Joint Computation Offloading and Resource Allocation in Green MEC-Assisted Software-Defined Island Internet of Things

Mobile edge computing (MEC) powered by renewable energy, is promising to provide green computing for the Internet of Things (IoT). However, the unpredictable renewable energy and computing demands usually cause a mismatch between system requirements and energy supply, resulting in wasted surplus energy or energy supply shortage. Hence, it is crucial to improve energy efficiency and system performance, that is, “make the best use of generated energy” and “make the best use of system’s talents” simultaneously. In this article, we focus on some islands far from the mainland, with growing computation requirements for environmental monitoring and navigation safety, and propose a device-to-device (D2D) collaboration-based software-defined network-MEC framework in Island IoT employing tidal energy. Following that, we formulate a multiobjective energy scheduling system performance association (MESPA) problem to minimize the long-term average task execution loss (TEL), including energy consumption per bit executed, overall execution latency, and energy waste, caused by underutilization of tidal energy, with the constraints of energy queue stability, peak transmission power, and central process unit-cycle frequency. To address this challenging problem, we propose a Lyapunov-based multidimensional resource allocation and computation offloading (LMDRACO) algorithm and transform the original problem into several individual subproblems in each time slot. These subproblems are then solved using convex decomposition and submodular methods. Theoretical research shows that the LMDRACO algorithm can achieve a [ $\mathcal {O}$ (1/V), $\mathcal {O}$ (V)] tradeoff between TEL and energy queue length. Numerical results show that the proposed algorithm significantly improves both system performance and energy efficiency compared to baseline schemes.