Optimizing VNF Service Provisioning in Mobile Edge Computing Networks Through PSO-RL

This study deals with setting up Virtual Network Function (VNF) services in Mobile Edge Computing (MEC) networks. It aims to balance different user needs in the everchanging MEC environment. Leveraging Particle Swarm Optimization-Reinforcement Learning algorithms (PSO-RL), our proposed system offers a dynamic provisioning approach adept at accommodating user mobility patterns and service delay prerequisites. The system's sequential flow comprises distinct stages designed for optimal functionality. The Firefly Algorithm efficiently maximizes utility by simplifying intricate issues and approximating optimal solutions. Additionally, our system implements Stochastic Gradient Descent and Simulated Annealing techniques for real-time throughput optimization, ensuring adaptability to ever-changing network conditions. Experimental simulations form a critical component of our study, providing a robust assessment of our proposed methodologies. Results obtained demonstrate substantial enhancements across critical metrics such as Throughput, Resource Utilization Efficiency, and End-to-End Delay. Varied scenarios encompassing different quantities of IoT users and mobile edge servers were simulated, reaffirming the efficacy of the PSO-RL method. These simulations underscore the potential of our proposed approach in significantly bolstering the performance and responsiveness of MEC networks. Ultimately, this study showcases promising prospects in addressing the intricate challenges associated with provisioning VNF services to mobile users in dynamic MEC environments.