Efficient data transmission method in Beidou satellite communication system

Abstract

The invention relates to the technical field of satellite communication, and discloses an efficient data transmission method in a Beidou satellite communication system, and the method comprises the following steps: initializing and deploying a federated learning model based on deep reinforcement learning on each satellite, and collecting channel information and real-time feedback data; generating a fractional order partial differential equation model for describing channel interference according to the collected channel information and real-time feedback data, and performing online parameter identification; quantum-classical hybrid optimization is implemented based on the established fractional order partial differential equation interference model, and the channel state is dynamically optimized through a reconfigurable quantum logic unit; and based on the optimized channel state, adaptively adjusting a modulation mode and a coding strategy according to the real-time channel quality. According to the method, the federated learning model based on deep reinforcement learning is adopted, the effects of efficiently collecting channel information and feeding back data in real time are achieved, the problem of data islands is solved, and the information sharing efficiency is remarkably improved.

Description

Technical Field

[0001] The present invention relates to the field of satellite communication technology, and in particular to a high-efficiency data transmission method in a Beidou satellite communication system. Background Art

[0002] At present, the Beidou system has gradually advanced from single modulation and coding optimization to a multi-level and multi-technology integration stage in terms of efficient data transmission, including antenna arrays, integrated satellite-ground networks, inter-satellite optical links and intelligent scheduling. In the future, with the further expansion of deep learning, quantum communication, satellite software-defined networks, higher frequency bands and constellation scale, the efficiency and capacity of Beidou data transmission will be greatly improved.

[0003] However, traditional data processing methods often face the problem of data silos. Information sharing between nodes is limited, and different devices or systems cannot efficiently communicate and integrate channel information. This leads to resource waste and reduced information utilization efficiency, and cannot form a comprehensive understanding of the channel state, thus affecting the performance of the entire network. Secondly, many existing systems use static models to describe channel interference. This approach fails to effectively adapt to dynamically changing channel environments. This means that when interference or channel conditions change, the system lacks an effective adaptive mechanism, resulting in a serious decline in signal availability and quality. Static models cannot capture instantaneous changes in channel conditions in a timely manner, which in turn makes it impossible for the system to perform real-time parameter identification and adjustment, thus losing the ability to cope with complex environments. In addition, classic optimization algorithms have slow calculation speeds when processing channel state optimization and cannot meet the requirements of real-time response. This performance bottleneck makes it difficult for the system to make timely adjustments under rapidly changing channel conditions, thereby affecting the efficiency and accuracy of data transmission. Finally, the modulation method in existing communication systems is often fixed, which lacks flexibility in the face of changing channel quality. Fixed modulation methods cannot be intelligently adjusted according to real-time channel information, resulting in reduced data transmission rate and reliability, especially in complex electromagnetic environments. Summary of the Invention

[0004] In response to the shortcomings of the existing technology, the present invention provides an efficient data transmission method in the Beidou satellite communication system, which solves the problems of inefficient data utilization, insufficient signal processing flexibility, slow calculation speed and insufficient optimization caused by isolated information propagation.

[0005] To achieve the above objectives, the present invention is implemented by the following technical solutions: an efficient data transmission method in a Beidou satellite communication system, comprising the following steps:

[0006] S1. Initialize and deploy a federated learning model based on deep reinforcement learning on each satellite to collect channel information and real-time feedback data;

[0007] S2. Generate a fractional-order partial differential equation model for describing channel interference based on collected channel information and real-time feedback data, and perform online parameter identification;

[0008] S3. Based on the established fractional-order partial differential equation interference model, quantum-classical hybrid optimization is implemented to dynamically optimize the channel state through reconfigurable quantum logic units;

[0009] S4. Based on the optimized channel state, adaptively adjust the modulation mode and coding strategy according to the real-time channel quality;

[0010] S5, based on adaptive adjustment of modulation mode and coding strategy, implements dynamic data transmission, responds to interference changes in real time and optimizes transmission rate, and conducts information feedback and coordination through inter-satellite communication;

[0011] S6. After completing dynamic data transmission, the model between satellites is aggregated and updated to optimize the next round of transmission strategy, while performance verification is performed in the process.

[0012] Preferably, the federated learning model of deep reinforcement learning in step S1 is a deep Q network, which includes at least 18 layers of convolutional neural networks and gated recurrent units.

