Intelligent Reflecting Surface Enhanced Wireless Powered Mobile Edge Computing

Abstract

Wireless powered mobile edge computing (MEC) has been a promising solution to improve the computation performance of the wireless networks. However, wireless devices (WDs) can not harvest sufficient energy and the link used for offloading tasks is hostile due to the doubly attenuation. Fortunately, the efficiency of wireless power transfer and spectrum can be improved significantly by intelligent reflecting surface (IRS), which can steer the incident signal collaboratively. This paper proposes a wireless powered MEC network assisted by the IRS, where the WDs follow a binary offloading rule. Our objective is to maximize the system computation rate by jointly optimizing the downlink and uplink passive beamforming of all IRSs, computing modes of the WDs and time allocation for wireless power transfer (WPT) and task offloading. The block coordinate descent (BCD) method is introduced to decompose the original problem into three sub-problems. The major difficulty is caused by the combinatorial nature of the WDs' computing mode selection. To solve this problem, we propose a duplex coordinate descent with dictionary (DCDD) method to obtain a sub-optimal solution with high efficiency. Numerical results show that the proposed scheme can achieve significant performance gains over the benchmark schemes without IRS.