Computation Bits Maximization for IRS-Aided Mobile-Edge Computing Networks With Phase Errors and Transceiver Hardware Impairments

Abstract

Intelligent reflecting surface (IRS) is a hopeful technique to improve the computation offloading efficiency for mobile-edge computing (MEC) networks. However, the phase errors (PEs) of IRS and transceiver hardware impairments (THIs) will greatly degrade the performance of IRS-assisted MEC networks. To overcome this bottleneck, this paper first investigates the computation bits maximization problem for IRSassisted MEC networks with PEs, where multiple Internet of Things (IoT) devices can offload their computation tasks to access points with the aid of IRS. By exploiting the block coordinate descent method, we design a multi-block optimization algorithm to tackle the non-convex problem. In particular, the optimal IRS phase shift, time allocation, transmit power and local computing frequencies of IoT devices are derived in closed-form expressions. Moreover, we further study the joint impact of PEs and THIs on the total computation bits of considered systems, where same methods in the scenario with PEs are used to obtain the optimal IRS phase shift and local computing frequencies of IoT devices, while an approximation algorithm and the variable substitution method are used to acquire the optimal transmit power and time allocation strategy. Finally, numerical results validate that our proposed methods can significantly outperform benchmark methods in terms of total computation bits.