Low earth orbit satellite communication AMC optimization method based on AI-rule fusion decision

Abstract

The invention relates to the technical field of wireless communication, and discloses a low earth orbit satellite communication AMC optimization method based on AI-rule fusion decision, which comprises the following steps: acquiring a measured value of a low earth orbit satellite load scheduling parameter; constructing a low earth orbit satellite communication adaptive modulation coding optimization model, and predicting an initial modulation coding scheduling result according to the measured value of the low earth orbit satellite load scheduling parameter; and generating a modulation and coding scheduling optimization result by adopting a rule-based adaptive modulation and coding optimization method, calculating a correction coefficient according to the initial modulation and coding scheduling result and the modulation and coding scheduling optimization result, and correcting the initial modulation and coding scheduling result by utilizing the correction coefficient to obtain a final modulation and coding scheduling result. According to the invention, the most suitable modulation and coding mode can be dynamically and adaptively selected, and the maximum system throughput rate can be realized as far as possible under the condition that the error rate is controlled in a relatively low range.

Description

Technical Field

[0001] This invention relates to the field of wireless communication technology, and specifically to an AI-rule fusion decision-making-based AMC optimization method for low-Earth orbit satellite communication. Background Technology

[0002] Low Earth Orbit (LEO) satellite communication refers to the provision of global communication services using satellite constellations at orbital altitudes of 500-2000 km. In recent years, with the rapid development of commercial space technologies (such as SpaceX Starlink, OneWeb, Telesat, etc.), LEO satellite communication has become an important component of 5G / 6G networks. Compared to traditional satellite communication (such as GEO / MEO) and terrestrial communication networks, their respective characteristics are shown in Table 1. Table 1 Comparison of Various Communication Technologies characteristic Low Earth Orbit (LEO) satellites Geostationary orbit (GEO) satellite terrestrial 5G / fiber Delay 20-50 ms 250 ms 5-10 ms (fiber optic) Coverage Globally (including remote areas) Wide area fixed coverage Urban / Densely Populated Areas Deployment costs Medium (requires large-scale constellations) High (expensive per star) High (base station / fiber optic cable deployment) Damage resistance Extremely high (distributed constellation) Medium (single point of failure risk) Low (dependence on ground facilities) Typical applications Global Internet, Emergency Communications Broadcasting and television, maritime communications Urban high-speed communication Low-Earth orbit (LEO) satellite communication, with its advantages of low latency, global coverage, and high spectral efficiency, holds the promise of achieving integrated air-space-ground-sea communication. Currently, LEO satellite communication, with its unique orbital characteristics and technical design, is reshaping the global communication architecture. In my country, LEO satellite communication technology has received significant attention. In recent years, my country has accelerated its development in this field, gradually building an independent and controllable satellite internet system through policy support, the rise of commercial aerospace, and technological innovation.

[0003] Low Earth orbit satellite communication technology also faces many challenges in its development, and a suitable adaptive modulation and coding (AMC) scheme is one of them.

[0004] AMC is a core technology in wireless communication systems. By dynamically adjusting the modulation scheme and coding rate, the system can optimize transmission performance according to real-time channel conditions. That is, it can achieve the maximum throughput while ensuring a low bit error rate. In simple terms, the core principle is to use high-order modulation and high code rate when the channel quality is good to maximize throughput, and to reduce to low-order modulation and low code rate when the channel quality is poor to ensure transmission reliability.

[0005] A good AMC solution can balance spectral efficiency and transmission reliability to achieve good system throughput. Conversely, a poor AMC solution will lead to a high bit error rate or severely reduce spectral efficiency during information transmission, ultimately resulting in a very low system throughput.

