Worst Case Latency Optimization-based Joint Computation Offloading and Scheduling for Interdependent Subtasks

Abstract

In this paper, we consider a mobile edge computing (MEC) system consisting of a number of MEC servers and mobile devices (MDs) where each MD generates a series of subtasks characterized by their dependency relationships. Stressing the worst case performance optimization, we examine the maximal completion time of all the subtasks, and formulate the joint computation offloading and scheduling problem as a maximal completion time minimization problem. As the original optimization problem cannot be solved conveniently, we propose a heuristic algorithm which successively designs the computation offloading and scheduling strategy for various sets of subtasks based on the dependency relationships of the subtasks. In particular, for the subtasks of which the preceding subtasks have been completed, a priority- based subtask computing algorithm is further proposed. Numerical results demonstrate the effectiveness of the proposed scheme.