摘要
星地卸载场景中,地面节点集向低轨卫星边缘服务器卸载计算任务时会导致传输能耗高的问题。为了提高节点集续航时间并有效利用网络中的计算资源,结合D2D(device-to-device)通信提出一种双边协作的部分计算卸载方案。考虑到计算任务的处理时延与低轨卫星边缘服务器的负载均受限,将优化问题建模为最小化地面节点集卸载过程中的总能耗,用改进的遗传算法将种群中个体按适应度值大小降序排列并依次两两交叉,且赋予适应度越小的个体越高的变异概率。仿真结果表明,改进遗传算法在不同的时延要求与任务节点数下均良好地收敛,有效减少了卸载过程中地面节点集的总能耗。
Aiming at the problem that high transmission energy consumption caused by offloading computation tasks from ground nodes set to low earth orbit satellite edge server in the satellite-ground computation offloading scenario.In order to improve the battery life of the nodes set and effectively utilize the computation resources in the network at the same time,a partial computation offloading scheme of bilateral cooperation is proposed combined with device-to-device communication.Considering that the processing delay of computation tasks and the load of low earth orbit satellite edge server are limited,the optimization problem is modeled as minimizing the total energy consumption of the ground nodes set during the offloading process,and the improved genetic algorithm is used to solve the problem.The algorithm arranged the individuals in the population by descending order according to the size of fitness value and crossed them in pairs,and give the individuals with smaller fitness higher mutation probability.The simulation results show that the improved genetic algorithm converges well under different delay requirement and task node numbers,and effectively reducing the energy consumption of ground nodes set during the offloading process.
作者
裴雪松
王俊义
李燕龙
PEI Xuesong;WANG Junyi;LI Yanlong(School of Information and Communication,Guilin University of Electronic Technology,Guilin 541004,China;Guangxi Key Laboratory of Wireless Wideband Communication and Signal Processing,Guilin University of Electronic Technology,Guilin 541004,China)
出处
《桂林电子科技大学学报》
2023年第2期149-156,共8页
Journal of Guilin University of Electronic Technology
基金
国家自然科学基金(61966007)
广西研究生教育创新计划(YCBZ2022106)。
关键词
边缘计算
D2D
部分卸载
改进遗传算法
edge computing
D2D
partial offloading
improved genetic algorithm
