超密集边缘计算网络中面向能耗优化的任务卸载方法

Task Offloading Method for Energy Consumption Optimization in Ultra-Dense Edge Computing Network

传统网络架构部署下的边缘服务器难以满足大规模用户设备的接入和通信质量要求。为增加网络容量,提高频谱利用率,通过密集化基站的部署,构建一种面向超密集边缘计算网络的任务卸载优化模型。面对信道状态的变化、移动设备的动态需求以及服务器和频谱资源的有限性对任务卸载带来的挑战,结合任务类型和服务器的计算能力,并考虑信道状态变化、移动设备的动态需求以及干扰约束对卸载策略的影响,提出一种基于自适应模拟退火遗传(AGASA)算法的任务卸载方法,在满足任务截止期限的同时,对任务卸载能耗进行优化。同时,为得到最优上传功率,采用黄金分割算法解决功率控制问题,从而降低传输能耗。实验结果表明,AGASA算法在信道状态变化时可保证通信质量和计算效率,相比混合遗传粒子群算法,能够在满足截止期约束的同时使卸载能耗降低15.56%。

The edge server deployed in a conventional network architecture exhibits difficulty meeting the requirements of large-scale user equipment access and communication quality.To increase network capacity and improve spectrum utilization,dense base station deployment is combined with Ultra-Dense Network(UDN) to develop a task offloading optimization model for an ultra-dense edge computing network.The reasons for changes in channel status,the dynamic requirements of mobile devices,and the limitations of servers and spectrum resources pose challenges for offloading. A genetic algorithm based on an Adaptive Genetic Algorithm with Simulated Annealing(AGASA)’s task offloading method optimizes the energy consumption of task offloading while meeting the task deadline by combining the task type and the computing power of the server and considering the influence of channel state changes,mobile device dynamic requirements,and interference constraints on the offloading strategy. Meanwhile,to improve upload power,this study solves the power control problem with the golden section algorithm,saving transmission energy consumption. The experimental results demonstrate that when the channel state changes,the proposed task offloading strategy ensures communication quality and computational efficiency.It can meet deadline constraints while reducing its offloading energy consumption by 15.56% when compared to the hybrid genetic particle swarm algorithm(GAPSO).