Semantic-based ltp protocol optimization method and system

Abstract

The application discloses a kind of based on semantic LTP protocol optimization method and system, method includes: original file is encapsulated as bundle format data and is issued to LTP layer;According to the semantic characteristics of the bundle format data, it is split and reorganized into multiple segments;Different channel conditions are used in the model training of semantic data transmission system model based on Transformer architecture, and the optimal segment combination mode under the corresponding scene is obtained according to the transmission accuracy under different segment combination modes;Actual transmission data is carried out in the optimal segment combination mode, if transmission error occurs, then retransmission standard judgment is carried out to error segment, if it meets retransmission judgment standard, then retransmission is carried out after adding semantic feature point to the error segment;During the retransmission process, the optimal segment combination is preferentially retransmitted.The application uses the least retransmission times to achieve reliable transmission while reducing the packet loss rate, and effectively improves information timeliness.

Description

Technical Field

[0001] This invention belongs to the field of satellite and deep space communication technology, specifically relating to a semantic-based LTP protocol optimization method and system. Background Technology

[0002] Existing DTN-based data transmission requires data to be transmitted and recovered (encoded and decoded) at the bit level. However, due to the development of deep space communication and the increase in the amount of data transmitted, excessive transmission latency cannot meet the needs. Especially in future exploration activities at the edge of the solar system, existing transmission technologies cannot meet the demands for highly time-sensitive data such as control command data. Therefore, some research has focused on improving existing LTP transmission mechanisms, mostly considering optimizing LTP segment size and protocol parameters such as retransmission timers. However, these methods all adhere to the existing LTP framework and transmission process, thus failing to overcome limitations. Neither approach fundamentally solves the problem of poor timeliness. Under given channel conditions, the packet loss rate has a clear lower bound as segment size changes, which constrains segment size optimization methods. Furthermore, existing methods do not consider the characteristics of the transmitted data itself and do not utilize semantic features between data to aid recovery and reduce retransmissions. To date, existing DTN protocols cannot solve the problem of high-time-sensitivity data transmission in long-distance space communication, necessitating the development of new transmission protocols. Summary of the Invention

[0003] To address the aforementioned problems, this invention provides a semantic-based LTP protocol optimization method and system. In satellite and deep space network communication scenarios, for single-hop data transmission, the method combines the semantic features of the transmitted file to perform semantic encoding and decoding of the original data and semantic-level transmission, and improves the retransmission mechanism of the existing LTP transmission protocol, effectively improving the timeliness of information under high error channels.

[0004] According to a first aspect of the present disclosure, a semantic-based LTP protocol optimization method is provided, the method comprising the following steps:

[0005] The original file is encapsulated into bundle format data and sent to the LTP layer;

[0006] Based on the semantic features of the bundle format data, it is split and reorganized into multiple segments;

[0007] The model of a semantic data transmission system based on the Transformer architecture is used to train the model under different channel conditions. Based on the transmission accuracy under different segment combination methods, the optimal segment combination method is obtained for the corresponding scenario.

[0008] The data is actually transmitted using the optimal segment combination. If a transmission error occurs, the erroneous segment is judged according to the retransmission criteria. If the retransmission criteria are met, semantic feature points are added to the erroneous segment before retransmission.

[0009] During retransmission, the optimal segment combination is retransmitted first.

[0010] In one embodiment, the semantic data transmission system model includes a semantic encoder and a semantic decoder. The semantic encoder includes a text embedding layer, a Transformer architecture, a fully connected layer, and a quantization layer. The semantic decoder includes a dequantization layer, a fully connected layer, a Transformer architecture, and a softmax layer.

[0011] In one embodiment, obtaining the optimal segment combination method includes:

[0012] Multiple minimal semantic segments are formed based on the semantic features of bundle format data;

[0013] The ways to combine all the smallest semantic segments;

[0014] Under different channel conditions, the semantic data transmission system transmits various combinations of methods and obtains the transmission accuracy in all cases;

[0015] The optimal segment combination is the one that achieves the highest transmission accuracy under each channel condition.

