

[0081] Step S203: The target semantic features and the target semantic check code are concatenated to obtain a semantic data packet, and the semantic data packet is sent to the receiving end. The receiving end extracts the semantic check code contained in the semantic data packet to obtain the received semantic check code, and calculates the distortion based on the received semantic check code and the target semantic check code to obtain retransmission indication information.

[0082] Specifically, after extracting target semantic features from the target source data using semantic coding technology, the sending end concatenates these features with a target semantic checksum to form a "semantic data packet" containing both semantic and checksum information. This data packet is then sent to the receiving end via the communication channel. Upon receiving the semantic data packet, the receiving end uses a semantic checksum extractor to extract a "received semantic checksum." Subsequently, the receiving end compares the received semantic checksum with the target semantic checksum provided by the sending end, calculating the distortion between the two, i.e., the degree of deviation or change in semantic information during transmission. Based on a comparison of the calculated distortion with a preset threshold, the receiving end can determine whether a retransmission request is needed. If the distortion exceeds the threshold, indicating significant semantic distortion during transmission, the receiving end generates a retransmission indication (e.g., NACK), instructing the sending end to retransmit the data packet. If the distortion is within acceptable limits, the receiving end generates an acknowledgment signal (e.g., ACK), indicating that the data packet has been correctly received and does not require retransmission. This process not only ensures the reliability of data transmission but also optimizes the utilization of communication resources and improves communication efficiency and overall system performance by precisely controlling retransmission conditions.

[0083] Step S204: Receive retransmission indication information sent by the receiving end, and determine whether to retransmit the semantic data packet based on the retransmission indication information.

[0084] Specifically, a "retransmission indication message" is a feedback message sent by the receiving end to inform the sending end whether the received semantic data packet needs to be retransmitted. Upon receiving this retransmission indication message, the sending end determines whether to retransmit the semantic data packet based on the signal type. Thus, the semantic content of the data is identified by acquiring the target source data and its corresponding target semantic checksum, and the target source data is semantically encoded to obtain target semantic features. These semantic features are then concatenated with the target semantic checksum to form a semantic data packet, which is sent to the receiving end. The receiving end extracts the received semantic checksum and compares it with the target semantic checksum to calculate the distortion, thereby obtaining the retransmission indication message. The sending end then decides whether to retransmit the data packet based on this signal. The beneficial effects of this process include: ensuring the semantic accuracy of data transmission by effectively detecting and correcting potential semantic distortions through checksum comparison; optimizing the use of network resources by retransmitting data only when necessary, reducing bandwidth waste and transmission latency; enhancing communication reliability by dynamically adjusting transmission strategies to adapt to different network conditions, thereby improving the system's adaptability and robustness to various environments; and ultimately, these measures work together to improve the overall performance of the semantic communication system, ensuring the accurate and error-free transmission of critical information, thus providing users with a more efficient and reliable communication experience.

[0085] In some optional embodiments, the target source data is semantically encoded to obtain target semantic features, including:

[0086] Semantic features are extracted from the target source data to obtain all semantic features;

[0087] The weight of each semantic feature is calculated based on the importance of all semantic features;

[0088] The semantic features are filtered according to the weight of each semantic feature to obtain the target semantic features.

[0089] Specifically, first, advanced data processing techniques, such as deep learning models, are used to analyze the target source data to identify and extract all relevant semantic features. These features may include key phrases in text, main objects in images, or important events in videos. Next, the weight of each extracted semantic feature is calculated based on its importance. This step may involve feature importance evaluation algorithms, such as model-based feature selection methods or statistical analysis, to determine which features are most critical for conveying the semantic content of the data. Then, based on the calculated weights, the semantic features are filtered, and features with higher weights are selected as "target semantic features" due to their importance in the semantic expression of the data. Finally, these target semantic features are used in subsequent semantic encoding or to generate semantic check codes to ensure that critical semantic information is preserved and accurately transmitted during data transmission.

