Delay-Optimal Temporal-Spatial Computation Offloading Schemes for Vehicular Edge Computing Systems
Dian Tang, Xuefei Zhang, Xiaofeng Tao · 2019
Vehicular edge computing (VEC) is a potential solution to meet delay-sensitive and computation-intensive vehicular applications demands. However, the nature of mobility brings a significant challenge, i.e., when and where to offload task can capture the optimal delay performance for VEC. To address this challenge, firstly, we characterizes the temporal-spatial correlation for VEC system. With the aid of this, we formulate the energy-constrainted delay minimization problem as a mixed integer nonlinear programming (MINLP). Due to the fact that it is hard to tackle the features about non-convex and coupling of designed MINLP problem, this paper transforms the original problem into task placement sub-problem and delayed offloading sub-problem. Specifically, the former is solved by two-stage decision tree algorithm and the latter is obtained by dynamic programming technique. On this basis, a temporal-spatial computation offloading scheme is proposed. Simulations demonstrate that our proposed scheme achieves superior delay performance and provides useful guides for choosing suitable schemes in different conditions.