[0006] Compared to terrestrial base stations, low-Earth orbit (LEO) satellite communication suffers from challenges such as a highly dynamic channel environment, long propagation delays, large-scale fading and obstruction of the satellite-to-ground link, Doppler shift, and phase noise. These issues significantly reduce the performance of traditional AMC (Accelerated Channel Management) methods for LEO satellite communication. This is primarily because while LEO satellite transmission delays are lower than those of geostationary satellites, they are still much higher than those on the ground. Furthermore, the high-speed movement of satellites causes rapid changes in the channel state between the user terminal and the satellite. Traditional AMC relies on periodic CSI (Channel State Information) feedback and bit error rate-based processing. In LEO satellite communication scenarios, the channel quality reported by the UE reaches the satellite payload after a significant transmission delay. Since the satellite channel changes faster than the feedback period, this channel quality may be outdated. Similarly, the historical bit error statistics obtained in this situation are also outdated and cannot provide accurate reference value for scheduling AMC execution at the current moment. Consequently, the satellite payload's MCS (Multi-Channel System Selection) cannot adapt to the actual situation, severely impacting system throughput. Summary of the Invention

[0007] To address the aforementioned shortcomings in existing technologies, this invention provides an AI-rule fusion decision-making method for low-Earth orbit satellite communication AMC optimization, aiming to dynamically and adaptively select the most suitable modulation and coding scheme for low-Earth orbit satellite payloads during scheduling, thereby maximizing system throughput while keeping the bit error rate within a low range.

[0008] To achieve the above-mentioned objectives, the technical solution adopted by this invention is as follows: A method for optimizing the AMC (Advanced Mobility Management) of low-Earth orbit satellite communications based on AI-rule fusion decision-making includes the following steps: Obtain measured values ​​of low-Earth orbit satellite payload scheduling parameters; An adaptive modulation and coding optimization model for low-Earth orbit satellite communication is constructed, and the initial modulation and coding scheduling results are predicted based on the measured values ​​of low-Earth orbit satellite payload scheduling parameters. A rule-based adaptive modulation and coding optimization method is used to generate modulation and coding scheduling optimization results. Correction coefficients are calculated based on the initial modulation and coding scheduling results and the optimization results. The initial modulation and coding scheduling results are then corrected using the correction coefficients to obtain the final modulation and coding scheduling results.

[0009] Furthermore, the low-orbit satellite payload scheduling parameters include signal-to-interference-plus-noise ratio (SINR), received signal strength indication, timing advance, and scheduling verification results.

[0010] Furthermore, the modulation and coding scheduling optimization results generated using a rule-based adaptive modulation and coding optimization method include: The scheduling index parameters are adjusted based on the scheduling verification results of uplink and downlink during the scheduling process. If the scheduling verification result passes, the scheduling index parameters are increased according to the set step size. If the scheduling verification result fails, the scheduling index parameters are decreased according to the set target bit error rate.

[0011] Furthermore, the calculation method for adjusting the scheduling index parameters according to the set target bit error rate is as follows: in, To reduce the adjustment amount of the scheduling index parameters, To set the step size, The target bit error rate is set.

[0012] Furthermore, when a valid initial hybrid automatic repeat request for the current modulation and coding update is received, adaptive modulation and coding optimization is performed. The current modulation and coding scheme update is valid. The initial hybrid automatic repeat request feedback is satisfied: When the Hybrid Automatic Repeat Request returns an error, the current modulation and coding scheme must be less than or equal to the modulation and coding scheme currently being used by the base station. When the hybrid automatic repeat request feedback is correct, the current modulation and coding must be greater than or equal to the modulation and coding currently being used by the base station. When the current modulation and coding is equal to the modulation and coding currently used by the base station minus 1 and the current scheduling index parameter is between the upper and lower boundaries of the index, the hybrid automatic repeat request feedback is valid regardless of whether it is correct or not.

[0013] Furthermore, the upper boundary of the index is the sum of half of the scheduling index parameter corresponding to the modulation and coding scheme currently being used by the base station and half of the scheduling index parameter corresponding to the modulation and coding scheme currently being used by the base station minus 1, and the lower boundary of the index is the scheduling index parameter corresponding to the modulation and coding scheme currently being used by the base station.

