Intelligent spectrum allocation and optimization method in satellite communication system

Abstract

The invention provides an intelligent spectrum allocation and optimization method in a satellite communication system, relates to the technical field of satellite communication, and solves the problems of low utilization rate, serious interference and poor dynamic adaptability in traditional spectrum allocation. An intelligent spectrum allocation and optimization method in a satellite communication system comprises the following steps: S1, multi-dimensional spectrum state real-time sensing and modeling: acquiring satellite spectrum occupation, interference temperature, service requirements and position information, and constructing a space-time-frequency three-dimensional state tensor; and S2, intelligent allocation decision based on deep reinforcement learning: defining an allocation action space and a reward function, and generating an optimal spectrum allocation strategy by adopting an attention enhancement DQN network. According to the invention, the dynamic response is enhanced, the burst service distribution delay is less than 1s, the service blocking rate is reduced from 25% to 4%, and the emergency communication guarantee capability is obviously improved; and cross-layer collaborative optimization: realizing end-to-end joint optimization from spectrum allocation to power control in combination with satellite orbit parameters and service QoS (Quality of Service) requirements.

Description

Technical Field

[0001] This invention belongs to the field of satellite communication technology, specifically a smart spectrum allocation and optimization method applicable to hybrid satellite communication systems of low Earth orbit (LEO) and geostationary orbit (GEO). Background Technology

[0002] With the continuous development of satellite communication technology, the services carried by satellite communication systems are becoming increasingly diverse, and users' demands for communication services are becoming more and more diversified, including time-sensitive services such as high-definition video transmission and real-time data interaction. However, the spectrum resources in satellite communication systems are limited. How to meet the growing service demands and improve the utilization efficiency of spectrum resources under limited spectrum resources has become an urgent problem to be solved.

[0003] Based on the above, the existing allocation method has the following drawbacks: 1. Static allocation is inefficient: Pre-allocation strategies based on fixed frequency bands, such as the FCC satellite spectrum allocation rules, have a spectrum utilization rate of only 40%-50%. In scenarios where coverage areas change due to satellite movement, spectrum congestion and resource idleness often coexist. 2. Lack of interference coordination mechanisms: Inter-satellite interference (ACLI) between satellites sharing the same frequency, such as between Starlink and OneWeb systems, is not effectively suppressed. When the distance between satellites is less than 500 km, the bit error rate (BER) can rise to 10. -3 above; 3. Insufficient dynamic adaptability: Traditional algorithms, such as greedy allocation and polling mechanisms, have update cycles of ≥10 seconds, making them unable to respond to sudden business demands, such as the real-time bandwidth requirements of disaster emergency communications, resulting in a service congestion rate as high as 25%. Summary of the Invention

[0004] The purpose of this invention is to address the above-mentioned problems by providing an intelligent spectrum allocation and optimization method in a satellite communication system. Through multi-dimensional spectrum state perception, reinforcement learning algorithms, and dynamic interference coordination mechanisms, this invention solves the problems of low utilization, severe interference, and poor dynamic adaptability in traditional spectrum allocation, achieving efficient utilization of spectrum resources and reliable communication. This results in a spectrum utilization rate increase of over 60%, co-channel interference suppression of over 3dB, and a service blocking rate reduction to below 5%.

[0005] To achieve the above objectives, the present invention provides the following technical solution: a method for intelligent spectrum allocation and optimization in a satellite communication system, comprising the following steps: S1, Real-time perception and modeling of multi-dimensional spectrum status: Collect satellite spectrum occupancy, interference temperature, service requirements and location information, and construct a spatiotemporal frequency three-dimensional state tensor; S2, intelligent allocation decision based on deep reinforcement learning: define the allocation action space and reward function, and use attention-enhanced DQN network to generate the optimal spectrum allocation strategy; S3, distributed interference coordination and power control: establish an interference temperature model, and realize the collaborative optimization of the transmission power between the same frequency satellites through game theory algorithm; S4, dynamic spectrum re-allocation and strategy update: re-allocation is triggered according to spectrum fluctuation, satellite state change or burst traffic, and incremental update of the strategy is realized by using rolling time domain optimization.

[0006] Further, the spectrum state collection accuracy in S1 meets: signal power spectral density resolution 0.5dBm / Hz, satellite position error ≤10m, and service demand parameter real-time reporting frequency ≥10Hz.

[0007] Further, the reward function in S2 includes multi-objective optimization of spectrum efficiency, same frequency interference and service blocking, and the allocation strategy meets the constraint of single satellite exclusive occupation of same frequency sub-band.

[0008] Further, in S3, the interference temperature model considers the transmission power of adjacent satellites, path loss and environmental noise, and the power control algorithm realizes information interaction between satellites through a consistency protocol with a communication period ≤200ms.

[0009] Further, in S4, the dynamic re-allocation trigger conditions include spectrum occupancy fluctuation >20%, satellite node change or burst traffic bandwidth demand >30%, and re-allocation time ≤1s.

[0010] Further, the steps S1-S4 are cooperatively operated to realize the following effects: spectrum utilization rate is improved by 50%-70%, same frequency interference is suppressed by more than 3dB, and burst traffic response time is ≤1s.

