Minimizing Energy Consumption for Mobile Edge Computing with Non-orthogonal Multiple Access

We consider resource allocation for offloading computational-intensive tasks in a mobile-edge computing (MEC) system, where each IoT device's task can be processed at the MEC server, access points (APs), or locally at IoT device itself. The envisioned system has both cooperative APs with a computing capability and multiple radio resource blocks (RRBs) and a MEC server. We aim to study the trade-off between minimizing the energy consumption and maximizing the effective system capacity, which is the number of IoT devices with a successful task processing. For this objective, we exploit non-orthogonal multiple access (NOMA) to schedule a set of IoT devices to the set of MEC's subcarriers and RRBs of the APs. Due to the intractability of the energy consumption minimization problem, we split it into two sub-problems. The first sub-problem considers the offloading decision and local computation allocation, while the second sub-problem considers the IoT device scheduling and power allocation. Leveraging graph-theory, we propose an approach for solving the two sub-problems. Numerical results are presented to depict the trade-off between minimizing the energy consumption and maximizing the effective system capacity of the proposed approach over benchmark schemes.