An access control and degree distribution joint optimization method and system for MEO satellite internet

Abstract

This invention discloses a joint optimization method for access control and degree distribution for MEO satellite internet. The method includes optimizing the access prohibition probability based on a Markov decision process to obtain optimized action decisions; optimizing the degree distribution based on a population optimization algorithm to obtain the optimal degree distribution, which is then fed back to the prohibition probability optimization process until the optimal prohibition probability is determined; subsequently, a generalized degree distribution strategy is generated based on the optimal prohibition probability and deployed to the satellite; satellite broadcasting and random access by ground user terminals complete the joint optimization. This scheme deeply integrates access control technology with the IRSA random access protocol to achieve adaptive response to dynamic environments and maximize throughput. It utilizes reinforcement learning algorithms to optimize access control parameters and differential evolution algorithms to optimize degree distribution parameters, achieving optimal strategy.

Description

Technical Field

[0001] This invention belongs to the field of satellite communications, specifically relating to a method and system for joint optimization of access control and degree distribution for MEO satellite internet. Background Technology

[0002] Medium Earth Orbit (MEO) satellites, with their wide coverage advantage, have become a crucial infrastructure for building a wide-area maritime internet. However, MEO satellite communications face dual challenges from both the physical environment and service characteristics: on the one hand, sudden surges in maritime monitoring traffic can easily overload the network; on the other hand, Ka-band links are highly susceptible to attenuation from rainfall, leading to physical link interruptions. To address these challenges, access control and random access protocols are key technologies for resolving network congestion and ensuring link transmission reliability, but their adaptability to MEO satellite internet scenarios faces severe tests.

[0003] Access control technology can dynamically manage the data transmission behavior of ground user terminals, effectively avoiding network congestion and ensuring system stability under high load. However, its implementation relies on frequent downlink signaling interactions, resulting in significant control command lag in MEO long-latency environments. It also forces energy-constrained terminals to frequently wait in standby mode to decode control frames, leading to substantial consumption of signaling resources. Traditional IRSA degree distribution is based on the assumption of an ideal additive white Gaussian noise channel and is optimized only for collision models. In Ka-band satellite communication, rain attenuation can cause data packet copies to be interrupted directly at the physical layer. Existing fixed degree distribution strategies cannot detect dynamic changes in channel interruption probability. Under severe weather conditions, the copy loss rate increases sharply, directly causing the iterative decoding process to terminate prematurely, resulting in a precipitous drop in system throughput. Summary of the Invention

[0004] To address the aforementioned problems, the present invention aims to provide a method and system for joint optimization of access control and degree distribution for MEO satellite internet. This method deeply integrates access control technology with the IRSA random access protocol to achieve adaptive response to dynamic environments and maximize throughput. The method utilizes reinforcement learning algorithms to optimize access control parameters and differential evolution algorithms to optimize degree distribution parameters, thereby achieving policy optimization.

[0005] The specific technical solution for achieving the objective of this invention is as follows:

[0006] A joint optimization method for access control and degree distribution for MEO satellite internet includes the following steps:

[0007] Step 1: Optimize the access prohibition probability based on the Markov decision process to obtain the optimized action decision;

[0008] Step 2: Optimize the degree distribution based on the population optimization algorithm to obtain the optimal degree distribution, and feed it back to the prohibition probability optimization process;

[0009] Step 3: Repeat steps 1 and 2 until the optimal prohibition probability is determined;

[0010] Step 4: Generate a generalized distribution strategy based on the optimal prohibition probability and deploy it to the satellite;

[0011] Step 5: Satellite broadcasting and ground user terminals are randomly connected to complete joint optimization.

[0012] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0013] (1) Adapt to the long latency characteristics of MEO satellite communication, reduce signaling interaction consumption through access control design, reduce standby decoding energy consumption of energy-constrained ground terminals, and ensure stable system operation under high load scenarios;

[0014] (2) To address the dynamic changes in channel status caused by rainfall attenuation in the Ka band of MEO satellite communication, access control technology is used to reduce the impact of data packet copy loss on the decoding process and ensure the stability of system throughput under complex weather conditions;

[0015] (3) To address the risks of access load fluctuations and link interruptions caused by sudden traffic surges, we will achieve coordinated optimization of access control and degree distribution to improve the transmission reliability of the system in dynamic environments.

[0016] (4) Based on this scheme, the ground user terminal only needs to read the generalized distribution of the broadcast once and perform probability sampling to simultaneously make decisions on whether to access and how to encode. This mechanism avoids the terminal frequently waking up to decode the control frame, reducing the communication energy consumption and signal processing complexity of the energy-constrained maritime terminal.

