MCVCO: Multi-MEC Cooperative Vehicular Computation Offloading
Jianhang Liu, Kunlei Xue, Qinghai Miao, Shibao Li, Xuerong Cui, Danxin Wang, Kewen Li
Abstract
Mobile edge computing (MEC) has been envisioned as a promising paradigm that provides processing resources for vehicular computation-intensive tasks to accommodate the strict latency requirement. However, there is still a need to further enhance system performance to overcome challenges such as poor efficiency of data transmission and limited system resources. To improve the quality of service, this article proposes a multi-MEC cooperative vehicular computation offloading (MCVCO) scheme. Firstly, we propose a heat-aware task offloading strategy to capture the time-varying multi-link relations between vehicle and MEC nodes. Secondly, we design a multi-MEC resource compensation method based on fountain code which cooperatively collects the task data and improves the efficiency of data reception in the edge layer. Finally, we develop a parallel transmission and execution based dynamic scheduling algorithm to make the most of available resources. Extensive simulation results and analyses demonstrate that MCVCO outperforms other baseline schemes in various experimental settings. MCVCO achieves a 32% increase in success rate, up to a 47% reduction in end-to-end latency, and a 24% improvement in uploading quality.