[0090] In this way, by extracting all semantic features from the target source data and calculating weights based on their importance, the most representative target semantic features are selected. The beneficial effects of this process are mainly reflected in the following aspects: First, it ensures that the most important semantic information is prioritized for preservation and transmission during data transmission, thereby improving the accuracy and reliability of information transmission; second, through weight calculation and feature selection mechanisms, the amount of data to be transmitted can be effectively reduced, bandwidth requirements can be lowered, and transmission latency can be reduced, thereby improving transmission efficiency; in addition, this method also helps to reduce the complexity of data processing at the receiving end, because only the most important features are transmitted and processed, which can reduce the consumption of computing resources and improve processing speed; finally, by focusing on key semantic features, this scheme enhances the system's robustness to noise and interference, maintaining high data transmission quality even in less than ideal communication environments, thus bringing users a more stable and high-quality communication experience.

[0091] In some optional embodiments, after filtering the semantic features according to the weight of each semantic feature to obtain the target semantic features, the method further includes:

[0092] The packet length is determined based on the maximum transmission unit allowed by the network link;

[0093] The target semantic features are grouped according to the data packet length to obtain multiple groups of target semantic features.

[0094] Specifically, the "Maximum Transmission Unit" (MTU) refers to the maximum data packet size that a network link can handle in a single transmission. This is an important network parameter that determines the maximum size of a data packet transmitted in the network. "Target semantic features" refers to the set of features extracted from the target source data and filtered by weights, representing the core semantic content of the data.

[0095] The specific implementation process of this scheme includes: First, determining the maximum allowable length of a single data packet based on the MTU of the network link. This ensures that data packets are not fragmented or dropped during transmission due to exceeding the processing capacity of network devices. Then, according to this determined data packet length, the target semantic features are grouped into multiple subgroups, with the number and size of features in each subgroup not exceeding the MTU limit. This grouping process ensures that each data packet can be effectively processed by network devices and facilitates reassembly and recovery of data packets at the receiving end. Since each data packet contains an appropriate amount of semantic features—neither too large, leading to transmission failure, nor too small, causing bandwidth waste—data packet transmission becomes more efficient and reliable. Furthermore, grouping allows for finer-grained control during data transmission. For example, transmission strategies can be dynamically adjusted based on the importance and transmission priority of each group of features, further optimizing network resource utilization and improving data transmission efficiency. In summary, this method of determining data packet length based on MTU and grouping target semantic features not only improves the reliability and efficiency of data transmission but also enhances the system's adaptability to different network environments, providing strong support for achieving efficient and reliable semantic communication.

[0096] This method ensures that the generated data packets are of moderate size, avoiding reduced transmission efficiency or packet loss due to excessively large packets, thus improving data transmission reliability. Secondly, through reasonable grouping, network bandwidth can be utilized more effectively, reducing bandwidth waste caused by excessively small packets and improving network resource utilization. Furthermore, since each data packet contains an appropriate amount of well-organized semantic features, it facilitates quick and accurate recovery of the original data; therefore, this method also helps to more efficiently reassemble data at the receiving end. Finally, by dynamically adjusting the data packet size and feature grouping, this scheme can adapt to different network environments and transmission requirements, demonstrating good flexibility and robustness, thus providing strong support for achieving efficient and reliable semantic communication. In summary, this method of determining data packet length and feature grouping based on MTU significantly improves the overall performance and user experience of the semantic communication system by optimizing data packet organization and transmission.

[0097] For ease of understanding of the scheme of the transmitting end embodiment of this disclosure, see [link to relevant documentation]. Figure 3 , Figure 3 This is a schematic diagram of the processing performed at the sending end in one embodiment of this disclosure. The process includes:

[0098] 1. Semantic Feature Extraction (Zs)

[0099] Input data: The cube on the left side of the diagram represents the input data, labeled Zs.

[0100] Feature extraction: Key semantic features are extracted from the input data Zs. These features represent the core semantic information of the data. In the diagram, points of different color depths represent the extracted semantic features, and points in each cube represent different features.

[0101] 2. Semantic feature grouping (Z^1_s, Z^2_s, ..., Z^K_s)

[0102] Feature grouping: The extracted semantic features are divided into multiple groups, and each group of features Z^i_s (where i represents the index of the group) contains a part of key semantic information.

[0103] Purpose of grouping: This grouping method helps to better manage and control the size of data packets during transmission, and also facilitates data reassembly and recovery at the receiving end.

[0104] 3. Generate Semantic Check Code (SCC)

[0105] SCC generation: A corresponding semantic check code (SCC) is generated for each set of semantic features. SCC is used to detect and correct semantic distortions that may occur during data transmission.