[0017] (5) This scheme can effectively cope with fluctuations in rainfall attenuation levels and access load. In heavy rain or high load scenarios, the algorithm will automatically adjust the access prohibition probability to prevent congestion, and optimize the degree distribution parameters to combat packet loss. Simulation results show that under different rainfall attenuation levels and sudden traffic surges, the simulation results are as follows (see attached figure). Figure 3 and attached Figure 4 As shown in the figure. The results show that the present invention can maintain a better throughput, effectively solving the problem of the precipitous drop in throughput of the traditional fixed-parameter IRSA protocol under severe weather conditions, and significantly improving the anti-interference ability and robustness of the system;

[0018] (6) This scheme transfers the high-complexity algorithm training process to the ground platform to generate a policy table covering the entire state space. When the satellite is in orbit, it only needs to perform a low-complexity table lookup operation based on the real-time measured signal-to-noise ratio and load estimation value to generate the broadcast policy, overcoming the long latency of MEO and the limitations of on-board computing power, and realizing low-latency real-time decision-making.

[0019] The present invention will be further described below with reference to specific embodiments. Attached Figure Description

[0020] Figure 1 This is a schematic diagram of the scenario architecture in which the present invention is applied.

[0021] Figure 2 This is a schematic diagram of the joint optimization method for access control and degree distribution for MEO satellite internet according to the present invention.

[0022] Figure 3 This is a comparison chart of normalized throughput performance under different rainfall attenuation levels in an embodiment of the present invention.

[0023] Figure 4 This is a simulation diagram comparing the dynamic response and stability of the standard IRSA protocol under burst traffic load and moderate rain in an embodiment of the present invention. Detailed Implementation

[0024] Example

[0025] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. The described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0026] As indicated in this application and claims, unless the context clearly indicates otherwise, the words "a," "an," "an," and / or "the" do not specifically refer to the singular and may also include the plural. Generally speaking, the terms "comprising" and "including" only indicate the inclusion of explicitly identified steps and elements, which do not constitute an exclusive list, and the method or apparatus may also include other steps or elements.

[0027] Unless otherwise specifically stated, the relative arrangement, numerical expressions, and values ​​of the components and steps described in these embodiments do not limit the scope of this application. It should also be understood that, for ease of description, the dimensions of the various parts shown in the drawings are not drawn to actual scale. Techniques, methods, and devices known to those skilled in the art may not be discussed in detail, but where appropriate, such techniques, methods, and devices should be considered part of the specification. In all examples shown and discussed herein, any specific values ​​should be interpreted as merely exemplary and not as limitations. Therefore, other examples of exemplary embodiments may have different values. It should be noted that similar reference numerals and letters in the following drawings denote similar items; therefore, once an item is defined in one drawing, it need not be further discussed in subsequent drawings.

[0028] To maintain high throughput under severe rain conditions, this paper proposes a joint optimization method for access control and degree distribution for MEO (Medium Earth Orbit) satellite internet. The overall design architecture of the satellite, users, and computing platform in this invention is as follows: Figure 1 As shown, the method of this invention expresses the access prohibition probability, a traditional MAC (Media Access Control) layer access control parameter, as a parameter with a degree of 0 in the physical layer degree distribution encoding, constructing a generalized degree distribution containing a "zero-degree coefficient." Simultaneously, it performs two-layer joint training using "Q"-Learning and DE (Differential Evolution) algorithms. The first layer optimization utilizes the "Q"-Learning algorithm to optimize the access prohibition probability, finding the most suitable access prohibition probability under the current environment to achieve effective control of access load. The second layer optimization utilizes the differential evolution algorithm to optimize the degree distribution, finding the optimal degree distribution parameter to effectively cope with rain attenuation. By constructing a system architecture of offline ground simulation training and online onboard decision-making, the training process is completed on a ground computing platform. After training, the generated generalized degree distribution table is deployed to the MEO satellite. Based on this generalized degree distribution table, the satellite achieves adaptive response to channel state and access load fluctuations, thereby dynamically adjusting the access strategy of ground user terminals.

