Predictive And Collaboration-based Edge Nodes Reconfiguration For The Industrial Internet Of Things

As the world is becoming increasing digital,more and more spaces are turned into smart environment,ranging from individual houses and offices to factories and hospital and even to whole cities.In order to construct the smart environment,an enormous number of sensors which are wirelessly connected with each other are densely deployed,especially in industry.Traditionally,industrial data generated by device sensors are collected and transported over the Internet to cloud,where large industrial data are centralized processed and analyzed.Nevertheless,many industrial applications are latency-sensitive,demands immediately response within an acceptable period of time.A small error or delay beyond the adequate limit might results in some catastrophe for various application.Obviously,in these cases,data processing in the cloud is impractical.Thanks to recent advancements on Edge computing,it is possible to overcome these tricky challenges accurately.Edge computing is a networking philosophy focused on bringing computing as close to the source of data as possible in order to reduce response latency and bandwidth usage.However,compared to cloud computing,the inherent resource limitations of distributed edge nodes such as energy,computing,storage and bandwidth,render it impracticable for all tasks to be processed on the edge simultaneously.To address these issues,this paper proposed a novel predictive and collaborative based edge nodes reconfiguration and selection(PCERS)strategy in IIoT.Our main contributions are as follows.Firstly,a prediction-based optimal service reconfiguration is proposed.This approach decomposes long-term prediction and configuration problem issues into several reconfiguration cycles.During the reconfiguration cycle,edge node predicts the possible arrival requests of the next cycle according to the devices' history information,and then configure the optimal services on edges.Thereafter,this paper proposes a task offloading policy based on edge node collaboration when the device selects the services of edge nodes.The strategy combined vertical offloading with horizontal offloading,and mainly focus on the latter.By collaborating and sharing computing resources with the edge nodes of one-hop neighbors,it can effectively reduce system network latency.Finally,given the long-term energy constraints of the entire network,the proposed approach was upgraded to maximize the network lifetime by limiting the use of energy during the period of each cycle.We divide the total energy of the edge nodes according to the expected time of use,and proposes an available energy to indicate priority.Simulation results show that the PCERS in this paper can significantly reduce network latency and improve the efficiency of edge computing while satisfy the long energy constraints.