基于最优运输理论的物联网边缘计算资源优化机制

Resource allocation based on optimal transport theory in IoT edge computing

随着物联网和边缘计算的发展,物联网设备可以将计算密集型任务卸载到边缘计算服务器上进行处理。由于物联网设备分布以及计算需求的变化,需要对边缘计算资源进行动态管理。利用最优运输理论对物联网中计算资源分配进行优化,提出一种基于物联网设备分布和边缘计算服务器位置的区域优化划分机制,在边缘计算服务器计算能力的约束下对物联网设备的能耗以及时延性能进行优化。仿真结果表明,与传统泰森多边形划分机制相比,该优化机制有更好的均衡性,并且物联网设备的平均能耗最多降低21%,平均时延最多降低45%。

With the development of the Internet of things(IoT)and edge computing,the computation-intensive tasks of IoT devices can be offloaded to edge devices and processed at the edge of networks.Due to the variation of the distribution and computation requirements of IoT devices,the computation resources of edge networks need to be managed dynamically.The optimal transport theory was adopted to optimize the computation resources allocation in IoT networks.An optimized regional partition mechanism was proposed based on the distribution of IoT devices and locations of edge computing devices.Under constraints on the computing capabilities of edge computing devices,the energy consumption and delay of IoT devices were optimized.The simulation results show that,compared with the traditional Voronoi partition scheme,the proposed optimization mechanism shows better balance.The average transmitting power can be reduced by 21%and the average delay can be reduced by 45%.