STAR-RIS-aided NOMA communication for mobile edge computing using hybrid deep reinforcement learning

Reconfigurable intelligent surface (RIS) is expected to be able to significantly reduce task processing delay and energy consumptions of mobile users (MUs) in mobile edge computing (MEC) by intelligently adjusting its reflecting elements’ phase-shifts and amplitudes. Nevertheless, both the passive and active RISs have the disadvantage of only reflecting the received signals, which means that the transmitters and receivers must be located on the same side of the RIS. This may be unrealistic due to the movement of MUs. Simultaneously transmitting and reflecting (STAR) RIS, which can simultaneously transmit and reflect incident signals to achieve full-area coverage, has been recognized as a revolutionary technique to solve the above-mentioned problem. For the STAR-RIS-aided non-orthogonal multiple access (NOMA) communication MEC, we first formulate an optimization problem to minimize the sum of weighted delay and energy consumptions of all MUs which can move randomly at low speeds. Then, under the practical coupled phase-shift model of STAR-RIS, we propose a hybrid deep reinforcement learning (DRL) scheme, in which we determine the amplitudes and phase-shifts of STAR-RIS, task offloading decisions of MUs, and computation resource allocations of MEC servers by using the deep deterministic policy gradient (DDPG) and Dueling deep Q learning (DQN). Finally, we validate and evaluate the performances of our proposed scheme through extensive simulations, which show that our proposed scheme outperforms the existing baseline schemes and its performance can indeed be improved due to the use of STAR-RIS.