Intelligent Resource Allocation Framework for 5G-Enabled Industrial IoT Using Deep Reinforcement Learning
Abstract
The quickly deployed 5G-enabled systems of IIoT have created systems with unparalleled connectivity, ultra-low Latency, and massively connected devices. However, effective and secure resource allocation continues to be a problem in the face of ever-changing traffic patterns. This paper presents an Intelligent Resource Allocation Framework (IRAF) to secure and optimally allocate traffic through spectrum, power, and computational resource distribution. This framework utilizes Deep Reinforcement Learning (DRL) to allocate traffic in settings with IIoT and 5G systems. The framework will use resource allocation as a Markov Decision Process (MDP) where the status of the network centers on its channels, the length of the queues, the priority of the devices, as well as the security risks. This policy is determined by a Deep Q-Network (DQN) based agent, which persistently interacts with the network environment and learns optimal allocation policies. Reward functions that are security-conscious are used so that anomalous traffic is avoided, and the potential for a traffic intrusion is not dangerous to the system. The results of the simulations indicated a 27.8% increase in the optimal spectrum usage, a 19.6% increase in the utilization of energy, and a 23.4% decrease in Latency as opposed to the heuristic-based traffic allocation systems of the past. Additionally, the system has a 21.3% increase in accuracy, and with the decreased vulnerability of the distributed denial of service (DDoS) Traffic, it correctly allocates secure traffic. 850 training episodes are the maximum for a system to be developed and deployed. The proposed framework is consistent with next-generation secure and intelligent computing models by merging adaptive learning, real-time optimization, and security awareness with 5G industrial networks. It provides a scalable and intelligent answer for the development of dependable, secure, and effective smart manufacturing and Industry 4.0 ecosystems.