[0013] Preferably, in step S2, a fractional-order partial differential equation model for describing channel interference is generated, and the mathematical expression of the fractional-order partial differential equation model is:

[0014]

[0015] In the formula, represents the α-order partial derivative of the function Ψ with respect to time t, Ψ represents the function to be solved, κ represents the diffusion coefficient, which affects the propagation rate of Ψ in space, and represents the second-order partial derivative of the function Ψ with respect to the spatial variables x and y, β represents a coefficient, represents the first-order partial derivative of Ψ with respect to the spatial variable z, describing the change of Ψ in that direction.

[0016] Preferably, the online parameter identification in step S2 uses a particle swarm optimization algorithm to identify the propagation coefficients κ and β in the interference model.

[0017] Preferably, quantum-classical hybrid optimization is implemented in step S3, and the Hamiltonian model of the quantum-classical hybrid optimization is:

[0018]

[0019] In the formula, represents the Hamiltonian, N represents the number of qubits in the system, and Ji represents the channel gain of the i-th qubit, represents the Pauli z operator of the i-th qubit, V ij represents the coupling strength between the i-th and j-th qubits, represents the Pauli x-operator for the i-th qubit.

[0020] Preferably, the reconfigurable quantum logic unit in step S3 is used to perform preset quantum operations, including quantum search and frequency domain transformation.

[0021] Preferably, the adaptively adjusting the coding strategy in step S4 further includes: selecting convolutional coding, Turbo coding or LDPC coding according to the channel state.

[0022] Preferably, the dynamic data transmission in step S5 is based on an adaptive modulation mode and coding strategy, and the adaptive modulation and coding mechanism is formed using a dynamic selection principle as follows:

[0023]

[0024] In the formula, Modulation represents the selected modulation mode, CQI represents channel quality indicator, BPSK represents binary phase shift keying, QPSK represents quadrature phase shift keying, and 16QAM represents 16-order quadrature amplitude modulation.

[0025] Preferably, in step S6, the model between satellites is converged and updated, and the model update formula is:

[0026]

[0027] In the formula, W global represents the global model weight, K represents the total number of satellites, W k represents the local model weight of the k-th satellite, M k represents the mask matrix of the kth satellite, and ⊙ represents the Hadamard product.

[0028] Preferably, the performance verification step in step S6 includes a satellite-ground joint test performed under simulated interference conditions to evaluate the transmission effect and optimization capability under different channel states.

[0029] The present invention provides an efficient data transmission method in the Beidou satellite communication system. It has the following beneficial effects:

[0030] 1. The present invention adopts a federated learning model based on deep reinforcement learning to achieve the effect of efficiently collecting channel information and real-time feedback data. Compared with traditional data processing methods in the existing technology, it overcomes the problem of data silos, significantly improves the efficiency of information sharing, and solves the problem of inefficient data utilization caused by isolated information dissemination.

[0031] 2. The present invention accurately describes channel interference and performs online parameter identification by generating a fractional-order partial differential equation model, which enables the system to dynamically adapt to changes in the channel environment. Compared with the limitations of the static model in the existing technology, this dynamism effectively reduces the impact of interference on the signal and solves the problem of insufficient signal processing flexibility.

[0032] 3. The present invention adopts quantum-classical hybrid optimization technology and significantly improves the channel state optimization capability by introducing a reconfigurable quantum logic unit. Compared with the classical optimization algorithm in the existing technology, it effectively solves the problem of slow computing speed, makes real-time response more efficient, and improves the system's processing capability.

[0033] 4. The present invention implements adaptive modulation methods and coding strategies to flexibly respond to changes in channel quality, thereby significantly improving data transmission rate and reliability. Compared with the limitations of fixed modulation methods in existing technologies, this innovation enhances the adaptability and intelligence of the system in complex environments and solves the problem of insufficient optimization. BRIEF DESCRIPTION OF THE DRAWINGS

[0034] Figure 1 Schematic diagram of the method of the present invention. DETAILED DESCRIPTION

[0035] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the drawings in the present specification. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0036] Please see the attached Figure 1 The embodiment of the present invention provides an efficient data transmission method in a BeiDou satellite communication system, including:

[0037] S1. Initialize and deploy a federated learning model based on deep reinforcement learning on each satellite to collect channel information and real-time feedback data;

[0038] Specifically, first, a federated learning model for deep reinforcement learning is established. The model is a deep Q-network (DQN), whose core includes at least 18 layers of convolutional neural networks (CNNs) and gated recurrent units (GRUs). The convolutional neural network is used to extract features from channel information, while the GRU is designed to capture dependencies in time series to achieve modeling of real-time feedback data.

