Managing mobility remains a challenge in the Internet of Things (IoT), especially for autonomous vehicles. This is addressed by vehicular fog computing (VFC), which executes tasks at surrounding fog nodes to speed up response time. However, mobility issues could worsen with rising numbers of mobile vehicles and fog nodes, increasing network overhead, and reducing response times due to frequent task migrations or large distances. This paper presents MARCH (Mobility-Aware Reinforcement Learning-based Computing Hierarchies), a novel reinforcement learning-based approach for improving edge and fog node management in mobile environments. MARCH focuses on avoiding task migrations and minimizing deadline misses for real-time tasks in an environment with stationary fog nodes, vehicular fog nodes, and cloud servers. Compared to state-of-the-art methods, MARCH reduces the deadline-miss ratio by up to 21% and lowers task migrations by up to 12%.

This project has contributed to the following academic papers:

1. MARCH: Mobility-Aware Reinforcement-Based Computing Hierarchies for Optimized Task Offloading in Vehicular Fog Networks
   • Authors: Elyas Oustad, Iman Mohammadi, Seyed Hossein Hosseini, Negar Babashah, Mohammad Mahdi Mirzaei, Sepideh Safari, Mohsen Ansari, Muhammad Shafique, Alireza Ejlali
   • Date: February 2022