[0016] In one embodiment, the retransmission criterion is that the semantic weight of the erroneous segment is a local minimum and less than or equal to 1, and the expected number of retransmissions of the erroneous segment is less than or equal to 2.

[0017] According to a second aspect of the present disclosure, a semantic-based LTP protocol optimization system is provided, the system comprising:

[0018] The encapsulation module is used to encapsulate the original file into bundle format data and send it to the LTP layer;

[0019] The splitting and recombining module is used to split and recombine the bundle format data into multiple segments based on the semantic features of the bundle format data.

[0020] The optimal segment combination method acquisition module is used to train the model under different channel conditions using the semantic data transmission system model based on the Transformer architecture, and obtain the optimal segment combination method for the corresponding scenario based on the transmission accuracy under different segment combination methods.

[0021] The transmission module is used to transmit data in the optimal segment combination. If a transmission error occurs, the erroneous segment is judged according to the retransmission criteria. If the retransmission criteria are met, semantic feature points are added to the erroneous segment and it is retransmitted. During the retransmission process, the optimal segment combination is retransmitted first.

[0022] In one embodiment, the semantic data transmission system model in the optimal segment combination acquisition module includes a semantic encoder and a semantic decoder. The semantic encoder includes a text embedding layer, a Transformer architecture, a fully connected layer, and a quantization layer. The semantic decoder includes a dequantization layer, a fully connected layer, a Transformer architecture, and a softmax layer.

[0023] In one embodiment, the acquisition of the optimal segment combination method in the optimal segment combination method acquisition module includes:

[0024] Multiple minimal semantic segments are formed based on the semantic features of bundle format data;

[0025] The ways to combine all the smallest semantic segments;

[0026] Under different channel conditions, the semantic data transmission system transmits various combinations of methods and obtains the transmission accuracy in all cases;

[0027] The optimal segment combination is the one that achieves the highest transmission accuracy under each channel condition.

[0028] In one embodiment, the retransmission criterion in the transmission module is to simultaneously satisfy the following conditions: the semantic weight of the erroneous segment is a local minimum and less than or equal to 1, and the expected number of retransmissions of the erroneous segment is less than or equal to 2.

[0029] The technical solution provided in this disclosure is a semantic-based LTP protocol optimization method and system. First, the semantic features of the original data to be transmitted are extracted and shared between the transmitting and receiving ends. Then, semantic segmentation and semantic encoding are performed based on these semantic features, and transmission and reconstruction are carried out at the semantic level. Using deep learning-based semantic communication technology, the transmitting and receiving ends share a semantic knowledge base containing all semantic features. Segmentation and combination standards are established, and the optimal combination method for different channels is obtained based on the training process. A semantic feature-based LTP transmission protocol and a semantic feature increment-based retransmission strategy are developed to replace the existing LTP transmission protocol. A semantic-based transmission quality detector is also designed to evaluate whether reliable transmission has been completed at the semantic level. This invention achieves reliable transmission with the fewest retransmissions while reducing packet loss rate, and effectively improves information timeliness.

[0030] It should be understood that the above general description and the following detailed description are exemplary and explanatory only, and are not intended to limit this disclosure. Attached Figure Description

[0031] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with the invention and, together with the description, serve to explain the principles of the invention.

[0032] Figure 1 This is a schematic diagram of the overall logic of the semantic-based LTP protocol optimization method in this embodiment of the invention;

[0033] Figure 2 This is a schematic diagram of the semantic-based LTP protocol optimization method in an embodiment of the present invention;

[0034] Figure 3 This is a schematic diagram of the semantic data transmission system structure in an embodiment of the present invention;

[0035] Figure 4 This is a schematic diagram illustrating the logic for obtaining the optimal segment semantic combination method in an embodiment of the present invention;

[0036] Figure 5 This is a schematic diagram of the SS-HARQ mechanism logic in an embodiment of the present invention;

[0037] Figure 6 This is a schematic diagram of the structure of the semantic-based LTP protocol optimization system in an embodiment of the present invention;

