The Research On Intelligent Resource Allocation Algorithm Based On Edge Computing In Leo Satellite Network

The low earth orbit(LEO)satellite network,with low delay and high stability,plays an important role for seamless global coverage in the future network.With the increase in the variety and complexity of services,especially the development and application of artificial intelligence,there are highly concurrent and diverse services.The traditional LEO satellite network is difficult to satisfy the requirement of intelligent services.Deploying multi-access edge computing(MEC)server in the LEO satellite increases the computing and caching capabilities of the network,which reduces the delay of services and expands the application scenarios of the satellite network.However,the communication and energy resources of the LEO satellite are limited,and the computing and caching resources of the MEC server are also limited.Therefore,it is necessary to study the intelligent resource allocation algorithm to improve the resource utilization rate in the LEO satellite network and satisfy the requirement of various services.As launch and maintenance costs of the satellite limit energy,computing,caching resources in the LEO satellite network,the number of computing tasks that the satellite can support is limited.The existing works on resource allocation in the LEO satellite network focus on the limitation of satellite resources and the diversification of service requirement.In practice,the mobility of the LEO satellite causes fluctuation in satellite energy,leading to changes in the capability to support computing services.Time-varying network capacity increases the complexity of resource allocation algorithm.In addition,when there is various access equipment such as satellites and base stations in the network,it is also necessary to carry out joint allocation for communication and computing resources in heterogeneous nodes.Hence,this thesis studies the intelligent resource allocation algorithm based on MEC in LEO satellite network,considering energy fluctuation caused by satellite mobility and the heterogeneity of the access equipment.The main work of this thesis is summarized as follow:Considering the energy fluctuation caused by satellite mobility,this thesis proposes an energy-aware task offloading and resource allocation algorithm to reduce the energy consumption of satellites.To solve the problem of limited resources of a single satellite,the cooperation mechanism with the stable intra-orbit inter-satellite links for inter-satellite cooperation and the satellite-terrestrial links for cloud-edge cooperation is adopted,which increases the available resources in the network.Analyzing the energy collected and consumed by the satellite during its periodic flight,the satellite energy aware model is established.The task offloading and resource allocation problem is proposed by combining the energy aware results and the delay of task offloading,with the aiming at minimizing the satellite energy consumption.The problem is solved by the improved non-dominated sorting genetic algorithm Ⅱ,where the constraint comparator is proposed to make feasible solutions satisfy trade-off constraints.The simulation results show that the proposed algorithm effectively reduces the satellite energy consumption.Considering the differences of resources and performance in heterogeneous equipment,this thesis investigates the joint communication,computing,and caching resource allocation problem with user equipment cooperation to reduce delay and improve the resource utilization.Considering the service caching environment and overhead of various computing services on MEC servers,the task offloading model and the service caching model are established.Then the strategy that the executive codes of services are transmitted by user equipment cooperation is proposed to reduce the transmission of repeated data in the network.By analyzing the resource utilization rate in the network,the communication,computing,and caching resource allocation problem is proposed with the target of minimizing the average delay of tasks and maximizing the utilization of resources.The above problem is formulated as a mixed-integer nonlinear programming problem and solved by genetic algorithm.The approximate optimal solution is obtained by searching for feasible solutions with limited iterations.Simulation results show that the proposed algorithm can reduce the computing resources required to execute the tasks and the average delay of tasks.