A Probabilistic Computation Offloading Strategy for Vehicle-to-Everything Networks

Abstract

Vehicle-to-everything (V2X) is regarded as a fundamental service offered by the studied B5G/6G wireless technology. However, data traffic services provided in V2X networks requires computation capabilities, presenting a tremendous challenge in ensuring excellent user experience. To address this demanding task, computation offloading emerges as an effective solution. In this paper, we propose a probabilistic computation offloading strategy for V2X networks. Specifically, in order to minimize the total network delay, we first investigate the characteristics of task types and then categorize them into different groups. In particular, a priority-based M/M/1 queueing model is designed to simulate the transmission and computation process for offloading various tasks, where a probabilistic offloading problem is then established to minimize the average response delay. Finally, the task’s optimal offloading probability problem is solved based on the sub-gradient search (SGS) algorithm. Simulation results demonstrate that by taking task characteristics and priorities into account, the constructed model can obtain the optimal offloading probability and also satisfy the requirement of low latency service.