Latency and Cost Minimization for Task Offloading with Energy Harvesting in UAV-MEC Network

Unmanned aerial vehicles (UAVs) are recently considered for various Internet of things (IoT) applications. Furthermore, mobile edge computing offers low-latency data computation and improves network performance by bringing computing resources closer to IoT devices. UAVs equipped with MEC capabilities are expected to play a critical role in providing reliable and efficient communication in future wireless networks. The low-power IoT devices can harvest energy for sustainable operations in different applications such as public safety. Resource management and efficient task offloading are the major challenges to take full advantage of UAV-MEC networks. In this paper, we formulate an optimization problem to minimize latency and cost as well as maximize the number of IoT devices connected with UAVs. We consider constraints on computation, caching, energy harvested, latency, and the maximum number of connections a given UAV allows. A heuristic with learning placement (HLP) algorithm is proposed based on an unsupervised learning algorithm and an iterative rounding algorithm to solve the problem with less complexity. Simulation results show the effectiveness of the proposed HLP algorithm in UAV-MEC networks compared with the branch and bound algorithm (optimal results).