[0039] The structure design of the deep Q network is:

[0040]

[0041] In the formula, Q(s,a) represents the value function of taking action a in state s, s represents the current channel state, including signal quality, interference level and delay, a represents the selected action, r represents the immediate reward, which is used to feedback the success rate of channel transmission and the efficiency of data transmission, γ represents the discount factor, and s ′ It represents the new state after the transition, that is, the update of the channel state after taking action a.

[0042] During initialization, the convolutional layer of the model is designed as follows:

[0043] Number of convolutional layers: at least 18 layers are used to ensure the feature learning ability of the deep network;

[0044] Convolution kernel size: A 3×33×3 convolution kernel can be used to extract local features;

[0045] Activation function: Use ReLU function to enhance nonlinear feature learning;

[0046] The data output by the convolutional layer is connected to the GRU through the fully connected layer to obtain time series feedback.

[0047] The design of GRU includes:

[0048] The input gate, forget gate, and output gate use the Sigmoid activation function to control the flow of information.

[0049] The output and hidden state updates are calculated using the following formula:

[0050]

[0051] In the formula, h t represents the state vector at the current time t, z t Represents the gate variable h at the current time t t-1 represents the state vector at the previous moment t-1, Represents the candidate state vector generated at the current time t, which is usually calculated based on the input information.

[0052] Calculated by the following formula:

[0053]

[0054] In the formula, represents the candidate state vector at the current time t, W h represents the input weight matrix, x t represents the input vector at the current time t, U h represents the state weight matrix, r t Represents resetting the gate variable, and tanh represents the hyperbolic tangent function.

[0055] During execution, the channel information dynamically collected by each satellite includes:

[0056] Signal strength (RSSI): Indicates the received signal strength in dBm;

[0057] Interference: The degree of interference from other signals in the channel;

[0058] Latency: The delay that occurs during data transmission.

[0059] During the merging process, these channel information will be fed back to the deep Q network through experience replay, so that the federated learning model is continuously updated.

[0060] Specifically, real-time feedback data is compared with the expected value of the model output to optimize the network's weight parameters. In this way, the model can continuously adapt to changing channel environments.

[0061] Particle swarm optimization (PSO) algorithm is used for online parameter identification to improve model performance.

[0062] The steps of particle swarm optimization are:

[0063] Randomly initialize the particle swarm, where particles represent learning rate and network weight;

[0064] Calculate the fitness of each particle.

[0065] Throughout the optimization process, particles update their positions and velocities according to their fitness, and eventually, the model parameters will converge to the optimal values ​​to improve channel utilization efficiency.

[0066] Through the implementation of the above technical solution, the federated learning model based on deep reinforcement learning can be directly implemented in specific applications and its practicality can be verified.

[0067] S2. Generate a fractional-order partial differential equation model for describing channel interference based on collected channel information and real-time feedback data, and perform online parameter identification;

[0068] Specifically, first, a fractional-order partial differential equation model is generated based on the collected channel information. The mathematical expression of the model is:

[0069]

[0070] In the formula, represents the α-order partial derivative of the function Ψ with respect to time t, Ψ represents the function to be solved, κ represents the diffusion coefficient, which affects the propagation rate of Ψ in space, and represents the second-order partial derivative of the function Ψ with respect to the spatial variables x and y, β represents a coefficient, represents the first-order partial derivative of Ψ with respect to the spatial variable z, describing the change of Ψ in that direction.

[0071] Secondly, after the model is generated, online parameter identification is performed to optimize the accuracy of the interference model. In this process, real-time channel feedback data is first collected, including signal strength (RSSI), environmental noise, interference signals, and delay information.

[0072] The real-time data collection process specifically includes:

[0073] Each satellite collects the signal strength through the receiving module and records it as RSSI;

[0074] Monitor the interference signal strength, determine the environmental background noise, and discuss algorithms for comparison.

