Research On Mobile Satellite Offloading Strategies Based On Edge Computing

With the rapid development of the Internet of Things,cloud computing faces challenges such as the difficulty of meeting users’ demands for large-scale data processing and real-time requirements,necessitating the use of auxiliary computing methods to optimize computing speed.Edge computing effectively reduces computing costs by offloading computing tasks to other devices within the access network.However,due to the difficulty or operational cost of establishing ground-based network infrastructure,the coverage of traditional ground-based networks is limited,resulting in a situation where task computing in some areas cannot meet the user’s latency requirements or cannot be performed at all.Satellite mobile edge computing serves as an effective complement to and extension of ground-based network computing,solving these problems.Therefore,this thesis proposes a satellite-land edge computing network architecture using satellites as edge nodes to provide auxiliary computing.By deploying satellite edge computing servers on satellites,the architecture provides computing services for ground-based user devices performing compute-intensive tasks.In this architecture,when the terminal computing capability of the user is insufficient or the coverage of nearby ground-based servers is inadequate,the user can choose to offload their tasks to one or more satellites for auxiliary computing.Although research on deploying computing servers at ground-based edge nodes or conducting auxiliary computing using intermediate platforms has been widely conducted,there are still many challenges to task scheduling,different user latency requirements,and system power consumption when computing offloading is performed without cloud server auxiliary computing.Since these problems are difficult to find an optimal offloading strategy in a short time when the number of tasks and devices increases.To solve these problems,multiple tasks are selected to be offloaded to a satellite or to remain for local computing as ordered data pairs in this thesis.By using a genetic algorithm-based offloading algorithm for cross-mutation and system performance indicator calculation,the task allocation strategy,offloading location adjustment,and offloading order are optimized to solve the latency,feedback,and energy consumption issues encountered during offloading.The time and energy required for offloading are weighted as a system performance indicator to achieve the best latency and energy consumption balance,and to optimize this indicator to achieve the lowest latency and energy consumption.Finally,this thesis validates the effectiveness and feasibility of satellite-land edge computing through experimental evaluations.From our experimental results we know that compared with traditional ground edge computing,satellite-land edge computing has lower latency and higher computing efficiency and can solve the limitations and problems of traditional ground edge computing.At the same time,the application of genetic algorithm in offloading tasks is also verified,as it can effectively optimize offloading strategies,improve offloading efficiency,and provide an effective processing method for computationally intensive tasks generated by mobile devices,with broad application prospects.