Computing And Communication Resources Optimization For Low Earth Orbit Satellite Network

As an important supplement to terrestrial network,Low-Earth-Orbit(LEO)satellite communication system has the advantages of wide coverage,free from geographical and weather conditions,and strong damage resistance,etc.,which is of great significance to the future development of communication system.However,due to the deployment characteristics of satellite,on-board resources are limited.Incorporating Mobile Edge Computing(MEC)technology into the LEO satellite network can effectively make up for the shortage of resources on the satellite,reduce the service response time,and improve the quality of communication service.However,in the satellite communication system based on MEC,how to efficiently manage the communication and computing resources of the system and improve the energy efficiency of the network are the key problems to be solved urgently.Usually,there will be multiple satellites coverage the user,so selecting the appropriate satellite for association can provide users with more sufficient resources.And due to the high-speed movement of LEO satellites,the satellite handover will also affect the performance of the system.Moreover,since there are multiple satellites and gateways in the system,efficient offloading and scheduling decision can reduce the latency and energy consumption generated by processing tasks.And for tasks offloading to the same node,efficient resource allocation approach can reduce latency of tasks and improve the resource utilization.Therefore,this paper will study the wireless resource management of the LEO satellite network from the aspects of the user association,satellite handover,offloading and scheduling decision of tasks,and the resource allocation.Firstly,for the quasi-static LEO satellite IoT system,the joint optimization problem of latency and energy consumption for a single batch task is formulated as a dynamic mixed-integer programming problem.And this problem is decomposed into two sub-problems.The first one is optimal resource allocation problem,and the second one is joint user association and offloading problem.This paper proposed a Lagrange multiplier-based algorithm for resource allocation sub-problem.The second sub-problem is further modeled as a Markov Decision Process(MDP),and an algorithm based on Deep Reinforcement Learning(DRL)is proposed to tackled this sub-problem.Simulation results show that the proposed algorithm can adapt to the dynamic system environment,and reduce the latency and energy consumption of the system efficiently.Secondly,for the dynamic LEO satellite constellation,in order to solve the cooperative processing problem of the continuously arriving tasks,this paper decomposed it into two sub-problems.The first one is fair resource allocation problem,which is formulated as a max-min fairness problem and solved by Dual Ascent method.The second one is dynamic task offloading and scheduling decision problem,which is modeled as an MDP and solved by DRL based algorithm.Simulation results show that the proposed algorithm can continuously and efficiently make decisions for the collaborative processing of LEO satellite constellation,and improve the task completion rate of the system.The research results of this paper provide a new idea for wireless resource management of LEO satellite network,and have reference value to the planning and construction of LEO satellite network in the future.