Resource Allocation Algorithm For Industrial Internet Of Things Based On Deep Reinforcement Learning

Industrial Internet of Things(IIoT)is one of the typical application scenarios of5 G mobile communications.It plays a very important role in improving industrial automation production efficiency,reducing product costs,and increasing production safety.However,the dynamic changes of the IIoT environment,limited spectrum resources,and the low latency,low energy consumption,and high security requirements of IIoT devices have posed tremendous challenges to the practical application of IIoT.This thesis focuses on the low latency,low energy consumption,high throughput,and high security requirements of IIoT communication.It studies the resource allocation issue in the IIoT environment and proposes short-packet communication resource allocation algorithms subject to security capacity constraints,joint allocation algorithms for uplink power and channel based on full-duplex networks,and edge computing task offloading and device allocation algorithms based on deep reinforcement learning methods.The main research contents of this thesis are as follows:To address the issues of energy consumption constraints and high security requirements in the industrial Internet of Things environment,this thesis proposes a power allocation and bandwidth allocation optimization problem for short packet communication,considering the minimum secure transmission capacity of communication information and the limited total available bandwidth.The goal is to minimize the total transmission power of IIoT devices.Based on the constraints and optimization problem,a bandwidth and power allocation algorithm using the Double Deep Q Network(DDQN)and the Deep Deterministic Policy Gradient(DDPG)algorithm is designed.Simulation results show that the proposed intelligent resource allocation algorithm under the security capacity constraint effectively reduces the total transmission power of short packet communication in the industrial IoT environment.For the high throughput and high security requirements of the industrial Internet of Things uplink,considering the security transmission rate and delay constraints of IIoT devices,this thesis constructed a maximum channel allocation and power optimization problem to maximize the total throughput of IIoT devices.For the problem solving,this thesis designed a sub-channel allocation network based on DDQN and a power distribution network based on near-end policy optimization.Simulation results show that the proposed channel allocation and power optimization algorithm based on DDQN and near-end policy optimization can effectively improve the total throughput of industrial IoT systems compared to other algorithms.To meet the low latency and high security requirements of industrial IoT tasks,a security rate guarantee calculation offloading model assisted by digital twins under the environment of IIoT is established,and the problem of minimizing the total task delay under the constraints of IIoT device transmission power,task delay,and secure transmission rate is constructed.This thesis proposes task offloading and device association algorithms based on single intelligent agent deep reinforcement learning and multi-agent reinforcement learning,designs a power allocation network based on DDPG and a device association factor allocation network based on DDQN,and solves the offloading ratio in combination with linear programming.Simulation results show that the task offloading and device association algorithm based on multi-agent deep reinforcement learning proposed in this thesis effectively reduces the maximum total task delay of the IIoT system.The thesis has 28 figures,15 tables,and 86 references.