[0014] Furthermore, the probability of passing the hybrid automatic repeat request verification corresponding to each modulation and coding scheme within a set number of scheduling attempts is dynamically counted. Based on the probability of passing the hybrid automatic repeat request verification corresponding to the modulation and coding scheme currently being used by the base station, the probability of passing the hybrid automatic repeat request verification corresponding to the modulation and coding scheme currently being used by the base station minus 1 is used to determine the ratio of modulation and coding schemes used when scheduling uplink and downlink.

[0015] Furthermore, the modulation and coding ratio used for scheduling uplink and downlink is calculated as follows: in, For the scheduling ratio, For the set target bit error rate, This refers to the scheduling index parameter corresponding to the modulation and coding scheme currently being used by the base station. The scheduling index parameter is the modulation and coding scheme currently being used by the base station minus 1.

[0016] Furthermore, the adaptive modulation and coding optimization model for low-Earth orbit satellite communication adopts a structure of several stacked Transformer encoder layers and maps them to different modulation and coding schemes through fully connected layers.

[0017] Furthermore, the low-orbit satellite communication adaptive modulation and coding optimization model sets a time period and a data collection period during model training. Modulation and coding scheduling is performed within the time period, and training data is collected from the beginning of the time period to the data collection period. Model training is performed from the data collection period to the end of the time period.

[0018] The present invention has the following beneficial effects: This invention establishes an AI-rule fusion decision-making-based adaptive modulation and coding optimization for low-Earth orbit satellite communication, which can predict the probability of passing the MCS scheduling verification result based on the measurement. At the same time, it establishes a rule-based MCS scheduling method and uses the rule-based scheduling result to correct the AI ​​output. This model takes into account both bit error rate and spectral efficiency, and can dynamically and adaptively select the most suitable modulation and coding method, so as to achieve the maximum system throughput while keeping the bit error rate within a low range. Attached Figure Description

[0019] Figure 1 This is a flowchart illustrating an AI-rule fusion decision-making-based AMC optimization method for low-Earth orbit satellite communication. Figure 2 This is a schematic diagram of a neural network; Figure 3 This is a diagram illustrating network updates. Detailed Implementation

[0020] The specific embodiments of the present invention are described below to enable those skilled in the art to understand the present invention. However, it should be understood that the present invention is not limited to the scope of the specific embodiments. For those skilled in the art, various changes are obvious as long as they are within the spirit and scope of the present invention as defined and determined by the appended claims. All inventions utilizing the concept of the present invention are protected.

[0021] like Figure 1 As shown in the figure, an AI-rule fusion decision-making method for low-Earth orbit satellite communication AMC optimization provided by an embodiment of the present invention includes the following steps S1 to S3: S1. Obtain the measured values ​​of low-Earth orbit satellite payload scheduling parameters; In an optional embodiment of the present invention, the measured values of the low-earth orbit satellite payload scheduling parameters obtained in step S1 include signal-to-interference-plus-noise ratio, received signal strength indication, timing advance, and scheduling verification result.

[0022] S2. Construct an adaptive modulation and coding optimization model for low-earth orbit satellite communication, and predict the initial modulation and coding scheduling result according to the measured values of the low-earth orbit satellite payload scheduling parameters; In an optional embodiment of the present invention, step S2 uses a neural network to construct an adaptive modulation and coding optimization model for low-earth orbit satellite communication. The neural network adopts a structure of several stacked Transformer encoder layers and is mapped to different modulation and codings through a fully connected layer.

