Intelligent and connected vehicles rely on edge computing to offload their perception and planning tasks, so the scheduling of communication and computing resources is critical to the driving safety and efficiency. However, the imbalanced distribution of road traffic and offloading demands impedes the quality of vehicular edge computing. In this paper, we propose a multi-timescale load balancing approach to improve the service quality and resource utility of vehicular edge computing. Specifically, vehicle mobility optimization is leveraged to perform long-term load balancing, and resource allocation is used to achieve real-time load balancing. As the multi-timescale optimization is confronted with the curse of dimensionality, multi-agent deep reinforcement learning is utilized to optimized vehicle mobility and resource allocation in parallel. Experimental results show that the proposed method can significantly reduce the service delay of vehicular edge computing.
A Multi-Timescale Load Balancing Approach in Vehicular Edge Computing
2020-11-01
1638313 byte
Conference paper
Electronic Resource
English
Vehicular Edge Computing for Multi-Vehicle Perception
IEEE | 2021
|Efficient Cellular Load Balancing Through Mobility-Enriched Vehicular Communications
Online Contents | 2016
|User Association for Load Balancing in Vehicular Networks: An Online Reinforcement Learning Approach
Online Contents | 2017
|Efficient Cellular Load Balancing Through Mobility-Enriched Vehicular Communications
Online Contents | 2016
|