Dynamic Slicing Strategy for Power Internet of Things Based on Deep Reinforcement Learning
Software-defined power internet of things supports the construction of Network Slice(NS)that can carry different business requirements.By deploying NS,end-to-end services can be provided to internet of things devices with specific business demands.The deployment of business NS involves two interrelated issues,namely the deployment of Virtual Network Function(VNF)and the determination of business transmission routings.In dynamic network scenario with massive business requirements,NS deployment solutions need to achieve intelligent and dynamically flexible deployment based on the network status.To address the aforementioned problems,the slicing strategy in the dynamic network scenario is explored and the complex joint optimization problem of VNF deployment and business transmission routing determination is solved based on deep reinforcement learning algorithm.The experimental findings demonstrate that the proposed strategy effectively adjusts the deployment plan based on the current state,controls the average energy loss,average reliability,and average remaining bandwidth occupation of the service path,and improves the overall transmission performance of the network.
software-defined power internet of thingssliceVNFroutingdeep reinforcement learning
国家科技期刊平台
NSTL
万方数据
