工业物联网中基于Sarsa算法的节能计算卸载方案

Energy-saving computation offloading scheme based on Sarsa algorithm in industrial internet of things

针对任务有完成截止时间要求的工业物联网系统,为了降低该类系统的总能源损耗,提出一种基于Sarsa算法的节能计算卸载方案。基于Sarsa算法的特点对能耗优化问题进行了解耦,由Sarsa算法进行外部状态值函数的迭代,选择合适的边缘计算服务器进行计算任务卸载,再由粒子群优化算法解决其中的资源分配问题,最后更新环境信息进行下一轮算法迭代,直至算法满足结束条件。仿真结果表明,本方案具有更快的收敛速度并且有效降低了系统能耗。

In order to reduce the total energy consumption of industrial internet of things systems with deadline requirements,a computing offloading scheme based on Sarsa algorithm was proposed.Based on the characteristics of Sarsa algorithm,the energy consumption optimization problem was coupled.The Sarsa algorithm iterated the external state value function,selected the appropriate edge computing server to offload the computing task,and then the particle swarm optimization algorithm solved the resource allocation problem.Finally,the environmental information was updated for the next round of algorithm iteration until the algorithm met the end conditions.The simulation results show that this scheme has faster convergence speed and effectively reduces the system energy consumption.