[0106] SCC Function: In the diagram, the generated SCCs are represented by darker colored cubes. Each SCC is associated with a set of semantic features Z^i_s to ensure the integrity and accuracy of the set of features during transmission.

[0107] 4. Data packet generation (Y1, Y2, ..., Yk)

[0108] Data packet concatenation: Each semantic feature Z^i_s is concatenated with its corresponding SCC to form a complete data packet Yi.

[0109] Data packet structure: Each data packet Yi contains two parts: one part is semantic features (light-colored dots), and the other part is SCC (dark-colored dots) used for error detection and correction.

[0110] 5. Data packet transmission

[0111] Sending end: The sending end sends the generated data packets Y1, Y2, ..., YK to the receiving end through the communication channel.

[0112] The above embodiments have described in detail the processing at the sending end and the receiving end. To facilitate a comprehensive understanding of the technical solution of this application, this disclosure describes the communication process of the solution as a whole. The method includes: the sending end acquiring target source data and a target semantic checksum, and semantically encoding the target source data to obtain target semantic features; concatenating the target semantic features with the target semantic checksum to obtain a semantic data packet, and sending the semantic data packet to the receiving end. The target semantic checksum is obtained by encoding the input random data using a semantic checksum encoder.

[0113] The receiving end extracts the semantic check code contained in the semantic data packet to obtain the received semantic check code, calculates the distortion based on the received semantic check code and the target semantic check code to obtain the retransmission indication information, and sends the retransmission indication information to the sending end.

[0114] The sending end receives the retransmission indication information sent by the receiving end, and determines whether to retransmit the semantic data packet based on the retransmission indication information.

[0115] Specifically, firstly, random data, such as text, images, or videos, is acquired, and a corresponding "target semantic checksum" is generated. This is an encoding used to identify and verify the integrity of the semantic content of the data. Next, the sending end performs "semantic encoding" on the target source data, extracting key semantic features to obtain "target semantic features." These features represent the core meaning of the data and are used in subsequent transmission and verification processes. Then, the sending end concatenates the target semantic features with the target semantic checksum to form a "semantic data packet" containing semantic and verification information, and sends this data packet to the receiving end via the communication network. Upon receiving the semantic data packet, the receiving end extracts the "received semantic checksum" and compares it with the target semantic checksum provided by the sending end, calculating the "distortion" between the two, i.e., the degree of deviation or change in the semantic information during transmission. Based on the comparison result of the distortion with a preset threshold, the receiving end generates a corresponding "retransmission indication message," indicating whether the data packet needs to be retransmitted, and feeds this signal back to the sending end. After receiving the retransmission indication message, the sending end decides whether to retransmit the semantic data packet based on the signal content.

[0116] In this process, the sending end acquires the target source data and target semantic checksum, performs semantic encoding on the source data to obtain target semantic features, and then concatenates these features with the checksum to form a semantic data packet, which is sent to the receiving end. The receiving end extracts the received semantic checksum and compares it with the target semantic checksum to calculate the distortion, and generates a retransmission indication message based on this, which is sent back to the sending end. The sending end decides whether to retransmit the data packet based on this signal. The beneficial effects of this process include: ensuring high reliability of data transmission, as retransmission is only triggered when the semantic distortion of the data packet exceeds a preset threshold, thus avoiding unnecessary bandwidth waste and transmission delays; optimizing the use of network resources, reducing network congestion caused by frequent retransmissions by precisely controlling retransmission conditions; improving communication efficiency, as key semantic information is delivered preferentially and accurately; and enhancing the robustness of the system, enabling it to adapt to different network environments and data characteristics, achieving more flexible and efficient semantic information transmission. In summary, this scheme significantly improves the overall performance and user experience of the semantic communication system through intelligent semantic verification and retransmission mechanisms.

[0117] See Figure 4 , Figure 4 This is a flowchart corresponding to the overall semantic communication method in one embodiment of this disclosure. The following is a detailed explanation of the working principle of this diagram:

[0118] 1. Transmitter

[0119] Data packet generation: The sending end generates a data packet Yk containing semantic features Zk_i and the corresponding semantic check code SCCk_i.

[0120] Data packet transmission: Data packet Yk is sent to the receiving end through the channel.

[0121] 2. Channel

[0122] Transmission process: Data packets are transmitted in the channel and may be affected by noise and interference, resulting in data distortion.