[0029] Combination Figure 2 A joint optimization method for access control and degree distribution for MEO satellite internet includes the following steps:

[0030] Step 1: Optimize the access prohibition probability based on Markov Decision Process (MDP) to obtain the optimized action decision:

[0031] Step 1-1: Construct an access prohibition probability optimization model based on Markov decision process:

[0032] Define access load and channel interruption probability As a state variable, the access load The average number of users requesting data transmission per unit time slot, representing the channel interruption probability. Used to characterize the probability of physical link interruption caused by Ka-band rainfall attenuation;

[0033] Define access denial probability As an action variable, it is used to characterize the probability that a ground user terminal is denied access;

[0034] Based on this, initialize a two-dimensional value function table: The table contains rows that correspond to state indices and columns that correspond to action indices; all initial... Set the value to 0;

[0035] Initialize and set the maximum number of iterations. Learning rate Initial exploration probability And explore the decay factor ;

[0036] This embodiment defines the access load. The state space range is This is then linearly quantized into 60 discrete levels. The channel interruption probability is defined. The set of states is These correspond to three typical meteorological environments: light rain, moderate rain, and heavy rain. The action space is defined as the access prohibition probability. The range of values ​​is This is discretized into 80 action levels. The initial dimension is... Value function table The table is initialized with all values ​​of 0. Set the reinforcement learning training parameters: maximum number of iterations. Learning rate Initial exploration probability Explore the attenuation factor .

[0037] Step 1-2: Begin the state traversal in the access prohibition probability optimization algorithm:

[0038] The trained model used in the prohibition probability optimization process is used as an agent to traverse every state in the state space. For each state, determine the corresponding access prohibition probability. and update The corresponding table value;

[0039] Action decision-making in the prohibition probability optimization algorithm, for the currently traversed state ,use - Greedy strategy (Epsilon-Greedy) to select the current access denial probability After selecting an action, the current access load will be... Current channel interruption probability and the probability of currently determined access denial. As input parameters, they are used in degree distribution optimization.

[0040] In this embodiment, a [missing information] is generated. random numbers in an interval ,when At that time, the agent executes exploration mode, randomly selecting one action with equal probability from 60 actions, based on the access prohibition probability. As the current access denial probability To traverse the unknown strategy space; when At that time, the intelligent agent executes the utilization mode and queries the value function table. Table, select status Down The probability of prohibiting access to the action corresponding to the element with the largest value As the current access denial probability After selecting an action, the agent will ( The input parameter is passed to the distributed optimization algorithm module.

[0041] Step 2: Optimize the degree distribution based on the population optimization algorithm to obtain the optimal degree distribution, and feed it back to the prohibition probability optimization process:

[0042] Constructing a degree distribution optimization model based on population optimization algorithm:

[0043] The interruption probability input into the degree distribution optimization model is defined as... Access denial probability is defined as Set population size Maximum number of generations (Gen), scaling factor Crossover probability Construct a population matrix, where each row of the matrix represents an individual in the population. ,individual The The column values ​​represent the degree distribution. coefficient Used to characterize the transmission of ground user terminals The probability of a copy;

[0044] The first individual in the population Set as the standard IRSA (segmented irregular slotted ALOHA) degree distribution vector Suitable for low-load, light rain-induced degradation environments; the second individual Set as a low-order degree distribution vector Suitable for high-load, moderate rain-attenuation environments; the third individual Set as a high-order robustness distribution vector Suitable for environments with severe rain attenuation;

[0045] The remaining individuals in the population are generated randomly and normalized;

[0046] At the same time, all individuals are forced The first column contains values ​​of 0, which indicates that the degree is in the degree distribution. =1 coefficient Set to 0 to suppress the short-loop effect from disrupting iterative decoding;

[0047] In this embodiment, the interruption probability passed to the degree distribution optimization algorithm module is defined as... Access denial probability is defined as Set population size Maximum number of generations scaling factor 0.6, crossover probability Construct a population matrix of 40 rows and 8 columns, where each row represents an individual. Each individual Given a 1×8 vector, the first... The column values ​​represent the degree distribution. coefficient The first individual in the population Set as standard IRSA degree distribution vector Suitable for low to medium loads and light rain environments; the second individual Set as a low-order degree distribution vector Suitable for high-load, moderate rain-attenuation environments; the third individual Set as a higher-order robustness distribution vector This is suitable for environments with severe rain attenuation. The remaining 27 individuals were randomly generated and normalized, while the degree of all individuals was forced. =1 coefficient .

[0048] In degree distribution optimization algorithms, the population traversal begins by iterating through each individual in the population, sequentially performing evolutionary operations, fitness evaluation, and evolutionary selection until the maximum number of evolutionary iterations (GEN) is reached. After the iterations are completed, the maximum fitness in the current population is determined. and the corresponding individuals Perform an inspection; if the maximum fitness is... Anomaly, reset optimality distribution And calculate the corresponding fitness. As the maximum fitness Otherwise, the individual As the optimal degree distribution ;

[0049] Distribution of optimality With maximum fitness As the output of the degree distribution optimization model;

[0050] The evolutionary operation is:

[0051] For each individual encountered during the iteration, it is treated as the parent individual. Then, a mutation vector is generated based on a random difference strategy. And perform a non-negativity constraint check;

[0052] Subsequently, the mutation vector was analyzed. With parental individuals Perform binomial cross-validation to generate test vectors After the crossover is completed, the test vector is calculated. If the sum of all components is less than a preset minimum value, then the test vector is reset. This is to ensure that a valid solution always exists in the population; otherwise, for the experimental vector Perform normalization processing.

