Total Computational Bits Maximization for STAR-RIS Aided Wireless Power Transfer Mobile Edge Computing Networks: TDMA or NOMA?

Due to the limited battery capacity and computational power of Internet-of-Things (IoT) devices, emerging IoT applications depend on mobile edge computing (MEC) networks for computational offloading and wireless power transfer (WPT) to sustain energy levels, with transfer performance influenced by the propagation environment and transmission strategies. Compared to the traditional reflecting-only reconfigurable intelligent surface (RIS), simultaneously transmitting and reflecting (STAR)-RIS extends coverage from half-space to full-space, significantly enhancing transferring efficiency. This paper compares the performance of three different uplink transmission schemes on a STAR-RIS WPT MEC network. On this basis, for each scenario, a non-convex optimization problem is formulated to maximize the total computational bits for the system, where energy transfer, local processing, uplink transmission allocated resources, and phase shift vectors of STAR-RIS are jointly optimized. To address the non-convexity, each scenario is decoupled into two main subproblems: phase shift vectors of STAR-RIS optimization and joint power, time, and computing frequency optimization. While a closed-form expression is derived to optimize the phase shift vector of STAR-RIS in time division multiple access (TDMA), semi-definite relaxation is employed to optimize the phase shift vectors under Hybrid TDMA-non-orthogonal multiple access (NOMA) and NOMA. Furthermore, block coordinate descent is employed to address the remaining resources. Simulation results indicate the superiority of the proposed scheme for all scenarios in terms of total computational bits compared to benchmark schemes. Moreover, among the three uplink transmission schemes, TDMA performs the best in total computational bits and user fairness owing to its ability to flexibly adapt STAR-RIS phase shift vectors.