[0038] Figure 7 This is a schematic diagram comparing file delivery delays in a simulation experiment of this invention. Figure 1 ;

[0039] Figure 8 This is a schematic diagram comparing file delivery delays in a simulation experiment of this invention. Figure 2;

[0040] Figure 9 This is a histogram comparing the average AoI values ​​in the simulation experiment of this invention embodiment;

[0041] Figure 10 This is a comparison chart of the average AoSI curve in the simulation experiment of this invention embodiment. Detailed Implementation

[0042] The present invention will now be described in further detail with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are for illustrative purposes only and are not intended to limit the scope of the invention. Furthermore, it should be noted that, for ease of description, only the parts relevant to the present invention are shown in the drawings, not the entire structure.

[0043] Before discussing the exemplary embodiments in more detail, it should be noted that some exemplary embodiments are described as processes or methods depicted as flowcharts. Although the flowcharts describe the steps as sequential processes, many of these steps can be performed in parallel, concurrently, or simultaneously. Furthermore, the order of the steps can be rearranged. The process can be terminated when its operation is complete, but may also have additional steps not included in the figures. The process can correspond to a method, function, procedure, subroutine, subroutine, etc.

[0044] This invention provides the following embodiments for a semantic-based LTP protocol optimization method and system:

[0045] Example 1 illustrates a semantic-based LTP protocol optimization method. See [link to example]. Figure 1 The diagram below illustrates the logic of the semantic-based LTP protocol optimization method in this embodiment. During data transmission from one node to another, the original file is first encapsulated into a bundle format and sent to the LTP layer. Based on the semantic features of the data, the bundle is split and reassembled into multiple segments. The optimal segment combination for the corresponding scenario is obtained based on the transmission accuracy of different segment combinations under different channel conditions during training. In actual transmission, data is transmitted using this optimal combination. During retransmission, the optimal combination itself is retransmitted first. If further retransmission is required, a semantic incremental retransmission mechanism is used based on the actual conditions.

[0046] For details, see Figure 2 The semantic-based LTP protocol optimization method includes the following steps:

[0047] S21. Encapsulate the original file into bundle format data and send it to the LTP layer;

[0048] S22. Based on the semantic features of the bundle format data, split and reorganize it into multiple segments;

[0049] S23. Use the semantic data transmission system model based on the Transformer architecture to train the model under different channel conditions, and obtain the optimal segment combination method under the corresponding scenario based on the transmission accuracy under different segment combination methods.

[0050] S24. Data is actually transmitted using the optimal segment combination. If a transmission error occurs, the erroneous segment is retransmitted according to the retransmission criteria. If the retransmission criteria are met, semantic feature points are added to the erroneous segment before retransmission. During the retransmission process, the optimal segment combination is retransmitted first.

[0051] See Figure 3 The semantic data transmission system model includes a semantic encoder and a semantic decoder. The semantic encoder includes a text embedding layer, a Transformer architecture, a fully connected layer, and a quantization layer. The semantic decoder includes a dequantization layer, a fully connected layer, a Transformer architecture, and a softmax layer. The semantic encoder and semantic decoder are symmetrical structures. The weights of each parameter obtained during the training process are stored in the encoder and decoder and applied in actual transmission.

[0052] See Figure 4 Obtaining the optimal segment combination includes:

[0053] Multiple minimal semantic segments are formed based on the semantic features of bundle format data;

[0054] The ways to combine all the smallest semantic segments;

[0055] Under different channel conditions, the semantic data transmission system transmits various combinations of methods and obtains the transmission accuracy in all cases;

[0056] The optimal segment combination is the one that achieves the highest transmission accuracy under each channel condition.

[0057] The semantic incremental retransmission mechanism SS-HARQ in the embodiment is described in [reference needed]. Figure 5 Specifically, it includes:

[0058] First, obtain the optimal segment combination method under the current channel conditions.