[0053] In this embodiment, each traversed individual is treated as a parent individual. Then based on Mutation strategy, from individuals without the original parent. Randomly select three distinct individual indices from the current population. Based on these three basis vectors, the mutation vector is calculated according to the following formula. : ;

[0054] After mutation is completed, the mutated vector is... Each dimension is checked for nonnegativity constraints; if a value in any dimension is found to be less than 0, it is corrected to 0. Subsequently, the mutation vector is... With the original paternal individual Perform binomial crossover operations to generate test vectors That is, for each dimension of the vector. Generate a random numbers in an interval ,like Less than the crossover probability or dimension index Equal to the randomly selected required dimension index Then the test vector The Dimension values ​​are taken from the mutation vector Otherwise, retain the parent individual. No. The original value of the dimension. After the crossover operation is completed, the trial vector is calculated. The sum of all components like Then the test vector Reset to standard IRSA degree distribution vector Otherwise, for the experimental vector Perform normalization:

[0055] ;

[0056] in, The original experimental vector before normalization The Column values, For the normalized first The column values ​​ensured ;

[0057] The fitness assessment is:

[0058] Define the current degree distribution vector as the candidate degree distribution vector. Using density evolution theory, the candidate degree distribution vector is evaluated. The performance, in order to Normalized throughput As fitness Evaluation indicators:

[0059] First, by calculating the probability of the ground user terminal sending d copies and the satellite receiving k copies using binomial probability, the actual equivalence distribution vector corresponding to the satellite receiver can be generated. And then according to Calculate the probability that the transmission of the ground user terminal is not interrupted by rain attenuation. ,like If the value is less than the threshold for determining the validity of the degree distribution, then the candidate degree distribution vector is directly determined. Ineffective, its fitness The value is 0; otherwise, the candidate degree distribution vector is determined. efficient;

[0060] For the effective candidate degree distribution vector According to the access load and access denial probability And the probability of transmission not being interrupted by rain attenuation. Calculate the actual payload that enters the decoder at the satellite receiver. Then set the initial single-copy decoding failure probability. According to the payload With equivalence distribution vector The iterative decoding process for serial interference cancellation is simulated using state evolution equations, and the decoding failure probability of a single replica after convergence is calculated. Then set the convergence threshold. If in the iteration of the access prohibition probability optimization algorithm, and after convergence... Exceeding this threshold Then the fitness value =0; otherwise according to and Calculate the transmission failure probability for each user. Finally, the normalized throughput is calculated. And use it as the candidate degree distribution vector fitness ,Right now ;

[0061] In this embodiment, density evolution theory is utilized to... Normalized throughput As fitness Evaluation metrics, evaluation candidate degree distribution vector The performance of [the system / mechanism] is assessed. The binomial theorem is used to calculate the probability of channel interruption. Next, the ground user terminal sends... One copy, actual satellite reception The probability of a copy :

[0062]

[0063] according to The distribution vector of the selection estimate can be obtained. The actual equivalent distribution vector at the satellite receiver end Among them, the equivalence distribution vector The Column values The calculation formula is as follows:

[0064]

[0065] Based on the equivalence distribution vector The probability that the transmission of the ground user terminal was not interrupted by rain attenuation was calculated. :

[0066]

[0067] like Less than the degree distribution validity threshold Then directly determine the candidate degree distribution vector. Ineffective, its fitness The value is 0. Otherwise, for the equivalence distribution Normalization is performed to generate a normalized equivalence distribution vector. The validity distribution vector The Middle numerical values The calculation formula is as follows:

[0068]

[0069] Based on access load Access Denial Probability The probability of transmission not being interrupted by rain attenuation Calculate the effective physical load that actually enters the decoder at the satellite receiver. :

[0070]

[0071] Subsequently, the iterative decoding process of serial interference cancellation (SIC) is simulated using state evolution equations. An initial single-copy decoding failure probability is set. Let the first... After the iteration, the probability of a single copy failing to decode is: Based on the payload Equivalence distribution after normalization Perform the following iterative updates:

[0072]

[0073] in Let the edge be the degree distribution from the perspective, and the calculation formula is as follows:

[0074]

[0075] Iterative computation Until convergence, the probability of a single copy failing to decode is obtained. Set the threshold for successful decoding. =0.01, if in the iteration of the access prohibition probability optimization algorithm, and Then the fitness value It should be 0. Otherwise, according to... and Calculate the transmission failure probability for each user. :

[0076]

[0077] Finally, the normalized throughput is calculated. :

[0078]

[0079] The calculated normalized throughput As the candidate degree distribution vector fitness ,Right now .