[0075] Then, perform parameter identification through the following steps:

[0076] Establish an empirical data set: Based on the collected real-time feedback data, an empirical data set is constructed to record the channel status and corresponding interference distribution. The structure of the data set includes signal strength, timestamp, and corresponding interference status.

[0077] Application of Particle Swarm Optimization (PSO): In parameter identification, the particle swarm optimization algorithm is used to iteratively adjust κ and β. First, the particle swarm is initialized. The particle positions represent the parameter values ​​of the current model. At the same time, the speed is set to control the parameter update step size.

[0078] Fitness calculation: The fitness function is calculated by optimizing the following formula:

[0079] f(p)=α·r success -β·r errors ;

[0080] In the formula, f(p) represents the fitness function value of the current particle, r success Indicates the ratio of effective received signals, r errorsIt represents the proportion of failures caused by signal interference, and α and β represent weight factors, which adjust the relative importance of success rate and error rate.

[0081] Particle update: According to the fitness calculation results, the position and velocity of the particle are updated. The update formula is as follows:

[0082]

[0083] In the formula, Indicates the velocity of the particle after update, represents the current particle's speed, w represents the inertia weight, c1 and c2 represent learning factors that adjust the particle's learning degree of its own optimal position and the global optimal position, r1 and r2 represent random numbers between 0 and 1, represents the current optimal position of particle i, g best represents the global optimal position.

[0084]

[0085] In the formula, Indicates the velocity of the particle after update, and Represent the updated and current positions of the particles respectively.

[0086] Finally, the changes in channel conditions are dynamically monitored, and the fractional-order partial differential equation model is adjusted according to the updated parameters to improve the accurate description of channel interference. The optimized model is fed back to the deep Q network to enhance its learning ability and enable it to maintain efficient channel utilization in a dynamic environment.

[0087] Through the implementation of the above technical solution, a fractional-order partial differential equation model suitable for specific channel conditions can be directly generated, and an online parameter identification process can be realized.

[0088] S3. Based on the established fractional-order partial differential equation interference model, quantum-classical hybrid optimization is implemented to dynamically optimize the channel state through reconfigurable quantum logic units;

[0089] Specifically, first, a quantum-classical hybrid optimization Hamiltonian model is established to describe the evolution of the system state.

[0090] Its mathematical expression is:

[0091]

[0092] In the formula, represents the Hamiltonian, N represents the number of qubits in the system, and J i represents the channel gain of the i-th qubit, represents the Pauli z operator of the i-th qubit, V ij represents the coupling strength between the i-th and j-th qubits, represents the Pauli x-operator for the i-th qubit.

[0093] Next, a reconfigurable quantum logic unit (rQLU) was constructed. Its hardware structure includes a quantum state initialization module, a quantum gate operation unit, a measurement module, and a quantum operation control module. This logic unit is used to control energy levels and state evolution in the Hamiltonian model, as well as perform preset quantum operations, including quantum search and frequency domain transformation.

[0094] Quantum state initialization: The qubit is initialized to the ground state |0>, and the initial J is set according to the current channel state. i and V ij ;

[0095] Quantum gate operation application: Operate quantum states through reconfigurable quantum gates. The steps include:

[0096] Single-qubit gates: Single-qubit gate operations, such as Hadamard gates and T-gates, are applied to each qubit to introduce a superposition state.

[0097] Two-qubit gate: Apply controlled NOT gate (CNOT) to achieve entanglement and interaction between qubits. Reconfigurable quantum operation:

[0098] Quantum search: Utilizes the Grover algorithm to efficiently search channel states and optimize signal propagation paths.

[0099] Frequency Domain Transformation: Apply the Quantum Fourier Transform (QFT) to quickly convert signals to the frequency domain for spectral analysis and optimization.

[0100] Hamiltonian evolution: The evolution of a quantum system is based on the Schrödinger equation, which is expressed as:

[0101]

[0102] In the formula, i represents the imaginary part in complex number operations, represents the reduced Planck constant, represents the rate of change of quantum state over time, |ψ(t)> represents the quantum state vector at time t, represents the Hamiltonian operator.

[0103] Measurement: After the quantum state evolution is completed, the quantum state information is obtained through the measurement module and converted into a classical signal. The measurement results reflect the changes in the channel state and its impact on signal transmission.

[0104] Optimization process: Optimize based on the existing Hamiltonian and combine it with classical algorithms.