[0023] The neural network uses an online training method to update parameters. The specific method is as follows: Set a time period TP and a data collection time period TC (TC < TP). Within the first TP, use the traditional MCS scheduling method for scheduling. At the same time, within the time period (0, TC), collect training data. The training data includes and the different MCS scheduling verification results at the corresponding moments. Within the time period (TC, TP), use the gradient descent algorithm to train the neural network. At this time, the initial weights of the neural network are randomly initialized weights . After training is completed, the weights of the neural network are . Within the time period (TP, 2TP), use the MCS selection method in S1 for scheduling. At the same time, within the time period (TP, TP + TC), re-collect training data, and within the time period (TP + TC, 2TP), re-train the neural network. At this time, the initial weights of the neural network are . In subsequent time periods, repeat the operations of (TP, 2TP).

[0024] The label generation method during data collection is as follows: Within the time period (0, TC), according to the MCS selected by the traditional method index0 , if the CRC verification result passes, the label corresponding to the MCS less than or equal to MCS index0 is 1, and the label corresponding to the MCS greater than MCS index0 is 0. If the CRC verification result fails, the label corresponding to the MCS less than MCS index0 is 1, and the label corresponding to the MCS greater than or equal to MCS index0 is 0.

[0025] Within the time period (TP, TP + TC), the neural network outputs the probability of each MCS passing CRC, and then selects the optimal MCS according to the MCS selection method in S1 index1The corresponding neural network output probability is p1. If the CRC check result passes, the probability is within the range [max(0.1, p1- p1]. If the label corresponding to the MCS in the interval (1) is 1, then the probability is in the range (0, max(0.1, p1-). The label corresponding to the MCS within the interval is 0. If the CRC check fails, the probability is in the range [min(0.9, p1+ ] ]. ), 1) The label corresponding to the MCS in the interval is 1, in (0, min(0.9, p1+ The label corresponding to the MCS within the interval is 0. Wherein, and These are pre-set parameters.

[0026] In this embodiment, the time period (TP) is set to 20 minutes, and the total time period (TC) is set to 10 minutes. During the first TP, scheduling is performed using the traditional MCS scheduling method. Simultaneously, training data is collected during the (0, TC) time period. The training data includes... The MCS scheduling verification results are different at the corresponding time. In this embodiment, k is set to 5. The label generation method when collecting data is as follows: within the (0, TC) time period, the MCS selected according to the traditional method is used. index0 If the CRC check result passes, then it is less than or equal to MCS. index0 The label corresponding to the MCS is 1, which is greater than the MCS. index0 The label corresponding to the MCS is 0. If the CRC check fails, it is less than the MCS. index0 The label corresponding to the MCS is 1, which is greater than or equal to the MCS. index0 The label corresponding to the MCS is 0.

[0027] Building such Figure 2 The neural network shown has M set to 2. The network's initial weights are initialized using a random method, and the gradient descent algorithm is used to update the network weights.

[0028] like Figure 3 As shown, this embodiment uses a neural network for scheduling while simultaneously collecting new training data: MCS scheduling is performed using a trained neural network and an optimized model, while new training data is collected. In the new training data, the labels are generated as follows: within the time period (TP, TP+TC), the neural network outputs the probability of each MCS passing the CRC check, and then selects the optimal MCS_index1 according to the MCS selection method, with the corresponding neural network output probability p1. If the CRC check passes, the probability is within the range [max(0.1, p1-...]. ), 1) If the label corresponding to the MCS in the interval is 1, then the probability is in the range (0, max(0.1, p1- The label corresponding to the MCS within the interval is 0. If the CRC check fails, the probability is in the range [min(0.9, p1+ ] ]. ), 1) The label corresponding to the MCS in the interval is 1, in (0, min(0.9, p1+ The label corresponding to the MCS within the interval is 0. In this embodiment, and Set it to 0.2.

[0029] S3. A rule-based adaptive modulation and coding optimization method is used to generate modulation and coding scheduling optimization results. Correction coefficients are calculated based on the initial modulation and coding scheduling results and the optimization results. The initial modulation and coding scheduling results are corrected using the correction coefficients to obtain the final modulation and coding scheduling results.