[0080] Evolutionary selection in degree distribution optimization algorithms: by using trial vectors With parental individuals As candidate degree distribution vectors respectively Calculate its fitness and Then, a greedy selection mechanism is used to select the experimental vector. With parental individuals Performance comparisons are performed to select individuals for the next generation of the population. Simultaneously, the maximum fitness of the current generation is updated. and the corresponding individuals .

[0081] In the degree distribution optimization algorithm, the end of population traversal is achieved by repeating the above evolutionary operations, fitness evaluation, and evolutionary selection until all individuals in the population have been traversed.

[0082] In degree-distribution optimization algorithms, the iterative convergence of the population is achieved by repeating the above process until the maximum number of evolutionary iterations is reached. After the iteration is complete, the maximum fitness in the current population is determined. and the corresponding individuals Perform an inspection. If the maximum fitness is... Anomaly, reset optimality distribution And calculate the corresponding fitness. As the maximum fitness Otherwise, the individual As the optimal degree distribution Finally, the optimal degree distribution is... With maximum fitness As the output of the degree distribution optimization algorithm module;

[0083] In this embodiment, after the iteration ends, the maximum fitness in the current population is... and the corresponding individuals Perform an inspection. If the maximum fitness is... Based on the channel interruption probability Directly set degree distribution: If Set a higher-order robustness distribution vector As the optimal degree distribution vector Otherwise, set the standard IRSA distribution vector. As the optimal degree distribution And recalculate fitness according to step seven. For fitness If fitness , will individuals As the optimal degree distribution vector Finally, the optimal degree distribution vector is... and the corresponding maximum fitness As the output of the degree distribution optimization algorithm module.

[0084] The optimality distribution is fed back into the prohibition probability optimization process, specifically as follows:

[0085] The maximum fitness determined by the degree distribution optimization model. As the probability of prohibiting access to the current action. Instant reward value ,Right now ;

[0086] use renew The table corresponds to "status" -action "Yes Value, and through the learning rate Adjust the update amplitude;

[0087] The updated formula is:

[0088]

[0089] Step 3: Repeat steps 1 and 2 until the optimal prohibition probability is determined.

[0090] Iterative training, after each round of training, the exploration probability is adjusted according to... Gradual decay;

[0091] When proceeding Iterative training rounds until... When the table converges, at this time The table already contains all states. Optimal access denial probability Information, including the optimal access denial probability For state Down The highest probability of access being blocked .

[0092] Step 4: Generate a generalized distribution strategy based on the optimal prohibition probability and deploy it to the satellite:

[0093] Optimal access denial probability Smoothing processing for channel outage probability For each discrete value, extract all access loads separately. Corresponding optimal access denial probability By using moving average filtering and local weighted regression algorithms, the system can achieve... The smoothing process ultimately generates multiple lines corresponding to different channel interruption probabilities. Continuous and smooth access control strategy curve ;

[0094] In this embodiment, a moving average filter with a window length of 5 is used to remove errors from the sequence. Local noise was then smoothed using a local weighted regression algorithm, with each access load... Based on this, a weighted neighborhood containing 15% of the sample data around the point is constructed. Within each neighborhood, the goal is to minimize the weighted residual sum of squares function. To optimize the objective function, the load point is solved. The best estimate below objective function The definition is as follows:

[0095]

[0096] in, As the initial distance weights, the sample points are far from the center. The closer the object, the greater its weight. To robustly correct the weights, they are set to 1 in the initial iteration, and the weights are increased as the sample fitting error decreases. , for data points The fitting residuals. This step is achieved by minimizing... This allows the fitted curve to approximate high-quality local sample points as closely as possible. Simultaneously, by fitting the residuals... Update robust weights The calculation formula is:

[0097]

[0098] in To fit the residuals The median of the absolute deviation. The updated robust weights. Substitute this into the next round of locally weighted fitting, and repeat the above "fitting-weight update" process until convergence. Finally, for different channel outage probabilities... According to the load point Compared with the best estimate Generate three continuous and smooth access control policy curves. .