[0123] 3. Receiver

[0124] Data packet reception: The receiving end receives the data packet Y^k_i and extracts the semantic features Z^k_i and the semantic check code SCC^k_i from it.

[0125] Distortion detection: The distortion Dc(k) between the received semantic check code and the original semantic check code is calculated using a distortion detector.

[0126] 4. Distortion Adjudicator

[0127] Decision logic: Based on the distortion degree Dc(k) and the preset threshold, determine whether the data packet needs to be retransmitted.

[0128] If the distortion exceeds the threshold, a negative acknowledgment (NACK) signal is generated, requesting the sender to retransmit the data packet.

[0129] If the distortion is within an acceptable range, an acknowledgment (ACK) signal is generated, indicating that the data packet has been correctly received.

[0130] 5. Retransmission decision

[0131] Retransmission indication information: The receiving end sends retransmission indication information (ACK or NACK) back to the sending end.

[0132] If a NACK signal is received, the sender will regenerate and send the data packet Yk.

[0133] If an ACK signal is received, the sender will continue to send the next data packet Yk+1.

[0134] 6. Multi-round transmission

[0135] Multi-round transmission: This process can be repeated multiple times until all data packets are successfully received (i.e., an ACK signal is received).

[0136] This diagram illustrates the complete process of packet generation, transmission, reception, and retransmission in a HARQ (Hybrid Automatic Repeat Request) system based on semantic checksums. By embedding semantic checksums into the packets and performing distortion detection and judgment at the receiving end, semantic distortion during data transmission can be effectively detected and corrected, improving the reliability and accuracy of data transmission. Furthermore, by dynamically adjusting the retransmission strategy, network resource utilization can be optimized, improving communication efficiency.

[0137] The above embodiments are merely illustrative of automatic transmission requests between end-to-end; see also... Figure 5 , Figure 5 This is a flowchart of a multi-node semantic communication method according to another embodiment of this disclosure. The flowchart describes a system architecture for a hybrid Automatic Repeat Request (HARQ) method based on Semantic Check Code (SCC) to improve the effectiveness and reliability of information transmission in a semantic communication system. The detailed working principle of this flowchart is as follows:

[0138] First, (1) Input information (Iin): The sending end receives random input data (such as text, images, videos, etc.).

[0139] (2) Semantic Check Code Encoder: Encodes the input random data to generate the target semantic check code (SCC).

[0140] (3) Shared Semantic Check Code

[0141] The sender, relay node, and receiver share the same semantic check code encoder model to ensure consistency in semantic distortion calculation.

[0142] At the sending end (Transmitter S):

[0143] (1) Input information (Im): The sending end receives raw information (such as text, images, videos, etc.).

[0144] (2) Joint Source Channel Encoder (JSCC Encoder): Semantically encodes the input information and extracts key semantic features (Xs).

[0145] (3) Feature Importance Extractor: Evaluates the importance of semantic features and generates weights (W).

[0146] (4) Feature Selector: Selects important semantic features (Zs) based on their weights.

[0147] (5) Packet Combiner: Combines the selected semantic features and the target semantic check code into a semantic data packet (Ys).

[0148] (6) Channel: Send semantic data packets to the relay node (Relay R) or directly to the receiver (Receiver D) via the wireless channel.

[0149] At the relay node (Relay R):

[0150] (1) Channel: Receives semantic data packets (Ys) from the sender.

[0151] (2) Semantic check code extractor: Extracts semantic check codes (SCCr) from received data packets.

[0152] (3) Semantic check code cache: caches the extracted semantic check codes (SCCr);

[0153] (4) Semantic packet cache: caches the extracted semantic features (Zr).

[0154] (5) Semantic check code distortion detector: detects the distortion (Dr) between the received semantic check code and the known semantic check code.

[0155] (6) Distortion Adjudicator: Determines whether retransmission is needed based on the degree of distortion and generates an indication signal (ACK / NACK).

[0156] (7) Packet Combiner: Based on the indication signal and the feedback from the destination node, it decides whether to combine the cached semantic features and semantic check codes into a new packet (Xr).

[0157] (8) Channel: Sends new data packets to the receiver (Receiver D).

[0158] At the receiving end (Receiver D):

[0159] (1) Channel: Receive semantic data packets (YR or YRD) from the sender or relay node.

[0160] (2) Semantic check code extractor: Extracts semantic check codes (SCCd) from received data packets.