[0105] The specific plan is:

[0106] Input channel information: input the current channel state data obtained by the fractional-order partial differential equation model into the Hamiltonian model;

[0107] Execute quantum optimization algorithms: Use the variational quantum eigensolver (VQE) or quantum annealing algorithm to solve the ground state of the Hamiltonian. The goal is to find the lowest energy state, which is mapped to the optimal channel state.

[0108] Reverse update: Based on the measured channel state, the optimization information is fed back to the fractional-order partial differential equation model to dynamically adjust the channel model parameters.

[0109] Through the implementation of the above technical solutions, combined with the high-speed parallel processing capabilities of quantum computing, the optimization efficiency of channel states can be significantly improved.

[0110] S4. Based on the optimized channel state, adaptively adjust the modulation mode and coding strategy according to the real-time channel quality;

[0111] Specifically, first, a real-time channel quality assessment model is established to obtain the current channel quality index using the following formula:

[0112]

[0113] In the formula, Q represents the channel quality index, N represents the environmental noise power, and I represents the interference signal power.

[0114] Real-time channel monitoring: The signal monitoring module continuously obtains channel status information, regularly calculates the channel quality index Q, and compares it with the preset quality threshold Q. high and Q low Make a comparison.

[0115] Modulation mode selection: Dynamically select the appropriate modulation mode based on the real-time channel quality indicator Q.

[0116] The specific decision-making process is:

[0117] When Q>Q high :

[0118] High-order modulation (such as 64-QAM) is used to support high data rate transmission and is suitable for use in good channel conditions.

[0119] When Q low <Q<Q high :

[0120] Select a medium-order modulation method (such as 16-QAM) to reduce the error rate while ensuring the transmission rate.

[0121] When Q low :

[0122] Low-order modulation methods (such as QPSK) are used to enhance the system's resistance to channel noise and interference to ensure the reliability of data transmission.

[0123] Coding strategy adjustment: After the modulation method is determined, the coding strategy is adaptively adjusted according to the channel conditions.

[0124] The specific strategies are:

[0125] Low channel quality (Q low ):

[0126] Strong redundancy coding methods (such as convolutional coding) are used to enhance the robustness of the signal and reduce the bit error rate by adding redundant bits.

[0127] Medium channel quality (Q low <Q<Q high ):

[0128] Standard coding methods (such as low-density parity check coding, LDPC) are selected to achieve a good balance between performance and complexity.

[0129] High channel quality (Q>Q high ):

[0130] A lighter encoding strategy (such as Turbo encoding) can be selected to reduce latency and improve data transmission efficiency.

[0131] Feedback mechanism: The modulation mode and coding strategy generated based on calculation and decision-making are fed back to the signal transmission module, so that it can update the transmission parameters in real time based on the optimization results.

[0132] The structure of the signal transmission module includes:

[0133] Modulator: converts the input bit stream into the corresponding modulated signal according to the selected modulation method;

[0134] Encoder unit: Executes the selected coding strategy and encodes the modulated signal to improve anti-interference capability.

[0135] Parameter update and optimization: The monitoring module continuously tracks changes in channel quality and inputs the latest channel quality data into the quality assessment model, thereby achieving real-time updates and dynamic adjustments to keep the system in optimal condition.

[0136] Through the implementation of the above technical solution, it is possible to evaluate the channel quality in real time based on the optimized channel status and adaptively adjust the modulation mode and coding strategy to optimize the efficiency and reliability of data transmission.

[0137] ​​S5, based on adaptive adjustment of modulation mode and coding strategy, implements dynamic data transmission, responds to interference changes in real time and optimizes transmission rate, and conducts information feedback and coordination through inter-satellite communication;

[0138] Specifically, first, a dynamic selection principle formula is established to determine the appropriate modulation method. The specific expression is:

[0139]

[0140] In the formula, Modulation represents the selected modulation mode, CQI represents channel quality indicator, BPSK represents binary phase shift keying, QPSK represents quadrature phase shift keying, and 16QAM represents 16-order quadrature amplitude modulation.

[0141] The calculation formula of CQI channel quality indicator is:

[0142]

[0143] In the formula, CQI represents channel quality indicator, SNR represents signal-to-noise ratio, which is obtained by the ratio of received signal strength to noise power, and N th Represents the threshold, which defines the minimum signal-to-noise ratio required under different modulation schemes.