[0030] In an optional embodiment of the present invention, step S3, which uses a rule-based adaptive modulation and coding optimization method to generate modulation and coding scheduling optimization results, includes: The scheduling index parameters are adjusted based on the scheduling verification results of uplink and downlink during the scheduling process. If the scheduling verification result passes, the scheduling index parameters are increased according to the set step size. If the scheduling verification result fails, the scheduling index parameters are decreased according to the set target bit error rate.

[0031] The calculation method for adjusting the scheduling index parameters based on the set target bit error rate is as follows: in, To reduce the adjustment amount of the scheduling index parameters, To set the step size, The target bit error rate is set.

[0032] This embodiment establishes a rule-based MCS scheduling process, specifically including setting the intermediate parameter cqi. state During the scheduling process, the scheduling verification results of uplink and downlink are used as a basis. Adjust cqi_state. Verification passed, cqi state Upgrade; verification failed, cqi state Adjust downwards. The values ​​for both upward and downward adjustments are based on the set target bit error rate (target). bler Settings. cqi state The adjustment step size should be less than or equal to mapcqi mcs In the table, cqi state Half of the minimum value of adjacent intervals.

[0033] In this embodiment, cqi state The adjustment step size is set to 0.15, with an upward adjustment of 0.15 and a downward adjustment of 0.15*(1-target). bler ) / target bler .

[0034] Finally, through CQI state Mapping table with MCS (mapcqi) mcs The table retrieves the MCS value to be used in the next scheduling iteration. mcs cqi state Mapping table with MCS (mapcqi) mcs The table is shown in Table 2.

[0035] Table 2cqi state Mapping table with MCS <![CDATA[cqi state ]]> MCS index 1 0 2 0 2.5 1 3 2 3.5 3 4 4 4.5 5 4 6 5.5 7 6 8 6.5 9 7 10 7.5 11 8 12 8.5 13 9 14 9.5 15 10 16 10.5 17 11 18 11.5 19 12 20 12.5 21 13 22 13.5 23 14 24 14.3 25 14.7 26 15 27 In an optional embodiment of the present invention, when step S3 receives a feedback of an initial hybrid automatic repeat request that is valid for the current modulation and coding update, adaptive modulation and coding optimization is performed. The current modulation and coding scheme update is valid. The initial hybrid automatic repeat request feedback is satisfied: When the Hybrid Automatic Repeat Request returns an error, the current modulation and coding scheme must be less than or equal to the modulation and coding scheme currently being used by the base station. When the hybrid automatic repeat request feedback is correct, the current modulation and coding must be greater than or equal to the modulation and coding currently being used by the base station. When the current modulation and coding is equal to the modulation and coding currently used by the base station minus 1 and the current scheduling index parameter is between the upper and lower boundaries of the index, the hybrid automatic repeat request feedback is valid regardless of whether it is correct or not.

[0036] The upper boundary of the index is the sum of half of the scheduling index parameter corresponding to the modulation and coding scheme currently being used by the base station and half of the scheduling index parameter corresponding to the modulation and coding scheme currently being used by the base station minus 1. The lower boundary of the index is the scheduling index parameter corresponding to the modulation and coding scheme currently being used by the base station.

[0037] This embodiment establishes an update mechanism for MCS scheduling. When a valid initial HARQ feedback for the current MCS update is received, CQI is performed. state Adjustments and updates to MCS.

[0038] The definition of valid HARQ feedback for current MCS updates is as follows: When HarQ returns an error, the scheduled condition must be met. mcs <=currmcs ; When the harq response is correct, the sched requirement must be met. mcs >= curr mcs ; When sched mcs equal to curr mcs-1 And currently CQI state Located in (mapcqi) mcs table[curr mcs +mapcqi mcs table[curr mcs-1 ] / 2) and mapcqi mcs table[curr mcs When the condition is between ], Harq feedback is valid regardless of whether it is correct or not. (sched) mcs For the current HarQ feedback, the corresponding MCS is curr. mcs This represents the MCS currently being used by the base station.

