A study on global deployment and multi-agent balanced scheduling in internet of vehicles powered by edge computing

The booming Internet of Vehicles (IoV) demands well-planned task offloading to edge nodes, with efficiency and balance paramount. This paper presents a novel framework for the strategic global deployment of edge servers and a multi-agent balanced scheduling mechanism. It enhances the offloading process within IoV ecosystems. First, we propose an edge-server deployment algorithm that exploits minimum-distance density. It guarantees both full coverage and the shortest possible access distance for every vehicle. The scheme significantly reduces the infrastructure cost and improved the timeliness of task offloading. Second, a Q-learning-based multi-agent balanced scheduling method is proposed to achieve the balance between offloading distance and offloading equilibrium. Simulations show that the proposed algorithm was well-performing. It is verified to be better than other benchmark algorithms, especially when the system scales up. Our proposed algorithm can achieve the minimum target value. It proved the effectiveness and stability.