[0161] (3) Semantic packet cache: caches the extracted semantic features (Xd).

[0162] (4) Semantic check code distortion detector: detects the distortion (Dd) between the received semantic check code and the known semantic check code.

[0163] (5) Distortion Adjudicator: Determines whether retransmission is needed based on the degree of distortion and generates an indication signal (ACK / NACK).

[0164] (6) Joint Source Channel Decoder (JSCC Decoder): After receiving the ACK indication information of all data packets, it retrieves all received semantic features from the semantic feature buffer and performs semantic decoding to recover the original information (Iout).

[0165] This architecture enables semantic communication in end-to-end or multi-node collaborative modes, and improves transmission reliability and system efficiency through dynamic retransmission decision-making strategies.

[0166] The following describes an apparatus embodiment of this application, which can be used to execute the semantic communication method in the above embodiments of this application. For details not disclosed in the apparatus embodiments of this application, please refer to the embodiments of the semantic communication method described above.

[0167] This disclosure also provides a semantic communication device 600, such as Figure 6 As shown, it is applied to the receiving end and includes:

[0168] The receiving module 601 is used to receive semantic data packets sent from the sending end;

[0169] The extraction module 602 is used to extract the semantic check code contained in the semantic data packet to obtain the received semantic check code;

[0170] The calculation module 603 is used to calculate the semantic distortion degree based on the received semantic check code and the target semantic check code in order to obtain retransmission indication information;

[0171] The indication module 604 is used to instruct the sending end whether to retransmit the semantic data packet according to the retransmission indication information.

[0172] In some optional embodiments, the calculation module 603 calculates the semantic distortion based on the received semantic check code and the target semantic check code to obtain retransmission indication information, including:

[0173] Receive the target semantic check code, which is obtained by encoding the input random data using a semantic check code encoder;

[0174] The degree of deviation between the received semantic check code and the target semantic check code is calculated to obtain the semantic distortion.

[0175] The semantic distortion is compared with a preset threshold, and the retransmission instruction information is obtained based on the comparison result.

[0176] In some optional embodiments, the calculation module 603 compares the semantic distortion with a preset threshold and obtains retransmission indication information based on the comparison result, including:

[0177] If the semantic distortion exceeds a preset threshold, the retransmission indication information is used to indicate that the data packet should be retransmitted.

[0178] In some optional embodiments, the calculation module 603 compares the semantic distortion with a preset threshold and obtains retransmission indication information based on the comparison result, and further includes:

[0179] If the semantic distortion does not exceed the preset threshold, the retransmission indication information is used to indicate that the data packet does not need to be retransmitted.

[0180] This disclosure also provides a semantic communication device 700, such as Figure 7 As shown, it is applied to the sending end and includes:

[0181] The acquisition module 701 is used to acquire the target source data and the target semantic check code. The target semantic check code is obtained by encoding the input random data using a semantic check code encoder.

[0182] Encoding module 702 is used to perform semantic encoding on the target source data to obtain target semantic features;

[0183] The sending module 703 is used to concatenate the target semantic features with the target semantic check code to obtain a semantic data packet, and send the semantic data packet to the receiving end. The receiving end extracts the semantic check code contained in the semantic data packet to obtain the received semantic check code, and calculates the distortion based on the received semantic check code and the target semantic check code to obtain retransmission indication information.

[0184] The determination module 704 is used to receive retransmission indication information sent by the receiving end, and determine whether to retransmit the semantic data packet based on the retransmission indication information.

[0185] In some optional embodiments, the encoding module 702 performs semantic encoding on the target source data to obtain target semantic features, including:

[0186] Semantic features are extracted from the target source data to obtain all semantic features;

[0187] The weight of each semantic feature is calculated based on the importance of all semantic features;

[0188] The semantic features are filtered according to the weight of each semantic feature to obtain the target semantic features.

[0189] In some optional embodiments, after the encoding module 702 filters the semantic features according to the weight of each semantic feature to obtain the target semantic features, it further includes:

[0190] The packet length is determined based on the maximum transmission unit allowed by the network link;

[0191] The target semantic features are grouped according to the data packet length to obtain multiple groups of target semantic features.

[0192] The acquisition, storage, and application of user personal information involved in the technical solution disclosed herein comply with the provisions of relevant laws and regulations and do not violate public order and good morals.