[0039] In an optional embodiment of the present invention, step S3 dynamically calculates the probability of passing the hybrid automatic repeat request verification corresponding to each modulation and coding within a set number of scheduling operations, and determines the ratio of modulation and coding used when scheduling uplink and downlink based on the probability of passing the hybrid automatic repeat request verification corresponding to the modulation and coding currently being used by the base station and the probability of passing the hybrid automatic repeat request verification corresponding to the modulation and coding currently being used by the base station minus 1.

[0040] To prevent excessively high bit error rates and their persistence across multiple slots when using certain MCSs, this embodiment establishes an MCS scheduling balancing mechanism: dynamically calculating the ACK rate of the corresponding HARQ in the most recent 100 scheduling iterations for each MCS, and based on the curr... mcs ACK rate and curr mcs -1 The ACK rate determines the proportion of MCS used when scheduling uplink and downlink, and denoted as a is the proportion of the current MCS used.

[0041] The calculation method for the scheduling ratios of MCS and MCS-1 is as follows: in, For the scheduling ratio, For the set target bit error rate, This refers to the scheduling index parameter corresponding to the modulation and coding scheme currently being used by the base station. The scheduling index parameter is the modulation and coding scheme currently being used by the base station minus 1.

[0042] In an optional embodiment of the present invention, step S3 ultimately establishes an AI-rule fusion MCS scheduling optimization model, expressed as: in, This represents the optimized MCS value. SINR (Signal-to-Interference-plus-Noise Ratio) represents the signal-to-interference-plus-noise ratio, RSSI (Received Signal Strength Indicator) represents the received signal strength indication, and TA (Timing Advance) represents the timing advance. This indicates the scheduling verification result (CRC for uplink, HARQ ACK for downlink). This represents the measured values ​​of each parameter within the interval from time tk to time t. This represents the probability that different MCS scheduling verification results will pass based on the current measurement. This indicates the spectral efficiency corresponding to different MCS.

[0043] in The calculation method is as follows: in, This represents the sigmoid function, which maps the output of function f to the interval (0,1) as a probability output.

[0044] This represents the correction factor, used to correct the AI ​​output. The calculation formula is: Where MCS represents the different MCS values ​​that can be scheduled. This represents the rule-based MCS scheduling value determined by S1-S3. It is a moderating factor used to adjust the weighting of the influence of AI and rules on the final result. In this embodiment, The value is 0.5.

[0045] The values ​​are shown in Table 3: Table 3 Value Lookup Table This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0046] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0047] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0048] Specific embodiments have been used to illustrate the principles and implementation methods of this invention. The descriptions of the embodiments above are only for the purpose of helping to understand the method and core ideas of this invention. At the same time, for those skilled in the art, there will be changes in the specific implementation methods and application scope based on the ideas of this invention. Therefore, the content of this specification should not be construed as a limitation of this invention.

[0049] Those skilled in the art will recognize that the embodiments described herein are intended to help the reader understand the principles of the invention, and should be understood that the scope of protection of the invention is not limited to such specific statements and embodiments. Those skilled in the art can make various other specific modifications and combinations based on the technical teachings disclosed in this invention without departing from the spirit of the invention, and these modifications and combinations are still within the scope of protection of this invention.

Claims

1. An AI-rule fusion decision-based low-orbit satellite communication AMC optimization method, characterized in that, The method comprises the following steps: Obtaining a measurement value of a low-orbit satellite payload scheduling parameter; Building a low-orbit satellite communication adaptive modulation and coding optimization model, and predicting an initial modulation and coding scheduling result according to the measurement value of the low-orbit satellite payload scheduling parameter; Generating a modulation and coding scheduling optimization result by using a rule-based adaptive modulation and coding optimization method, calculating a correction coefficient according to the initial modulation and coding scheduling result and the modulation and coding scheduling optimization result, and correcting the initial modulation and coding scheduling result by using the correction coefficient to obtain a final modulation and coding scheduling result.