[0144] After the real-time channel quality assessment is completed, dynamic data transmission and interference monitoring are implemented. The specific steps are as follows:

[0145] Real-time monitoring of network interference: The interference monitoring module detects interference in the channel. This module updates interference information in real time based on signal power and noise level. The information can be expressed as:

[0146] I=P interference +P noise ;

[0147] In the formula, I represents the integrated interference power, P interference Represents the measured interference signal power, P noise Represents the ambient noise power.

[0148] Channel parameter adjustment: Dynamically adjust channel parameters based on real-time monitored interference data to optimize transmission efficiency.

[0149] The specific adjustment process is as follows:

[0150] When increased interference is detected, the modulation mode can be reduced to improve the robustness of the transmission;

[0151] Adjust the transmit power to increase the strength of the effective signal. The formula is:

[0152]

[0153] In the formula, SNR represents the signal-to-noise ratio, P signal Indicates the strength of the useful signal transmitted. noise Indicates the background noise intensity in the receiving system, P interference Indicates the strength of other signals that affect signal transmission.

[0154] Dynamic data transmission implementation: After determining the modulation method and coding strategy, data transmission is carried out through the inter-satellite communication network.

[0155] The transfer process includes:

[0156] The data to be transmitted is encoded by the selected encoder to reduce the bit error rate. The encoding method can be convolutional coding, Turbo coding or LDPC coding, depending on the channel quality.

[0157] The coded signal is modulated into the corresponding signal format by a modulator and sent via a satellite link.

[0158] Information feedback and coordination: Through the feedback mechanism between satellites, information on received signals is collected in real time and fed back to the signal processing module at the sending end.

[0159] Feedback includes:

[0160] the quality of the received signal;

[0161] Other relevant status information in the network to optimize overall network performance.

[0162] Feedback-driven dynamic adjustment: Based on feedback information, the scheduling control module can update the modulation method and channel parameters in real time to ensure optimal efficiency of signal transmission under interference conditions.

[0163] The implementation includes:

[0164] By monitoring the downstream channel quality, the CQI is updated in real time and the selected modulation mode is re-evaluated;

[0165] Adjust encoding strategies to improve overall system performance and data transfer rates.

[0166] By implementing the above technical solution, dynamic data transmission can be implemented based on real-time channel quality evaluation, thereby responding to interference changes in the network in real time and optimizing the transmission rate.

[0167] S6. After completing dynamic data transmission, the inter-satellite model is aggregated and updated to optimize the next round of transmission strategy, while also performing performance verification in the process.

[0168] Specifically, first, a formula for dynamic model updating is established to optimize the transmission strategy for the next round.

[0169] The model update formula is:

[0170]

[0171] In the formula, W global represents the global model weight, K represents the total number of satellites, W k represents the local model weight of the k-th satellite, M k represents the mask matrix of the kth satellite, and ⊙ represents the Hadamard product.

[0172] After the model update is completed, the performance verification step is performed.

[0173] This step includes joint satellite-ground testing under simulated interference conditions. The specific steps are as follows:

[0174] Simulate interference environment construction: Use the simulation module to create interference environments under different channel conditions. This environment can be configured using model parameters such as signal power, interference type, and noise level.

[0175] Satellite-ground joint test execution: sending data, receiving signals and evaluating communication effects under simulated conditions.

[0176] The testing process includes:

[0177] Send data under set interference conditions and measure signal quality and bit error rate.

[0178] The formula is:

[0179]

[0180] In the formula, BER represents the bit error rate, E r Indicates the number of error bits received, and N indicates the total number of bits sent.

[0181] Evaluate transmission effectiveness and optimization capabilities: Evaluate the optimization capabilities of the model by comparing test data with preset performance indicators, including transmission efficiency, latency, and error rate.

[0182] The evaluation formula is:

[0183]

[0184] In the formula, Efficiency represents the amount of data successfully transmitted per unit time, D represents the amount of data successfully transmitted, and T represents the total transmission time.

[0185] Model optimization feedback: Feedback the performance verification results to the performance optimization module, and adjust the model parameters W according to the test results. globalAnd provide data support for the next transmission strategy.

[0186] By implementing the above steps, effective data transmission model convergence and optimization can be achieved, and the model effect can be verified through actual satellite-ground joint testing, thereby ensuring that the method has good transmission effect and optimization capability under different channel conditions.

