Energy-Efficient Resource Allocation for Mobile Edge Computing With Multiple Relays

Abstract

In recent years, mobile edge computing (MEC) has attracted tremendous research thanks to its advantage in handling computation intensive latency critical tasks. To overtake the bad channel condition in the process of task offloading, multiple-relay assisted MEC system is considered in this paper. In specific, three cases including TDMA scenario, FDMA scenario in decode-and-forward (DF) mode and amplify-and-forward (AF) mode are investigated. The target is to minimize the overall energy consumption of mobile user and relays by jointly optimizing offloading data amount, transmit power and slot duration (in TDMA, or bandwidth allocation in FDMA, or amplitude gain in AF). In the scenario of TDMA, we show the associated problem is convex and solve it in a easier way through the manner of bi-level optimization. In the upper level, the optimal data amount for offloading is acquired, which corresponds to a simpler convex optimization problem, while in the lower level, the optimal solution of the rest of variable are found via KKT conditions. In the scenario of FDMA, the associated optimization problem is non-convex. Global optimal solution is found with the help of bi-level optimization and monotonic programming. For AF mode, bi-level optimization is also utilized in which neither of the two levels is convex. To this end, geometric programming and successive convex approximation (SCA) is used to find the convergent solution of the lower level while monotonic programming is adopted in the upper level. Numerical results proves the effectiveness of the proposed strategies under various scenarios investigated in this paper.