Latency Minimization for IRS-Aided NOMA MEC Systems With WPT-Enabled IoT Devices

Abstract

Mobile edge computing (MEC) and intelligent reflecting surface (IRS) are envisioned as two promising technologies that enable massive connectivity in the future Internet of Things (IoT) networks. MEC allows IoT devices (IDs) to offload their computation intensive tasks, and thus, can prolong their lifespan. In contrast, the IRS can enhance the channel condition between IDs and the access points (APs), which are colocated with the MEC server. Wireless power transfer technique enabling energy harvesting for IDs helps realizing sustainable IoT network. This paper applies IRS in a multi-ID MEC system for better latency performance. We first propose a multiple access scheme with hybrid frequency division and non-orthogonal access technologies and then design a timing protocol for the IDs. Based on the above design, we study the latency optimization problem with joint optimization of power allocation, the IRS phase shift matrix, and uplink and downlink beamformer under maximum power constraint for the IDs and AP. To tackle the formulated multi-variable non-convex problem, we split the target problem into several sub-problems and provide a nearoptimal low-complexity ID clustering scheme. Afterwards, we derive optimal solutions to these sub-problems, and a lowcomplexity fast-convergence alternating algorithm is proposed to minimize the overall latency. Presented simulation results verify the convergence of the alternating algorithm, and its superiority over the benchmarks.