[0193] According to embodiments of this disclosure, this disclosure also provides an electronic device, a readable storage medium, and a computer program product.

[0194] Figure 8 A schematic block diagram of an example electronic device 800 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 may also represent various forms of mobile devices, such as personal digital processors, cellular phones, smartphones, wearable devices, and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely illustrative and are not intended to limit the implementation of the present disclosure described and / or claimed herein.

[0195] like Figure 8As shown, the electronic device 800 includes a computing unit 801, which can perform various appropriate actions and processes according to a computer program stored in a read-only memory (ROM) 802 or a computer program loaded from a storage unit 808 into a random access memory (RAM) 803. The RAM 803 may also store various programs and data required for the operation of the device 800. The computing unit 801, ROM 802, and RAM 803 are interconnected via a bus 804. An input / output (I / O) interface 805 is also connected to the bus 804.

[0196] Multiple components in device 800 are connected to I / O interface 805, including: input unit 806, such as keyboard, mouse, etc.; output unit 807, such as various types of monitors, speakers, etc.; storage unit 808, such as disk, optical disk, etc.; and communication unit 809, such as network card, modem, wireless transceiver, etc. Communication unit 809 allows device 800 to exchange information / data with other devices through computer networks such as the Internet and / or various telecommunications networks.

[0197] The computing unit 801 can be a variety of general-purpose and / or special-purpose processing components with processing and computing capabilities. Some examples of the computing unit 801 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various special-purpose artificial intelligence (AI) computing chips, various computing units running machine learning model algorithms, a digital signal processor (DSP), and any suitable processor, controller, microcontroller, etc. The computing unit 801 performs the various methods and processes described above, such as semantic communication methods. For example, in some embodiments, the semantic communication method may be implemented as a computer software program tangibly contained in a machine-readable medium, such as storage unit 808. In some embodiments, part or all of the computer program may be loaded and / or installed on device 800 via ROM 802 and / or communication unit 809. When the computer program is loaded into RAM 803 and executed by the computing unit 801, one or more steps of the applet distribution described above may be performed. Alternatively, in other embodiments, the computing unit 801 may be configured to perform semantic communication methods by any other suitable means (e.g., by means of firmware).

[0198] Various embodiments of the systems and techniques described above 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), systems-on-a-chip (SoCs), payload-programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various embodiments may include implementations in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which may be a dedicated or general-purpose programmable processor, capable of receiving data and instructions from a storage system, at least one input device, and at least one output device, and transmitting data and instructions to the storage system, the at least one input device, and the at least one output device.

[0199] The program code used to implement the methods of this disclosure may be written in any combination of one or more programming languages. This program code may be provided to a processor or controller of a general-purpose computer, special-purpose computer, or other programmable data processing apparatus, such that when executed by the processor or controller, the program code causes the functions / operations specified in the flowcharts and / or block diagrams to be implemented. The program code may be executed entirely on a machine, partially on a machine, as a standalone software package partially on a machine and partially on a remote machine, or entirely on a remote machine or server.

[0200] In the context of this disclosure, a machine-readable medium can be a tangible medium that may contain or store a program for use by or in conjunction with an instruction execution system, apparatus, or device. A machine-readable medium can be a machine-readable signal medium or a machine-readable storage medium. Machine-readable media can be, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination of the foregoing. More specific examples of machine-readable storage media include electrical connections based on one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination of the foregoing.

[0201] To provide interaction with a user, the systems and techniques described herein can be implemented on a computer having: a display device for displaying information to the user (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor); and a keyboard and pointing device (e.g., a mouse or trackball) through which the user provides input to the computer. Other types of devices can also be used to provide interaction with the user; for example, 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 sound input, voice input, or tactile input).

[0202] The systems and technologies described herein can be implemented in computing systems that include backend components (e.g., as a data server), or computing systems that include middleware components (e.g., an application server), or computing systems that include frontend components (e.g., a user computer with a graphical user interface or web browser through which a user can interact with embodiments of the systems and technologies described herein), or any combination of such backend, middleware, or frontend components. The components of the system can be interconnected via digital data communication of any form or medium (e.g., a communication network). Examples of communication networks include local area networks (LANs), wide area networks (WANs), and the Internet.

[0203] Computer systems can include clients and servers. Clients and servers are generally located far apart and typically interact via communication networks. Client-server relationships are created by computer programs running on the respective computers and having a client-server relationship with each other. Servers can be cloud servers, servers in distributed systems, or servers incorporating blockchain technology.

