Research On Task Offloading And Service Placement Strategy Based On Satellite Edge Computing

Low Earth Orbit(LEO)satellites are an important part of 6G key technologies,which can provide more comprehensive network coverage for mobile communications.However,traditional satellite network resources are limited and it is difficult to meet the quality of user experience(QoE)for computing-intensive tasks.Referring to terrestrial mobile edge computing,cloud computing resources are deployed to LEO satellites to form satellite edge computing nodes,which can provide users with low-latency and highly reliable computing services.However,the high dynamics and limited resources of LEO satellites make dynamic task offloading and service placement in satellite.edge computing scenarios challenging.Based on deep reinforcement learning algorithms,this paper investigates task offloading and service placement strategies in satellite edge computing scenarios from different perspectives,and the main research contents and results are as follows.(1)Research on cache-assisted task offloading and resource allocation strategies for QoEAiming at the limited service of traditional LEO satellites,this paper constructs a edge computing architecture based on LEO satellite network.By introducing computing and storage resources at the edge of LEO satellites,the joint optimization problem of cache-assisted task offloading and resource allocation is studied.According to the differences in user preferences,this paper formulates the optimization objective as the problem of maximizing user QoE-aware utility.In order to solve the above problems,a cache-assisted task offloading and resource allocation algorithm based on deep reinforcement learning is proposed,which decouples the target problem into multiple sub-problems for solution.Experimental results show that the strategy has good performance in terms of user QoE.(2)Research on task offloading and service placement strategy for LEO satellite service coverageThe high-speed movement of LEO satellites makes the ground coverage time shorter,so service deployment at the edge of LEO satellites faces great challenges.In this paper,the joint optimization problem of task offloading and service placement for LEO satellites is studied by introducing a constellation of LEO satellites with cooperative services to achieve cost-aware service coverage maximization.Facing the challenges of time-varying satellite topology and high dynamics of LEO satellites,a task offloading and service placement algorithm based on multi-agent reinforcement learning is proposed,which divides the target problem into two sub-problems for solution.The simulation results show that the scheme can maximize LEO satellite service coverage under the limitation of service cost.