[0187] While embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions, and variations may be made to these embodiments without departing from the principles and spirit of the invention, and that the scope of the invention is defined by the appended claims and their equivalents.

Claims

1. An efficient data transmission method in a BeiDou satellite communication system, characterized in that: The following steps are involved: S1. Initialize and deploy a federated learning model based on deep reinforcement learning on each satellite to collect channel information and real-time feedback data; S2. Generate a fractional-order partial differential equation model for describing channel interference based on collected channel information and real-time feedback data, and perform online parameter identification; S3. Based on the established fractional-order partial differential equation interference model, quantum-classical hybrid optimization is implemented to dynamically optimize the channel state through reconfigurable quantum logic units; S4. Based on the optimized channel state, adaptively adjust the modulation mode and coding strategy according to the real-time channel quality; S5, based on adaptive adjustment of modulation mode and coding strategy, implements dynamic data transmission, responds to interference changes in real time and optimizes transmission rate, and conducts information feedback and coordination through inter-satellite communication; S6. After completing dynamic data transmission, the model between satellites is aggregated and updated to optimize the next round of transmission strategy, while performance verification is performed in the process.

2. The efficient data transmission method in the BeiDou satellite communication system according to claim 1, wherein: The federated learning model of deep reinforcement learning in step S1 is a deep Q network, which includes at least 18 layers of convolutional neural networks and gated recurrent units.

3. The efficient data transmission method in the BeiDou satellite communication system according to claim 1, wherein: In step S2, a fractional-order partial differential equation model is generated to describe the channel interference. The mathematical expression of the fractional-order partial differential equation model is: In the formula, represents the α-order partial derivative of the function Ψ with respect to time t, Ψ represents the function to be solved, κ represents the diffusion coefficient, which affects the propagation rate of Ψ in space, and represents the second-order partial derivative of the function Ψ with respect to the spatial variables x and y, β represents a coefficient, represents the first-order partial derivative of Ψ with respect to the spatial variable z, describing the change of Ψ in that direction.

4. The efficient data transmission method in the BeiDou satellite communication system according to claim 1, wherein: In step S2, the online parameter identification uses a particle swarm optimization algorithm to identify the propagation coefficients κ and β in the interference model.

5. The efficient data transmission method in the BeiDou satellite communication system according to claim 1, characterized in that: In step S3, quantum-classical hybrid optimization is implemented, and the Hamiltonian model of the quantum-classical hybrid optimization is: In the formula, represents the Hamiltonian, N represents the number of qubits in the system, and J i represents the channel gain of the i-th qubit, represents the Pauli z operator of the i-th qubit, V ij represents the coupling strength between the i-th and j-th qubits, represents the Pauli x-operator for the i-th qubit.

6. The efficient data transmission method in the BeiDou satellite communication system according to claim 1, characterized in that: The reconfigurable quantum logic unit in step S3 is used to perform preset quantum operations, including quantum search and frequency domain transformation.

7. The efficient data transmission method in the BeiDou satellite communication system according to claim 1, characterized in that: The adaptive adjustment of the coding strategy in step S4 further includes: selecting convolutional coding, Turbo coding or LDPC coding according to the channel state.

8. The efficient data transmission method in the BeiDou satellite communication system according to claim 1, characterized in that: The dynamic data transmission in step S5 is based on an adaptive modulation scheme and coding strategy. The adaptive modulation and coding mechanism is formed using the dynamic selection principle as follows: In the formula, Modulation represents the selected modulation mode, CQI represents channel quality indicator, BPSK represents binary phase shift keying, QPSK represents quadrature phase shift keying, and 16QAM represents 16-order quadrature amplitude modulation.

9. The efficient data transmission method in the BeiDou satellite communication system according to claim 1, characterized in that: In step S6, the models between satellites are converged and updated, and the model update formula is: In the formula, W global represents the global model weight, K represents the total number of satellites, W k represents the local model weight of the k-th satellite, M k represents the mask matrix of the kth satellite, and ⊙ represents the Hadamard product.

10. The efficient data transmission method in the BeiDou satellite communication system according to claim 1, characterized in that: The performance verification step in step S6 includes a satellite-ground joint test conducted under simulated interference conditions to evaluate the transmission effect and optimization capability under different channel states.

Images