[0204] It should be understood that the various forms of processes shown above can be used to rearrange, add, or delete steps. For example, the steps described in this disclosure can be executed in parallel, sequentially, or in different orders, as long as the desired result of the technical solution disclosed in this disclosure can be achieved, and this is not limited herein.

[0205] The specific embodiments described above do not constitute a limitation on the scope of protection of this disclosure. Those skilled in the art should understand that various modifications, combinations, sub-combinations, and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this disclosure should be included within the scope of protection of this disclosure.

Claims

1. A semantic communication method applied at a receiving end, the method comprising: Receive semantic data packets sent from the sending end. The semantic data packets are generated by the sending end in the following manner: extract semantic features from the target source data to obtain all semantic features, calculate the weight of each semantic feature according to the importance of all semantic features, and filter the semantic features according to the weight of each semantic feature to obtain the target semantic features. The data packet length is determined based on the maximum transmission unit allowed by the network link, and the target semantic features are grouped according to the data packet length to obtain multiple groups of target semantic features; For each set of target semantic features, a corresponding target semantic check code is generated. Each set of target semantic features is concatenated with its corresponding target semantic check code to obtain multiple semantic data packets. The semantic checksum contained in the semantic data packet is extracted to obtain the received semantic checksum; The semantic distortion is calculated based on the received semantic check code and the target semantic check code to obtain retransmission indication information; The retransmission indication information is used to instruct the sending end whether to retransmit the semantic data packet; The step of calculating the semantic distortion based on the received semantic check code and the target semantic check code to obtain retransmission indication information specifically includes: The target semantic check code is received. The target semantic check code is obtained by encoding the input random data using a semantic check code encoder. It can uniquely identify and represent the semantic information of the random input data. The degree of deviation between the received semantic check code and the target semantic check code is calculated to obtain the semantic distortion degree, which reflects the change or loss of semantic information in the data during transmission. The semantic distortion is compared with a preset threshold, and retransmission indication information is obtained based on the comparison result. The sending end and the receiving end share the target semantic check code to ensure consistency in semantic distortion calculation.

2. The method according to claim 1, wherein, The step of comparing the semantic distortion with a preset threshold and obtaining retransmission indication information based on the comparison result includes: If the semantic distortion exceeds the preset threshold, the retransmission indication information is used to indicate that the data packet should be retransmitted.

3. The method according to claim 2, wherein, The step of comparing the semantic distortion with a preset threshold and obtaining retransmission indication information based on the comparison result further includes: If the semantic distortion does not exceed the preset threshold, the retransmission indication information is used to indicate that the data packet does not need to be retransmitted.

4. A semantic communication method applied at a sending end, the method comprising: Obtain target source data and target semantic check code, wherein the target semantic check code is obtained by encoding the input random data using a semantic check code encoder; Semantic features are extracted from the target source data to obtain all semantic features; The weight of each semantic feature is calculated based on the importance of all semantic features; The semantic features are filtered according to the weight of each semantic feature to obtain the target semantic features; The packet length is determined based on the maximum transmission unit allowed by the network link; The target semantic features are grouped according to the data packet length to obtain multiple groups of target semantic features, wherein a corresponding target semantic check code is generated for each group of target semantic features. Each set of target semantic features is concatenated with its corresponding target semantic checksum to obtain multiple semantic data packets, which are then sent to the receiving end. The receiving end extracts the semantic checksum contained in the semantic data packets to obtain a received semantic checksum, and calculates the distortion degree based on the received semantic checksum and the target semantic checksum to obtain retransmission indication information. Specifically, the receiving end calculates the semantic distortion degree based on the received semantic checksum and the target semantic checksum to obtain retransmission indication information, which includes: receiving the target semantic checksum, which is obtained by encoding the input random data using a semantic checksum encoder and can uniquely identify and represent the semantic information of the random input data; calculating the degree of deviation between the received semantic checksum and the target semantic checksum to obtain the semantic distortion degree, which reflects the change or loss of semantic information during data transmission; comparing the semantic distortion degree with a preset threshold, and obtaining retransmission indication information based on the comparison result. The receiver receives the retransmission indication information sent by the receiving end and determines whether to retransmit the semantic data packet based on the retransmission indication information; wherein the sending end and the receiving end share the target semantic check code to ensure the consistency of semantic distortion calculation.

