Optimal Task Allocation in Vehicular Fog Networks Requiring URLLC: An Energy-Aware Perspective

Abstract

In order to make intelligent transportation systems (ITSs) come true, execution of a large amount of data needs to be migrated from the cloud centers to the edge nodes, especially in the scenarios requiring ultra reliable low latency communications (URLLC). In this article, we propose to study the energy-aware task allocation problem in the vehicular fog networks considering URLLC. Specifically, a requester who has some bursty computation tasks which cannot be finished within a required time by itself, needs to decide whether the nearby computation nodes can meet the latency and reliability requirements, and which nodes should be chosen. Given the required latency and reliability, the maximum computation capacity of each fog node is first calculated based on the martingale-theory-derived delay bound. Then, if the available fog nodes can accommodate the computation tasks, two different optimization problems concerning the energy efficiency maximization and the energy consumption minimization are constructed further. The corresponding solutions are also provided. Specifically, the optimal solution in maximizing the energy efficiency is not unique, while the optimal solution in minimizing the energy consumption is unique. Moreover, the latter solution is provided as a truncated-channel-inversion like policy. At last, numerical results are illustrated to demonstrate effectiveness of the proposed optimal task allocation schemes from the perspectives of the energy efficiency and the energy consumption.