Research On Doppler Frequency Offst Problem In LEO-MIMO Satellite Systems

Satellite communication sytems can provide cost-effective,seamless and ubiquitous service which contribute to near 100% geographical coverage in 6G communication network.However,due to the features of high-speed and resource-constraint in the satellite system,the problems of Doppler frequency offset(DFO),resource allocation and handover management will affect the performance of the system.Considering the impact of DFO in satellite communication system,the spectrum efficiency(SE)performance is analyzed and optimized in this paper.Meanwhile,considering the problem of resource allocation and handover management in the multi-satellite collaboration system,the paper solves such a mixed variable optimization problem by using deep reinforcement learning(DRL)algorithm.The simulation results show that the proposed algorithms significantly improve system performance.Firstly,two types of satellite channel model are introduced,which are deterministic channel modeling and random channel modeling.Several Doppler frequency offset compensation algorithms for satellite systems are studied.Through numerical simulations,the performances between different DFO compensation algorithms are compared.Then,some typical channel estimation methods are introduced,including pilotbased estimation,compressed sensing-based estimation and angle domain estimation.Next,we investigate the impact of DFO in low earth orbit multiple-input and multipleoutput(LEOMIMO)satellite communication systems.Considering the impact of DFO,we conduct the minimum mean squared error(MMSE)channel estimation and derive the closed-form expressions for the downlink achievable rates with maximum ratio transmission(MRT)and maximum average signal-to-leakage-plus-noise ratio(MASLNR)precoding.In order to relieve the impact of DFO,an adaptive DFO compensation algorithm based on historical data is designed.The algorithm is divided into pre-compensation stage and fine compensation stage.we can acquire value of satellite DFO based on the satellite’s orbit information and user’s location information in pre-compensation stage,which is low-complexity and easy to accomplish fastly.Beam alignment(BA)algorithm based on oversampling FFT is adopted in fine compensation,which can achieve high estimation accuracy.Two stages of proposed algorithm are beneficial to realize the balance of complexity and accuracy.Numerical results verify the effectiveness and robustness of proposed DFO compensation algorithm.Finally,we investigate the problem of power allocation,user selection and handover mangement in multisatellite collaboration system.Considering the impact of DFO,the expression of downlink sum achieveable rate,spectral efficiency(SE)and energy effieciency(EE)in multi-satellite collaboration system are derived.To maximize the sum achievable rate of system,we establish the multi-constrained function and mixed variables optimization problems.Based on the Markov Decision Process process(MDP),the observation space,action space and reward function of agents are created.Then,we proposed a design of joint optimized user selection and power allocation algorithm based on multi-agent deep deterministic policy gradient(MADDPG).Considering the problem of power allocation and handover management in dynamic multi-satellite collaboration system,to maximize sum EE in service period,we proposed a design of joint optimized handover management and power allocation algorithm based on MADDPG.Compared with the genetic algorithm(GA)and random algorithm,the design of joint optimized user selection and power allocation algorithm based on MADDPG achieves a better performance.