[0059] Step 1: The first round of transmission determines the optimal segment combination under the current channel conditions;

[0060] Step 2: Transmit various combinations of methods through a semantic data transmission system under different channel conditions and obtain the transmission accuracy under all conditions;

[0061] Step 3: If a transmission error occurs, determine the retransmission standard for the erroneous block and whether to use the SS-HARQ mechanism.

[0062] Step 4: If using SS-HARQ, add semantic feature points and then retransmit.

[0063] The retransmission criteria are as follows: the semantic weight of the erroneous segment is a local minimum and less than or equal to 1, and the expected number of retransmissions of the erroneous segment is less than or equal to 2.

[0064] Another embodiment illustrates a semantic-based LTP protocol optimization system, see [link to documentation]. Figure 6 The system 600 includes:

[0065] The encapsulation module 610 is used to encapsulate the original file into bundle format data and send it to the LTP layer;

[0066] The splitting and recombining module 620 is used to split and recombine the bundle format data into multiple segments according to the semantic features of the bundle format data.

[0067] The optimal segment combination method acquisition module 630 is used to train the model under different channel conditions using the semantic data transmission system model based on the Transformer architecture, and obtain the optimal segment combination method in the corresponding scenario based on the transmission accuracy under different segment combination methods.

[0068] The transmission module 640 is used to transmit data in the optimal segment combination. If a transmission error occurs, the erroneous segment is judged according to the retransmission criteria. If the retransmission criteria are met, semantic feature points are added to the erroneous segment and then it is retransmitted. During the retransmission process, the optimal segment combination is retransmitted first.

[0069] The semantic data transmission system model in the optimal segment combination acquisition module 630 includes a semantic encoder and a semantic decoder. The semantic encoder includes a text embedding layer, a Transformer architecture, a fully connected layer, and a quantization layer. The semantic decoder includes a dequantization layer, a fully connected layer, a Transformer architecture, and a softmax layer.

[0070] The acquisition of the optimal segment combination method described in module 630 includes:

[0071] Multiple minimal semantic segments are formed based on the semantic features of bundle format data;

[0072] The ways to combine all the smallest semantic segments;

[0073] Under different channel conditions, the semantic data transmission system transmits various combinations of methods and obtains the transmission accuracy in all cases;

[0074] The optimal segment combination is the one that achieves the highest transmission accuracy under each channel condition.

[0075] The retransmission criterion in the transmission module 640 is that the semantic weight of the erroneous segment is a local minimum and less than or equal to 1, and the expected number of retransmissions of the erroneous segment is less than or equal to 2.

[0076] In addition to the above module, system 600 may also include other components; however, since these components are not relevant to the content of this disclosure, their illustrations and descriptions are omitted here.

[0077] Other specific working processes of the semantic-based LTP protocol optimization system 600 are described in the above-described embodiment of the semantic-based LTP protocol optimization method, and will not be repeated here.

[0078] To further demonstrate the effectiveness of the invention, the inventors conducted simulation experiments, the simulation performance indicators of which included:

[0079] File delivery delay: The total delay required for a fixed-size file to be reliably received.

[0080] Average Age of Information (AoI): In a data transmission system, the average freshness of a series of transmitted data;

[0081]

[0082] In the above formula, t represents the current time, and u(t) represents the time when the latest updated data was generated.

[0083] Average Age of Semantic Information (AoSI): AoSI is a new metric used to evaluate the timeliness contribution of different semantic blocks in the document delivery process at the semantic level of the raw data.

[0084]

[0085] In the above formula, λ k Represents the k-th data block sk The semantic weight value, and the semantic quality value Related to the total number of semantic blocks M:

[0086]

[0087] Semantic quality is obtained by the following formula:

[0088]

[0089] in, This indicates the result of comparison between the sending and receiving ends (whether reliable transmission has been completed). If they match, it is 1; otherwise, it is 0.

