Hierarchical Energy-Efficient Mobile-Edge Computing in IoT Networks

Abstract

The ever-growing demand of the Internet of Things (IoT) imposes great challenges in the existing cellular systems and calls for novel approaches for the wireless network design. In this article, we develop a joint energy and computation optimization paradigm in an IoT network. The tasks collected at local IoT devices can be computed at hierarchical mobile-edge computing facilities. Both nonorthogonal multiple access (NOMA) and frequency-division multiple access (FDMA) are used for computation offloading. The system model considers both long-term and short-term system behaviors and makes the best decision for energy consumption and computation efficiency. The long short-term memory (LSTM) network is applied to predict the long-term workload, based on which the number of active process units in the edge layer is optimized. In the short-term model, a resource optimization problem is formulated. Due to the dynamic arrival workload and nonconvex features of the problem, the Lyapunov optimization approach and successive convex approximation for the low-complexity method are applied to solve this problem. The simulation results show that the proposed scheme can significantly improve the delay and energy consumption performance.