Latency Minimization in Uplink Non-Orthogonal Multiple Access-based Mobile Edge Computing

We consider the problem of latency minimization in non-orthogonal multiple access (NOMA)-based mobile edge computing (MEC) system. All the users are equipped with local computing power and offload their recorded data to the central MEC server following the power-domain NOMA technique. We find the optimal power allocation and the part of data offloaded by each user such that the overall latency is reduced. Towards this end, we formulate an optimization problem that minimizes the overall worst-case latency, subject to constraints on maximum transmission power and maximum energy available for local computation and offloading. We show that this optimization problem is non-convex, but can be reduced to a series of feasibility problems that are solved using the bisection search method. Through Monte Carlo simulations, we show that the partial offloading-enabled NOMA-based MEC system significantly outperforms the fully offloading-enabled NOMA-based MEC system and the fully offloading-enabled OFDM-based MEC system.