Research On Optimization Strategy Of 5G-based Industrial Internet Of Things Networking

Industrial Internet of things(IIoT)connects and integrates all elements of industrial production process to achieve the goal of further improving the level of industrial automation,industrial upgrading and efficient utilization of various production factors.However,the IIoT scene has the characteristics of massive device connection,computing task intensive and delay sensitive,which poses a huge challenge to the existing communication network.The fifth generation mobile networks(5G)and mobile edge computing(MEC)can provide massive and reliable communication connection and computing offloading services for the IIoT,while ensuring the delay requirements of services.They are regarded as the key technologies to support the IIoT.Due to the increasing shortage of radio resources and the limited computing capabilities of MEC server,how to design the optimal radio and computing resource allocation strategy to meet the needs of IIoT scenarios is an urgent problem.First,this thesis systematically analyzes the research status of radio and computing resource allocation in IIoT scenarios,and then proposes a machine-to-machine(M2M)communication assisted cooperative offloading MEC network structure,and on this basis,studies the joint resource allocation of radio and computing resources.Secondly,considering that there may be many idle devices in the network,this thesis introduces the sleep-scheduling mechanism.That is,on the premise of ensuring sufficient computing resources of network,the system dynamically schedule some devices to switch to the sleep mode of shutting down the wireless communication module,to save system energy consumption.Based on this,takes the wireless resources,computing resources and maximum delay as constraints,this thesis models a problem of optimizing sleep scheduling strategy and joint allocation of radio and computing resources with the objective of minimizing the total energy consumption of the system.To solve the optimization problem,this thesis decomposes the original optimization problem into two subproblems:(1)sleep scheduling strategy and(2)joint allocation of radio and computing resources.Then,a DRL(deep reinforcement learning)-based sleep-scheduling strategy is proposed,and an iterative-optimization-based joint resource allocation algorithm is designed.Finally,the simulation results show that compared with other schemes,the proposed scheme can effectively reduce the system total energy consumption;and compared with the randomly sleep-scheduling strategy,the proposed sleep-scheduling strategy can better ensure the execution success rate of computing tasks,while saving more energy consumption of the system.Finally,considering the problem that multiple devices multiplexing the same subchannel to transmit data at the same time will cause mutual interference and reduce the transmission rate,this thesis considers that the computing task of the device can be divided into multiple subtasks,and two devices occupying the same subchannel transmit subtasks to the offloading target in turn by time division multiplexing.Moreover,this thesis defines the weighted sum of energy consumption and delay of computing task execution as system cost.On the basis,takes the wireless resources,computing resources and maximum delay as constraints,this thesis models a problem of optimizing offloading strategy and joint allocation of radio and computing resources with the objective of minimizing the total system cost.To solve the optimization problem,by combining the deep Q network(DQN)algorithm with deep deterministic policy gradient(DDPG)algorithm,this thesis proposes a DRL-framework-based offloading strategy and joint resource allocation algorithm.Finally,the simulation results show that compared with other benchmark schemes,the scheme and algorithm proposed in this thesis can effectively reduce the total system cost.