[0011] Compared with the prior art, the beneficial effects of the present application are as follows: 1. Spectrum efficiency breakthrough: intelligent allocation increases the spectrum utilization rate from 50% to 85%, and the average throughput of a single satellite increases by 1.8 times; 2. Effective interference suppression: distributed power control reduces the same frequency interference by 3.5dB, and the bit error rate is stabilized at 10 -6 Below; 3. Enhanced dynamic response: burst traffic allocation delay <1s, service blocking rate reduced from 25% to 4%, and emergency communication guarantee capability significantly improved; 4. Cross-layer cooperative optimization: combined with satellite orbit parameters and service QoS requirements, end-to-end joint optimization from spectrum allocation to power control is realized. BRIEF DESCRIPTION OF DRAWINGS

[0012] Figure 1This is a schematic diagram of the overall framework structure of the present invention; Figure 2 This is a schematic diagram of the S1 multi-dimensional spectrum state real-time perception and modeling process structure of the present invention; Figure 3 This is a schematic diagram of the intelligent allocation decision-making process based on deep reinforcement learning in S2 of the present invention. Figure 4 This is a schematic diagram of the S3 distributed interference coordination and power control process structure of the present invention; Figure 5 This is a schematic diagram of the S4 dynamic redistribution process structure of the present invention; Figure 6 This is a schematic diagram of the process structure for improving spectrum utilization in this invention. Detailed Implementation

[0013] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention.

[0014] Example: Figures 1 to 6 As shown, this invention provides an intelligent spectrum allocation and optimization method in a satellite communication system, comprising the following steps: S1, Real-time perception and modeling of multi-dimensional spectrum status: Collect satellite spectrum occupancy, interference temperature, service requirements and location information, and construct a spatiotemporal frequency three-dimensional state tensor; S2, Intelligent allocation decision based on deep reinforcement learning: Define the allocation action space and reward function, and use an attention-enhanced DQN network to generate the optimal spectrum allocation strategy; S3, Distributed Interference Coordination and Power Control: Establish an interference temperature model and use game theory algorithms to achieve coordinated optimization of transmit power among satellites at the same frequency; S4, Dynamic Spectrum Reallocation and Policy Update: Based on spectrum fluctuations, changes in satellite status, or sudden service triggers, the policy is updated incrementally using rolling time-domain optimization.

[0015] Among them, the spectrum status acquisition accuracy in S1 meets the following requirements: signal power spectral density resolution 0.5dBm / Hz, satellite position error ≤10m, and real-time reporting frequency of service requirement parameters ≥10Hz.

[0016] The reward function in S2 includes multi-objective optimization of spectral efficiency, co-channel interference, and service blocking, and the allocation strategy satisfies the single-star exclusive constraint of the co-channel sub-band.

[0017] Among them, the interference temperature model in S3 comprehensively considers the transmission power of neighboring satellites, path loss and environmental noise, and the power control algorithm realizes information exchange between satellites through a consensus protocol, with a communication cycle of ≤200ms.

[0018] The dynamic re-distribution trigger condition in S4 includes a spectrum occupancy fluctuation > 20%, a satellite node change, or a burst traffic bandwidth demand > 30%, and a re-distribution time ≤ 1s.

[0019] The steps S1-S4 are cooperatively operated to achieve a spectrum utilization rate improvement of 50%-70%, a same-frequency interference suppression of more than 3dB, and a burst traffic response time ≤ 1s.

[0020] The specific use mode and role of the embodiment are as follows: S1, multi-dimensional spectrum state real-time perception and modeling; State parameter collection: At the satellite end, the signal power spectral density of the current frequency band is collected in real time through the on-board spectrum sensor, with an accuracy of 100 kHz and a PSD resolution of 0.5 dBm / Hz. At the same time, the satellite position is acquired by using GNSS positioning, with an error ≤ 10 m, and the transmit power margin is monitored in real time by the power amplifier, with an accuracy of 1%. At the ground end, the service demand reported by the user terminal, the bandwidth B{req}, the time delay tolerance T{tol}, and the bit error rate requirement P{ber} are received, and the regional interference temperature data are acquired by the interference monitoring station, with a resolution of 2 GHz. Three-dimensional model construction: the spectrum resources are divided into F sub-bands, with a bandwidth of 5-200 MHz adjustable, and a space-time frequency state tensor {S}(t,f,s) is constructed:

[0021] Wherein P{occ}(f) is the sub-band occupancy rate, I{temp}(f,s) is the interference temperature of the sub-band f at the satellite s, and {Pos}(s) is the satellite position coordinate. S2, intelligent allocation decision based on deep reinforcement learning; Action space definition: the allocation strategy π(f,s) is a binary matrix, and π(f,s)=1 indicates that the sub-band f is allocated to the satellite s, with the constraint condition being:

[0022] Reward function design:

[0023] Wherein is the sub-band spectrum efficiency (bps / Hz), I{co} is the same-frequency satellite-to-satellite interference strength, B{block} is the service blocking number, is a weight coefficient; Network architecture: a DQN network enhanced by an attention mechanism is adopted, the input layer is the state tensor S, the output layer is the allocation strategy π, the experience replay buffer capacity is 10 5 , and the target network update period is 500 steps. S3, Distributed Interference Coordination and Power Control; Interference temperature modeling: The interference temperature of satellite s to sub-band f is calculated as:

[0024] Where P_{s'} is the transmit power of the neighboring satellite, G_{s's} is the gain, L_{s's} is the path loss, and T_{env} is the environmental thermal noise. Power optimization algorithm: A distributed game theory model is used, with each satellite aiming to maximize its spectral efficiency while minimizing the interference to neighboring satellites. The power control vector {P}^* is iteratively solved:

[0025] Where is the interference penalty factor, and the information exchange between satellites is achieved through a consensus algorithm with a communication period of ≤200ms. S4, Dynamic Spectrum Re-allocation and Strategy Update; Triggering conditions: Re-allocation is initiated when the following conditions occur: a. Sub-band occupancy fluctuation exceeds threshold, for 1s; b. New satellite launch or old satellite decommission, detected through inter-satellite link, response time ≤5s; c. Sudden traffic request, bandwidth demand > 30% of current available resources; Incremental update: Rolling horizon optimization (MPC) technology is used to re-allocate affected sub-bands based on retaining 80% of the existing allocation strategy, with a re-allocation time of ≤1s; Strategy iterative optimization: The ground control center collects network spectrum usage data daily, updates network parameters through offline reinforcement learning (PPO algorithm), and forms periodic strategy optimization.

[0026] In the specific implementation, taking the Ku band 10.7-12.7GHz allocation of the Starlink-GEO satellite hybrid system as an example: 1. State acquisition: LEO satellites report their current positions every 500ms, synchronize through inter-satellite laser links, and report sub-band occupancy. There are 128 sub-bands, each with a bandwidth of 15.625MHz. The ground station collects a sudden video traffic demand bandwidth of 1.2GHz in a certain area with a time delay of ≤100ms; 2. Intelligent allocation: The DQN network detects that there are 4 idle sub-bands in the 10.9-11.5GHz frequency band, with a total bandwidth of 937.5MHz. These sub-bands are preferentially allocated to the 3 LEO satellites closest to the service area, avoiding the same frequency working frequency band of GEO satellites; 3. Interference coordination: adjacent LEO satellites adjust the transmission power through a consistency algorithm to control the interference temperature below-150dBm / Hz, ensuring the signal-to-noise ratio SNR≥20dB; 4. Dynamic adjustment: when one of the LEO satellites enters the shadow area, causing insufficient power margin, the system will transfer the satellite-borne services to the adjacent satellite within 500ms, and the service interruption time during spectrum reallocation is less than 10ms.

Claims

1. A method for intelligent spectrum allocation and optimization in a satellite communication system, characterized in that: Includes the following steps: S1, Real-time perception and modeling of multi-dimensional spectrum status: Collect satellite spectrum occupancy, interference temperature, service requirements and location information, and construct a spatiotemporal frequency three-dimensional state tensor; S2, Intelligent allocation decision based on deep reinforcement learning: Define the allocation action space and reward function, and use an attention-enhanced DQN network to generate the optimal spectrum allocation strategy; S3, Distributed Interference Coordination and Power Control: Establish an interference temperature model and use game theory algorithms to achieve coordinated optimization of transmit power among satellites at the same frequency; S4, Dynamic Spectrum Reallocation and Policy Update: Based on spectrum fluctuations, changes in satellite status, or sudden service triggers, the policy is updated incrementally using rolling time-domain optimization.

2. The intelligent spectrum allocation and optimization method in a satellite communication system according to claim 1, characterized in that: The spectrum status acquisition accuracy in S1 meets the following requirements: signal power spectral density resolution 0.5dBm / Hz, satellite position error ≤10m, and real-time reporting frequency of service requirement parameters ≥10Hz.

3. The intelligent spectrum allocation and optimization method in a satellite communication system according to claim 1, characterized in that: The reward function in S2 includes multi-objective optimization of spectral efficiency, co-channel interference, and service blocking, and the allocation strategy satisfies the single-star exclusive constraint of the co-channel sub-band.

4. The intelligent spectrum allocation and optimization method in a satellite communication system according to claim 1, characterized in that: The interference temperature model in S3 comprehensively considers the transmission power of neighboring satellites, path loss, and environmental noise. The power control algorithm realizes information exchange between satellites through a consensus protocol, with a communication cycle of ≤200ms.

5. The intelligent spectrum allocation and optimization method in a satellite communication system according to claim 1, characterized in that: The dynamic reallocation triggering conditions in S4 include spectrum occupancy fluctuation > 20%, satellite node change or sudden service bandwidth demand > 30%, and reallocation time ≤ 1s.

6. The intelligent spectrum allocation and optimization method in a satellite communication system according to claim 1, characterized in that: The steps S1-S4 work together to achieve a 50%-70% increase in spectrum utilization, a reduction of more than 3dB in co-channel interference, and a burst service response time of ≤1s.

Images