Research On Computation Offloading In Satellite Edge Computing

Satellite communication as a key technology to achieve future 6G air-space integration and seamless global coverage,but traditional satellite communication networks are limited in on-satellite resources,making it difficult to meet user demand for new service quality.Combined with the ground Mobile Edge Computing(MEC)technology,the satellite edge computing scenario is formed to break through the defects brought by satellite relay in the traditional central cloud mode and provide users with low latency and highly reliable computing services through edge satellite nodes.However,how to rationally manage system resources and improve the quality of user experience in the satellite edge scenario with multi-layer and heterogeneous computing and storage resources has become a new challenge.Therefore,this thesis investigates the edge computing core problem of computation offloading and resource allocation in the satellite edge computing scenario,which is mainly as follows.First,this thesis analyzes the satellite edge node characteristics for the satellite edge computing scenario and then models the system to represent the joint optimization of delay and energy consumption as a mixed integer nonlinear programming problem.The problem is split into an optimal resource allocation problem and a computation offloading decision problem.The resource allocation subproblem is solved by the algorithm of Lagrange multiplier and dichotomous search,while the computation offloading subproblem is represented as a noncooperative game among multiple users,and the existence of Nash Equilibrium(NE)is proved by the potential game.Finally,a game theory-based joint resource allocation algorithm for computation offloading is proposed,and it is verified through simulation that the algorithm can effectively reduce the system latency and energy consumption under different number of users,task computation intensity and satellite computation capacity,and still maintain good computation offloading performance and task success rate under multi-user competition.Secondly,in the face of complicated and higher computational intensity task land requests,the satellite cloud computing nodes are added,forming a satellite cloud edge collaborative computing scenario.A deep reinforcement learning based offloading decision and resource allocation strategy is proposed to minimize the computational cost in the context of multi-layer satellite cloud edge collaboration scenario with different computational capabilities,high dynamics of satellite nodes and fast channel fading.The strategy uses empirical playback with randomly selected samples to speed up the network update and accelerates the convergence of the algorithm by setting up an order-preserving quantization and a dynamic quantization of K values.Finally,the effectiveness of the proposed algorithm is verified by simulation results,and the comparison verifies that the algorithm can reduce the complexity and operation cost while ensuring the optimal average computational cost,which is more suitable for highly dynamic cloud-side cooperative satellite communication network systems.