Joint Task Offloading and Resource Assignment in Satellite Edge Computing Networks

As effective complements to terrestrial edge computing, satellite edge computing networks (SECNs) can deliver reliable communication and computing services by deploying computing resources on the satellites. This paper addresses the joint task offloading and resource assignment problem in an SECN that processes batch-processing tasks offloaded from ground users through inter-satellite collaboration. The optimization goal is to minimize the average response time of usersubmitted tasks with the constraints of the total amounts of energy and resource. We formulate the joint optimization problem as an NP-hard integer programming and decouple it into an offloading decision and a resource assignment subproblem. For resource assignment, we design a cost-effective task assignment policy to determine the execution order of tasks offloaded to the satellite and provide a theoretical analysis to justify the proposed assignment policy. We devise a genetic metaheuristic that incorporates the task assignment policy to evaluate the fitness value of each offloading solution and to iteratively explore better solutions. The end result of the proposed metaheuristic is the most appropriate offloading solution leading to the shortest average response time. The evaluation results on testing instances of different scales demonstrate the superiority of our proposed approach over baseline methods.