[0099] Generate a generalized distribution table for the access control strategy curve. In according to The values ​​are sampled discretely within the range to obtain 60 discrete access loads. Then, based on the three access control policy curves Obtain the probability of interruption for different channels Each access load Corresponding access denial probability Then the parameters ( , , In the input degree distribution optimization model, the optimal degree distribution is recalculated. Finally, the probability of different channel interruptions will be determined. Next, access load point Corresponding optimal access denial probability , and the recalculated optimal degree distribution Integration, forming a generalized distribution :

[0100]

[0101] According to different channel interruption probabilities Classify and build a system that covers all states. Generalized distribution table;

[0102] In this embodiment, all load sampling points are categorized and organized. Corresponding generalized distribution Construct three maps containing "discrete access load sampling points" - Generalized distribution "A generalized distribution table of mapping relationships, each table containing 60 rows of discrete access load sampling points." Data and 9 columns of generalized distribution The data is then used to upload the three generalized degree distribution tables in batches to the on-board storage module of the MEO satellite through the policy uploading interface of the ground computing platform, thus completing the policy deployment.

[0103] Finally, the generated generalized degree distribution table is deployed to the satellite.

[0104] Step 5: Satellite broadcasting and ground user terminals are randomly connected to complete joint optimization.

[0105] During satellite operation, the uplink signal-to-noise ratio is measured in real time by the onboard receiver and mapped to the discrete channel interruption probability. Simultaneously estimate the current time slot access load. ;

[0106] The satellite is based on the above status Information, retrieve the corresponding generalized degree distribution from the on-board generalized degree distribution table. It is broadcast to ground user terminals within the beam coverage area via the downlink channel;

[0107] Ground user terminals read the broadcast generalized distribution when they have data transmission needs. Then, probabilistic random sampling is performed. Based on the sampling results, it is determined whether to access the network and the data packet copy encoding method. Finally, it is decided whether to send data and how many data copies to send. That is, the ground user terminal first generates a... random numbers in an interval ,like Smaller than the generalized distribution The zero coefficient in If the ground user terminal determines that access is not allowed, it will keep the radio frequency front-end off and not transmit any data; if Greater than or equal to degrees Within the probability interval corresponding to 2, the ground user terminal determines that access is allowed and directly generates a corresponding number of data packet copies for encoding and transmission.

[0108] Simulation results show that, under different rainfall attenuation levels and sudden flow impacts, the simulation results are as follows (see attached figure). Figure 3 and attached Figure 4 As shown in the figure. The results show that the present invention can maintain a relatively high throughput, effectively solving the problem of the precipitous drop in throughput of the traditional fixed-parameter IRSA protocol under severe weather conditions, and significantly improving the system's anti-interference capability and robustness.

[0109] This invention proposes a joint optimization mechanism for access control parameters and channel coding parameters using a generalized degree distribution model. Unlike existing technologies that require separate transmission of access prohibition probabilities and degree distribution parameters, this invention dynamically integrates the access prohibition probability into the zero-degree coefficient of the degree distribution through an optimization algorithm. After acquiring this unified policy, the terminal does not need to perform independent handshake or parsing processes; it only needs to make a decision between "silence" and "transmission" through a single probability sampling, significantly reducing the terminal's communication energy consumption and processing complexity.

[0110] The joint algorithm proposed in this invention possesses dynamic environmental adaptability. Through a two-layer collaborative optimization mechanism combining "Q"-Learning and differential evolution algorithms, the dynamic changes in access load and channel outage probability are integrated into the parameter optimization process. Compared to the limitations of traditional fixed-parameter or single-algorithm approaches that struggle to adapt to complex and ever-changing scenarios, this joint algorithm can capture environmental dynamics such as rain attenuation switching and sudden traffic surges in real time, automatically adjusting the combination of access prohibition probability and degree distribution parameters. It maintains optimal transmission performance under different channel conditions and load fluctuation scenarios, significantly improving anti-interference capability and robustness in complex maritime environments.

[0111] This invention addresses the contradiction between long propagation delays and limited computing resources on satellite platforms in medium Earth orbit satellite communication by establishing a system architecture that combines offline ground simulation training with onboard online table lookup decision-making. This architecture transfers the complex training process of reinforcement learning and differential evolution algorithms to a ground computing platform, deploying the generated optimal policy table covering a multi-dimensional state space onto the satellite. During satellite operation in orbit, only low-complexity table lookup operations are required to complete real-time decisions. This design avoids the risk of algorithm non-convergence that may occur with online learning, while significantly reducing the computational burden on the onboard processor, ensuring the system can respond in real-time to dynamically changing channel environments.

[0112] This solution also provides a joint optimization system for access control and degree distribution for MEO satellite internet, including the following modules:

[0113] Access prohibition probability optimization module: used to optimize the access prohibition probability based on Markov decision process to obtain optimized action decisions;

[0114] Degree distribution optimization module: used to optimize the degree distribution based on the population optimization algorithm, obtain the optimal degree distribution, and feed it back to the prohibition probability optimization process;

[0115] Generalized policy generation module: used to generate generalized distribution policies based on the optimal prohibition probability and deploy them to satellites;

[0116] Random access module: Satellite broadcasting and ground user terminals randomly access each other, and joint optimization is completed.