5. A semantic communication device, applied at a receiving end, comprising: The receiving module is used to receive semantic data packets sent from the sending end. The semantic data packets are generated by the sending end in the following way: extracting semantic features from the target source data to obtain all semantic features, calculating the weight of each semantic feature according to the importance of all semantic features, and filtering the semantic features according to the weight of each semantic feature to obtain the target semantic features. The data packet length is determined based on the maximum transmission unit allowed by the network link, and the target semantic features are grouped according to the data packet length to obtain multiple groups of target semantic features; For each set of target semantic features, a corresponding target semantic check code is generated. Each set of target semantic features is concatenated with its corresponding target semantic check code to obtain multiple semantic data packets. The extraction module is used to extract the semantic check code contained in the semantic data packet to obtain the received semantic check code; The calculation module is used to calculate semantic distortion based on the received semantic checksum and the target semantic checksum to obtain retransmission indication information. Specifically, the calculation module is used to receive the target semantic checksum, which is obtained by encoding the input random data using a semantic checksum encoder, and can uniquely identify and represent the semantic information of the random input data; calculate the degree of deviation between the received semantic checksum and the target semantic checksum to obtain the semantic distortion, which reflects the change or loss of semantic information during data transmission; compare the semantic distortion with a preset threshold, and obtain retransmission indication information based on the comparison result. The indication module is used to instruct the sending end whether to retransmit the semantic data packet according to the retransmission indication information, wherein the sending end and the receiving end share the target semantic check code to ensure the consistency of semantic distortion calculation.

6. The apparatus according to claim 5, wherein, The calculation module compares the semantic distortion with a preset threshold and obtains retransmission indication information based on the comparison result, including: If the semantic distortion exceeds the preset threshold, the retransmission indication information is used to indicate that the data packet should be retransmitted.

7. The apparatus according to claim 6, wherein, The calculation module compares the semantic distortion with a preset threshold and obtains retransmission indication information based on the comparison result. It also includes: If the semantic distortion does not exceed the preset threshold, the retransmission indication information is used to indicate that the data packet does not need to be retransmitted.

8. A semantic communication device, applied at a sending end, comprising: The acquisition module is used to acquire target source data and target semantic check code. The target semantic check code is obtained by encoding the input random data using a semantic check code encoder. The encoding module is used to extract semantic features from the target source data to obtain all semantic features; and to calculate the weight of each semantic feature based on the importance of all semantic features. The semantic features are filtered according to the weight of each semantic feature to obtain the target semantic features; The data packet length is determined based on the maximum transmission unit allowed by the network link; the target semantic features are grouped according to the data packet length to obtain multiple groups of target semantic features; wherein, for each group of target semantic features, a corresponding target semantic check code is generated. The sending module is used to concatenate each set of target semantic features with its corresponding target semantic checksum to obtain multiple semantic data packets, and send the semantic data packets to the receiving end. The receiving end extracts the semantic checksum contained in the semantic data packets to obtain the received semantic checksum, and calculates the distortion degree based on the received semantic checksum and the target semantic checksum to obtain retransmission indication information. Specifically, the receiving end calculates the semantic distortion degree based on the received semantic checksum and the target semantic checksum to obtain retransmission indication information, which includes: receiving the target semantic checksum, which is obtained by encoding the input random data using a semantic checksum encoder and can uniquely identify and represent the semantic information of the random input data; calculating the degree of deviation between the received semantic checksum and the target semantic checksum to obtain the semantic distortion degree, which reflects the change or loss of semantic information during data transmission; comparing the semantic distortion degree with a preset threshold, and obtaining retransmission indication information based on the comparison result. The determining module is used to receive the retransmission indication information sent by the receiving end, and determine whether to retransmit the semantic data packet according to the retransmission indication information; wherein the sending end and the receiving end share the target semantic check code to ensure the consistency of semantic distortion calculation.

9. An electronic device, comprising: 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 to enable the at least one processor to perform the method of any one of claims 1-4.

10. A non-transitory computer-readable storage medium storing computer instructions, wherein, The computer instructions are used to cause the computer to perform the method according to any one of claims 1-4.

11. A computer program product comprising a computer program that, when executed by a processor, implements the method according to any one of claims 1-4.

Images