[0090] In the simulation verification, the inventors considered the performance comparison in the high-orbit satellite scenario and the Mars exploration scenario, respectively. The specific indicators included file delivery delay, average information age and average semantic information age. The comparison schemes included (1) comparison of encoding schemes: traditional bit-level encoding and semantic-level encoding proposed in this invention; (2) comparison of retransmission standards: CRC check that requires complete reliability for all data, Sim32 detector that requires consistent semantic reliability for all data, and semantic quality detector SQD proposed in this invention that has different semantic reliability requirements for data with different semantic importance; (3) comparison of retransmission mechanisms: standard ARQ mechanism, semantic-level HARQ mechanism SCHARQ, and semantic incremental retransmission mechanism SS-HARQ.

[0091] First, the inventors studied the performance of document delivery latency. For example... Figure 7 , Figure 8 The figure shows a comparison of file delivery latency under different transmission protocols and standards. When transmitting files with the same content and size, the traditional bit-level encoding scheme using CRC checksum has the highest delivery latency in both environments. Since semantic communication can recover the original information using stored semantic features under high channel error rates, the delivery latency is significantly reduced after using semantic transmission instead of bit-level encoding. After replacing CRC with the Sim32 semantic detector, the retransmission standard is relaxed to a certain extent, and some data with high semantic reliability no longer needs to be retransmitted, thus further reducing the file delivery latency. On this basis, after introducing semantic feature increments during the retransmission process, the added redundant semantic information makes the semantic features more complete. Therefore, the increased number of semantic features contained in the retransmitted data leads to an improvement in recovery accuracy. Furthermore, the Semantic Quality Detector (SQD) can stratify semantic data according to the degree of semantic importance and set different levels of retransmission conditions, further reducing unnecessary retransmissions. Thus, the method of this invention has the lowest file delivery latency.

[0092] like Figure 9The figure shows a performance comparison of the average AoI for four schemes. Since AoI is related to the interval between data block transmissions, γ in the figure represents the interval time coefficient for sequentially transmitting semantic data blocks at the source node; a larger γ indicates a longer interval. To observe and compare the processing behavior of various nodes in real-world scenarios, the range of γ is set to [0.2, 1.0]. Figure 8 As can be seen, under all transmission intervals, the average AoI of the four schemes gradually decreases, while the scheme of this invention, which utilizes semantic transmission, SS-HARQ, and SQD, always has the smallest average AoI.

[0093] like Figure 10 The figure shows a performance comparison of the average AoSI for the three schemes. Because semantic-level metrics are considered, a comparison with traditional bit-level transmission schemes is no longer made; only the performance of semantic schemes is analyzed. Compared to AoI, the AoSI value is also related to the contribution of each semantic block to the overall semantic recovery of the information. It can be seen that the scheme proposed in this invention consistently has the lowest average AoSI, but the performance improvement becomes less significant as the transmission time interval γ increases. This is because the waiting latency at the node increases compared to the processing latency, and the proportion of the transmission latency metric optimized in this invention becomes smaller overall, thus resulting in a diminishing improvement.

[0094] Based on the technical solutions provided in the above embodiments, the semantic-based LTP protocol optimization method and system first extracts the semantic features of the original data to be transmitted and shares them between the transmitting and receiving ends. Then, semantic segmentation and semantic encoding are performed based on these semantic features, and transmission and reconstruction are carried out at the semantic level. Using deep learning-based semantic communication technology, the transmitting and receiving ends share a semantic knowledge base containing all semantic features, and segmentation and combination standards are established. The optimal combination method for different channels is obtained based on the training process. A semantic feature-based LTP transmission protocol and a semantic feature increment-based retransmission strategy are developed to replace the existing LTP transmission protocol. A semantic-based transmission quality detector is also designed to evaluate whether reliable transmission has been completed at the semantic level. This invention achieves reliable transmission with the fewest retransmissions while reducing packet loss rate, and effectively improves information timeliness.

[0095] In this document, the terms “comprising,” “including,” or any other variations thereof are intended to cover non-exclusive inclusion, such that a step or method that comprises a list of elements includes not only those elements but also other elements not expressly listed or inherent to such a step or method.

