基于能量收集技术的协作卸载计算方案

Collaborative offloading computing scheme based on energy harvesting technology

近年来,物联网(IoT)应用对设备可使用的能量要求不断提高,能量收集(EH)技术成为缓解边缘计算中设备能量短缺问题并延长电池寿命的重要途径。然而,当环境中可再生能源不充足时,设备电量耗尽会导致任务中断,影响物联网性能。为了解决这一问题,提出了一种联合能量收集和设备间(D2D)通信技术的任务卸载框架,采用基于深度强化学习(DRL)的边缘协作卸载计算方案,自主进行决策并使用模拟退火算法解决资源分配问题,以最小化系统运行总成本。对稳定和极端两种能量环境进行仿真,结果表明,该方案在单用户多设备场景下可稳定运行且具有成本效益。

In recent years,the energy requirements for devices in internet of things(IoT)applications have increased,making energy harvesting(EH)technology an important way to alleviate the energy shortage problem in edge computing and extend the battery life of devices.However,when there was insufficient renewable energy in the environment,the depletion of device power can cause task interruption and affect the performance of IoT.To solve this problem,a task offloading framework that combined energy harvesting and device-to-device(D2D)communication technology was pro‐posed,using a deep reinforcement learning(DRL)-based edge collaborative offloading computing scheme to make au‐tonomous decisions and solve resource allocation problems using simulated annealing algorithms to minimize the total cost of system operation.Simulation results on stable and extreme energy environments show that the proposed scheme can run stably and cost-effectively in single-user multiple-device scenarios.