Dynamic Offloading Based on Lyapunov Optimization for UAV-Assisted Maritime IoT-MEC Networks

With the rapid expansion of Maritime Internet of Things(M-IOT), a number of computation-intensive and latency-sensitive tasks are increasing. To cope with this challenge, the convergence of maritime Unmanned Aerial Vehicle (UAV) and Mobile Edge computing (MEC) has become a much-anticipated solution for intelligent maritime systems, which can effectively improve the service by delivering computationally intensive tasks to UAVs for processing. In this paper, we consider a UAV-assisted M-IoT network in which task data arrive randomly. Our goal is to design an online algorithm that can minimize devices' transmission energy and UAV propulsion energy while satisfying queue stability and making decisions for each time slot without knowing the future data arrivals. We formulate the problem as a multi-phase linear programming, and proposed a low-complexity delay-aware dynamic offloading (DADO) based on Lyapunov optimization to solve the coupling effect of the control actions in consecutive time. This algorithm transforms the long-term decisions into deterministic optimization within each time slot. The algorithm can realize the trade-off between energy and delay and make adaptive decisions based on the state of the network. Finally, the experimental results show that, comparing with other algorithms, our proposed algorithm effectively reduces the energy consumption and delay while also ensuring queue's long-term stability.