2. The low-orbit satellite communication AMC optimization method based on AI-rule fusion decision according to claim 1, characterized in that, The low-orbit satellite payload scheduling parameter comprises a signal-to-interference-and-noise ratio, a received signal strength indication, a time advance, and a scheduling check result.

3. The AMC optimization method for low-orbit satellite communication based on AI-rule fusion decision according to claim 1, characterized in that, Generating the modulation and coding scheduling optimization result by using the rule-based adaptive modulation and coding optimization method comprises: Judging whether to adjust a scheduling index parameter according to scheduling check results of uplink and downlink in a scheduling process; if the scheduling check results pass, increasing the scheduling index parameter according to a set step; and if the scheduling check results do not pass, decreasing the scheduling index parameter according to a set target bit error rate.

4. The AMC optimization method for low-orbit satellite communication based on AI-rule fusion decision according to claim 3, characterized in that, The adjustment amount calculation mode of the scheduling index parameter according to the set target bit error rate is: wherein, is an adjustment amount for down-regulating the scheduling index parameter, is a set step size, is a set target bit error rate.

5. The AMC optimization method for low-orbit satellite communication based on AI-rule fusion decision according to claim 1, characterized in that, When a first transmission hybrid automatic repeat request feedback for a current modulation and coding update is received, adaptive modulation and coding optimization is performed; The first transmission hybrid automatic repeat request feedback for the current modulation and coding update satisfies: When the hybrid automatic repeat request feedback is incorrect, the current modulation and coding needs to be less than or equal to a modulation and coding currently used by a base station; When the hybrid automatic repeat request feedback is correct, the current modulation and coding needs to be greater than or equal to the modulation and coding currently used by the base station; When the current modulation and coding is equal to the modulation and coding currently used by the base station minus 1 and the current scheduling index parameter is between an index upper boundary and an index lower boundary, the hybrid automatic repeat request feedback is valid regardless of whether it is correct or incorrect.

6. The AMC optimization method for low-orbit satellite communication based on AI-rule fusion decision according to claim 5, characterized in that, The index upper boundary is a sum of one half of a scheduling index parameter corresponding to the modulation and coding currently used by the base station and a scheduling index parameter corresponding to the modulation and coding currently used by the base station minus 1, and the index lower boundary is the scheduling index parameter corresponding to the modulation and coding currently used by the base station.

7. The AMC optimization method for low-orbit satellite communication based on AI-rule fusion decision according to claim 1, characterized in that, A probability of a hybrid automatic repeat request check passing corresponding to each modulation and coding within a set scheduling number of times is dynamically counted, and a modulation and coding ratio used when scheduling uplink and downlink is determined according to a probability of the hybrid automatic repeat request check passing corresponding to the modulation and coding currently used by the base station and a probability of the hybrid automatic repeat request check passing corresponding to the modulation and coding currently used by the base station minus 1.

8. The AMC optimization method for low-orbit satellite communication based on AI-rule fusion decision according to claim 7, characterized in that, The modulation and coding ratio used when scheduling uplink and downlink is calculated in the following mode: wherein, is a scheduling ratio, is a set target bit error rate, is a scheduling index parameter corresponding to a modulation and coding pair currently used by the base station, is a scheduling index parameter corresponding to a modulation and coding pair currently used by the base station minus 1.

9. The AMC optimization method for low-orbit satellite communication based on AI-rule fusion decision according to claim 1, characterized in that, The low-orbit satellite communication adaptive modulation and coding optimization model adopts a plurality of stacked Transformer encoder layer structures, and is mapped to different modulation and coding through a fully connected layer.

10. The AMC optimization method for low-orbit satellite communication based on AI-rule fusion decision according to claim 1, characterized in that, The low-orbit satellite communication adaptive modulation and coding optimization model sets a time period and a data collection time period during model training, performs modulation and coding scheduling within the time period, collects training data from a start time of the time period to the data collection time period, and performs model training from the data collection time period to an end time of the time period.

Images