[0117] The embodiments described above are merely one implementation method of this application, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of the invention patent. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this patent application should be determined by the appended claims.

Claims

1. An access control and degree distribution joint optimization method for MEO satellite Internet, characterized in that, Includes the following steps: Step 1: Optimize the access prohibition probability based on the Markov decision process to obtain the optimized action decision; Step 2: Optimize the degree distribution based on the population optimization algorithm to obtain the optimal degree distribution, and feed it back to the prohibition probability optimization process; Step 3: Repeat steps 1 and 2 until the optimal prohibition probability is determined; Step 4: Generate a generalized distribution strategy based on the optimal prohibition probability and deploy it to the satellite; Step 5: Satellite broadcasting and ground user terminals are randomly connected to complete joint optimization.

2. The access control and degree distribution joint optimization method for MEO satellite internet according to claim 1, characterized in that, The optimized action decision obtained in step 1 is specifically as follows: Step 1-1: Construct an access prohibition probability optimization model based on Markov decision process: Definition of access load and channel outage probability as a state variable, the access load characterizes the average number of users per time slot that request to send data, the channel outage probability for characterizing the physical link outage probability due to rain attenuation in Ka band; Definition of access barring probability As an action variable, used to characterize the probability that a ground user terminal is barred from accessing; On this basis, a two-dimensional value function table is initialized: Table, the row values correspond to state indexes, and the column values correspond to action indexes. All initial values are set to 0. values are set to 0. Initialize and set the maximum number of iterations. Learning rate Initial exploration probability And explore the decay factor ; Step 1-2: Traverse each state in the state space For each state, determine the corresponding access prohibition probability. and update The corresponding table value; For the current traversed state ,use - Greedy strategy to select the current access prohibition probability After selecting an action, the current access load will be... Current channel interruption probability and the probability of currently determined access denial. As input parameters, they are used in degree distribution optimization.

3. The joint optimization method for access control and degree distribution for MEO satellite internet according to claim 2, characterized in that, The degree distribution optimization based on the population optimization algorithm in step 2 is specifically as follows: Constructing a degree distribution optimization model based on population optimization algorithm: The interruption probability input into the degree distribution optimization model is defined as... Access denial probability is defined as Set population size Maximum number of generations (Gen), scaling factor Crossover probability Construct a population matrix, where each row of the matrix represents an individual in the population. ,individual The The column values ​​represent the degree distribution. coefficient Used to characterize the transmission of ground user terminals The probability of a copy; The first individual in the population Set as standard IRSA degree distribution vector Suitable for low-load, light rain-induced degradation environments; the second individual Set as a low-order degree distribution vector Suitable for high-load, moderate rain-attenuation environments; the third individual Set as a high-order robustness distribution vector Suitable for environments with severe rain attenuation; The remaining individuals in the population are generated randomly and normalized; At the same time, all individuals are forced The first column contains values ​​of 0, which indicates that the degree is in the degree distribution. =1 coefficient Set to 0 to suppress the short-loop effect from disrupting iterative decoding; The process iterates through each individual in the population, performing evolutionary operations, fitness assessments, and evolutionary selections sequentially until the maximum number of evolutionary iterations (GEN) is reached. After the iterations are completed, the maximum fitness in the current population is determined. and the corresponding individuals Perform an inspection; if the maximum fitness is... Anomaly, reset optimality distribution And calculate the corresponding fitness. As the maximum fitness ; Otherwise, the individual As the optimal degree distribution ; Distribution of optimality With maximum fitness As the output of the degree distribution optimization model.

4. The joint optimization method for access control and degree distribution for MEO satellite internet according to claim 3, characterized in that, The evolutionary operation is: For each individual encountered during the iteration, it is treated as the parent individual. Then, a mutation vector is generated based on a random difference strategy. And perform a non-negativity constraint check; Subsequently, the mutation vector was analyzed. With parental individuals Perform binomial cross-validation to generate test vectors After the crossover is completed, the test vector is calculated. If the sum of all components is less than a preset minimum value, then the test vector is reset. This is to ensure that a valid solution always exists in the population; otherwise, for the experimental vector Perform normalization processing.

