Optimizing Resource Allocation in Mobile Edge Computing for IIoT through Reinforcement Learning and Multiobjective Particle Swarm Optimization

Abstract

Mobile networks are flexible enough to support a range of resource allocation models and service-based choices in computing domains, which affects both virtual reality and the Industrial Internet of Things (IIOT). Virtual resource management is made easier by the Mobile Edge Computing (MEC) paradigm, which also makes edge connectivity between data terminals and execution in the core network under heavy load possible. Meeting customer requirements efficiently is achieved through thoughtful planning, aided by cognitive agents. User data, incorporating behavioral patterns, is amalgamated to cater to IIOT service types. Neural caching for memory during task execution is made easier by the use of swarm intelligence and reinforcement learning techniques. Prediction strategies optimize business operations and caching to reduce execution delays. Predictive models evaluate performance by taking into account variables like nearby user equipment and mobile edge computing resources. This method's efficacy is demonstrated by a cognitive agent model that manages resource distribution and creates networks of communication to improve service quality. For accurate resource distribution among end users, reinforcement learning techniques—more especially, Multi-Objective Particle Swarm Optimization (MOPSO) algorithms—are used. This includes building cost mapping tables and optimizing allocation in MEC. The proposed method outperforms existing algorithms like Task-Offloading and Resource Allocation Strategy and achieves a better throughput value of 785.32.