Joint Energy-Computation Management for Electric Vehicles Under Coordination of Power Distribution Networks and Computing Power Networks

Weifeng Zhong, Wei Su, Xumin Huang, Jiawen Kang, Chau Yuen, Ruilong Deng, Yan Zhang, Shengli Xie

Abstract: This paper explores the integration of electric vehicles (EVs) into the power distribution network (PDN) and computing power network (CPN), leveraging EVs’ inherent energy storage and computing resources. A conceptual hub called a charging and computing station (CCS) is introduced, enabling parked EVs to interact with the PDN and CPN simultaneously. The CPN is composed of the EVs and edge servers, whose computing resources are collectively utilized for processing computation tasks from various applications. The EVs and edge servers in CCSs consume energy at different nodes of the PDN. A two-stage framework is proposed for joint energy and computation management in the EV-PDN-CPN coordination. In Stage 1, a day-ahead system cost minimization problem is formulated with decisions on EV charging/discharging energy scheduling and computation task reallocation among edge servers. A fast algorithm based on the convex-concave procedure is developed to solve the Stage-1 problem whose nonconvexity stems from network constraints of both the PDN and CPN. In Stage 2, EV computing resources are utilized to achieve real-time task offloading, coping with the prediction errors of computation tasks in Stage 1 and minimizing the use of extra energy and computing resources. A linear search algorithm is proposed to solve the nonconvex Stage-2 problem. Results show that the proposed algorithms are more computationally efficient than off-the-shelf solvers, and the proposed EV-PDN-CPN coordination model can save 4.7% and 91.9% of costs in the two stages, respectively, compared to uncoordinated models.