Cost-Efficient Edge Cloud Deployment Method for Autonomous Driving

When edge clouds are deployed at all road side unit (RSU), autonomous vehicles (AVs) can offload the tasks and receive the results of the tasks with low latency. However, too excessive deployment of edge clouds can lead significant capital expenditure (CAPEX) of an offloading service provider. In this paper, we propose a cost-efficient edge cloud deployment method where the deployment locations of edge clouds are minimally decided by considering task offloading rates of road segments. To minimize the number of deployed edge clouds while supporting all generated traffic volumes, we formulate an integer non-linear programming (INLP) problem. For the practical deployment even at a huge target area, we propose a low-complexity heuristic algorithm called traffic-aware deployment algorithm (TADA). Evaluation results demonstrate that TADA can achieve a similar deployment cost with the optimal solution while handling all generated traffic.