Research On Dynamic Beam Scheduling Technology In Multi-beam Satellite Communication System

Satellite communication system,as a supplement to the infrastructure of terrestrial communication networks,has been widely used by researchers because of its strong global coverage,long communication distance,high system capacity,resistance to major natural disasters,and the provision of fixed and mobile communication services.Value and support from the state.With the increasing demand for its capacity and the continuous consumption of spectrum resources,a multi-beam satellite communication system is proposed,which uses multiple high-gain narrow beams to cover a large area together,which can effectively improve spectrum resource utilization and system performance.However,the more spot beams a satellite provides,the more transmitters are required,but the onboard equipment is very limited and expensive.The dynamic beam scheduling technology can effectively solve the problem by using a small number of beams to cover multiple cells by time division multiplexing,but different beam scheduling strategies directly affect the transmission delay and throughput of the data packet.Therefore,how to improve system performance through dynamic beam scheduling technology is an urgent problem to be solved.In this regard,the paper has conducted the following two studies.First,the beam scheduling technique based on the steady-state time average is studied.When a small number of beams cover multiple cells,how to set the traffic backoff parameter of each cell to reduce system congestion and improve throughput,by replacing the binary variable indicating whether the cell obtains the beam with the steady state time average,The beam scheduling problem is modeled as a double nested linear programming problem constrained by average latency,channel capacity,traffic size,number of beams,and user fairness to maximize system throughput.The simulation results show that compared with the existing algorithms,the system throughput is significantly improved while ensuring the user's fairness.Secondly,the dynamic beam scheduling technique based on deep reinforcement learning is studied.In the beam scheduling technique based on the steady-state time average,the results obtained by linear programming only apply to the current state,and once the environment changes slightly,it needs to be recalculated,and the delay optimization is not considered.Aiming at the above problems,this paper combines the ability of deep learning to extract the characteristics of channel capacity,user traffic and delay,and the characteristics of enhanced learning for beam scheduling decision.A dynamic beam scheduling algorithm based on deep reinforcement learning is proposed to minimize all The average delay of the cell transmission packet and increase the system throughput.Firstly,the dynamic beam scheduling problem is modeled as a Markov decision process,which accurately characterizes the correlation between decisions at each moment,then reconstructs the state of the environment into multi-dimensional tensors,characterizes the spatio-temporal features of the traffic,and then passes the volume.The neural network extracts relevant features and performs beam scheduling decisions.This part of the content has a generalized beam scheduling strategy,which can adapt to the dynamic changes of the environment in real time,and brings more flexible solutions to the dynamic resource scheduling problem.The simulation results show that compared with the existing algorithms,the proposed algorithm can effectively reduce the packet transmission delay and improve the system throughput.