Min-Max Latency Minimization for Energy-Constrained Multi-UAV Mobile Edge Computing

This paper considers a multi-unmanned aerial vehicle (UAV)-mounted mobile edge computing (MEC) framework, in which a set of devices offload all or a fraction of their computational tasks to the UAV-mounted MEC servers. To enable simultaneous offloading from the devices to MEC servers, a non-orthogonal multiple access transmission is considered such that each device transmits a superposed message, and a successive interference cancellation-enabled decoding is performed at each server. A min-max latency minimization problem is formulated to optimize the UAVs' placement, the edge computing resources, the power allocation of each device, and the offloading volume from each device to the servers. Pertinent practical constraints are imposed on the edge computing capabilities of the UAV-mounted servers and the energy consumption. An alternating optimization approach is developed using the block coordinate descent (BCD) technique. A more computationally efficient heuristic solution is also introduced, which involves a weighted K-nearest neighbor approach and ϵ-scaled power allocation. The case of sufficient energy availability is considered as well. Numerical results illustrate that the heuristic solution provides a good performance close to the BCD technique performance. Results also demonstrate the effectiveness of the considered framework compared to the device-UAV association scenario