[0096] The above description, in conjunction with specific preferred embodiments, provides a further detailed explanation of the present invention. It should not be construed that the specific implementation of the present invention is limited to these descriptions. For those skilled in the art, various simple deductions or substitutions can be made without departing from the concept of the present invention, and all such modifications and substitutions should be considered within the scope of protection of the present invention.

Claims

1. A semantic-based LTP protocol optimization method, characterized in that, The method includes the following steps: The original file is encapsulated into bundle format data and sent to the LTP layer; Based on the semantic features of the bundle format data, it is split and reorganized into multiple segments; The model of a semantic data transmission system based on the Transformer architecture is used to train the model under different channel conditions. Based on the transmission accuracy under different segment combination methods, the optimal segment combination method is obtained for the corresponding scenario. The data is actually transmitted using the optimal segment combination. If a transmission error occurs, the erroneous segment is judged according to the retransmission criteria. If the retransmission criteria are met, semantic feature points are added to the erroneous segment before retransmission. During retransmission, the optimal segment combination is retransmitted first.

2. The semantic-based LTP protocol optimization method according to claim 1, characterized in that, The semantic data transmission system model includes a semantic encoder and a semantic decoder. The semantic encoder includes a text embedding layer, a Transformer architecture, a fully connected layer, and a quantization layer. The semantic decoder includes a dequantization layer, a fully connected layer, a Transformer architecture, and a softmax layer.

3. The semantic-based LTP protocol optimization method according to claim 1, characterized in that, Obtaining the optimal segment combination method includes: Multiple minimal semantic segments are formed based on the semantic features of bundle format data; The ways to combine all the smallest semantic segments; Under different channel conditions, the semantic data transmission system transmits various combinations of methods and obtains the transmission accuracy in all cases; The optimal segment combination is the one that achieves the highest transmission accuracy under each channel condition.

4. The semantic-based LTP protocol optimization method according to claim 1, characterized in that, The retransmission criterion is that the semantic weight of the erroneous segment is a local minimum and less than or equal to 1, and the expected number of retransmissions of the erroneous segment is less than or equal to 2.

5. A semantic-based LTP protocol optimization system, characterized in that, The system includes: The encapsulation module is used to encapsulate the original file into bundle format data and send it to the LTP layer; The splitting and recombining module is used to split and recombine the bundle format data into multiple segments based on the semantic features of the bundle format data. The optimal segment combination method acquisition module is used to train the model under different channel conditions using the semantic data transmission system model based on the Transformer architecture, and obtain the optimal segment combination method for the corresponding scenario based on the transmission accuracy under different segment combination methods. The transmission module is used to transmit data in the optimal segment combination. If a transmission error occurs, the erroneous segment is judged according to the retransmission criteria. If the retransmission criteria are met, semantic feature points are added to the erroneous segment and it is retransmitted. During the retransmission process, the optimal segment combination is retransmitted first.

6. The semantic-based LTP protocol optimization system according to claim 5, characterized in that, The semantic data transmission system model in the optimal segment combination acquisition module includes a semantic encoder and a semantic decoder. The semantic encoder includes a text embedding layer, a Transformer architecture, a fully connected layer, and a quantization layer. The semantic decoder includes a dequantization layer, a fully connected layer, a Transformer architecture, and a softmax layer.

7. The semantic-based LTP protocol optimization system according to claim 5, characterized in that, The acquisition of the optimal segment combination method in the optimal segment combination method acquisition module includes: Multiple minimal semantic segments are formed based on the semantic features of bundle format data; The ways to combine all the smallest semantic segments; Under different channel conditions, the semantic data transmission system transmits various combinations of methods and obtains the transmission accuracy in all cases; The optimal segment combination is the one that achieves the highest transmission accuracy under each channel condition.

8. The semantic-based LTP protocol optimization system according to claim 5, characterized in that, The retransmission criterion in the transmission module is that the semantic weight of the erroneous segment is a local minimum and less than or equal to 1, and the expected number of retransmissions of the erroneous segment is less than or equal to 2.

Images