5. The joint optimization method for access control and degree distribution for MEO satellite internet according to claim 3, characterized in that, The fitness assessment is: Define the current degree distribution vector as the candidate degree distribution vector. Using density evolution theory, the candidate degree distribution vector is evaluated. The performance, in order to Normalized throughput As fitness Evaluation indicators: First, by calculating the probability of the ground user terminal sending d copies and the satellite receiving k copies using binomial probability, the actual equivalence distribution vector corresponding to the satellite receiver can be generated. And then according to Calculate the probability that the transmission of the ground user terminal is not interrupted by rain attenuation. ,like If the value is less than the threshold for determining the validity of the degree distribution, then the candidate degree distribution vector is directly determined. Ineffective, its fitness The value is 0; otherwise, the candidate degree distribution vector is determined. efficient; For the effective candidate degree distribution vector According to the access load and access denial probability And the probability of transmission not being interrupted by rain attenuation. Calculate the actual payload that enters the decoder at the satellite receiver. Then set the initial single-copy decoding failure probability. According to the payload With equivalence distribution vector The iterative decoding process for serial interference cancellation is simulated using state evolution equations, and the decoding failure probability of a single replica after convergence is calculated. Then set the convergence threshold. If in the iteration of the access prohibition probability optimization algorithm, and after convergence... Exceeding this threshold Then the fitness value =0; otherwise according to and Calculate the transmission failure probability for each user. Finally, the normalized throughput is calculated. And use it as the candidate degree distribution vector fitness ,Right now ; By using experimental vectors With parental individuals As candidate degree distribution vectors respectively Calculate its fitness and Then, a greedy selection mechanism is used to select the experimental vector. With parental individuals Performance comparisons are performed to select individuals for the next generation of the population. Simultaneously, the maximum fitness of the current generation is updated. and the corresponding individuals .

6. The joint optimization method for access control and degree distribution for MEO satellite internet according to claim 3, characterized in that, The step 2, feeding back the optimality distribution to the prohibition probability optimization process, specifically involves: The maximum fitness determined by the degree distribution optimization model. As the probability of prohibiting access to the current action. Instant reward value ,Right now ; use renew The table corresponds to "state" -action "Yes Value, and through the learning rate Adjust the update range.

7. The joint optimization method for access control and degree distribution for MEO satellite internet according to claim 3, characterized in that, In step 3, determining the optimal prohibition probability specifically involves: When proceeding Iterative training rounds until... When the table converges, at this time The table already contains all states. Optimal access denial probability Information, including the optimal access denial probability For state Down The highest probability of access being blocked .

8. The joint optimization method for access control and degree distribution for MEO satellite internet according to claim 3, characterized in that, The strategy for generating a generalized distribution based on the optimal prohibition probability in step 4 is as follows: Optimal access denial probability Smoothing processing for channel outage probability For each discrete value, extract all access loads separately. Corresponding optimal access denial probability By using moving average filtering and local weighted regression algorithms, the system can achieve... The smoothing process ultimately generates multiple lines corresponding to different channel interruption probabilities. Continuous and smooth access control strategy curve ; Generate a generalized distribution table for the access control strategy curve. In Discrete sampling is performed to obtain discrete access loads. Then, based on the three access control policy curves Obtain the probability of interruption for different channels Each access load Corresponding access denial probability Then the parameters ( , , In the input degree distribution optimization model, the optimal degree distribution is recalculated. Finally, the probability of different channel interruptions will be determined. Next, access load point Corresponding optimal access denial probability , and the recalculated optimal degree distribution Integration, forming a generalized distribution ; According to different channel interruption probabilities Classify and build a system that covers all states. Generalized distribution table; Finally, the generated generalized degree distribution table is deployed to the satellite.

9. The joint optimization method for access control and degree distribution for MEO satellite internet according to claim 3, characterized in that, The random access of satellite broadcasting and ground user terminals in step 5 is specifically as follows: During satellite operation, the uplink signal-to-noise ratio is measured in real time by the onboard receiver and mapped to the discrete channel interruption probability. Simultaneously estimate the current time slot access load. ; The satellite is based on the above status Information, retrieve the corresponding generalized degree distribution from the on-board generalized degree distribution table. It is broadcast to ground user terminals within the beam coverage area via the downlink channel; Ground user terminals with data transmission needs read the broadcast generalized distribution. Then, probabilistic random sampling is performed, and the sampling results determine whether to access the network and the encoding method of the data packet replicas. Finally, it is decided whether to send data and how many data replicas to send.

10. A joint optimization system for access control and degree distribution for MEO satellite internet, characterized in that, Includes the following modules: Access prohibition probability optimization module: used to optimize the access prohibition probability based on Markov decision process to obtain optimized action decisions; Degree distribution optimization module: used to optimize the degree distribution based on the population optimization algorithm, obtain the optimal degree distribution, and feed it back to the prohibition probability optimization process; Generalized policy generation module: used to generate generalized distribution policies based on the optimal prohibition probability and deploy them to satellites; Random access module: Satellite broadcasting and ground user terminals randomly access each other, and joint optimization is completed.

Images