Utility Maximization Based Trade-Off Policy for MEC-Enabled NB-IoT Networks in Smart Cities

To meet intensive computing needs of various IoT devices in smart cities, such as smart sensors and smart cameras, narrowband Internet of Things (NB-IoT) and mobile edge computing (MEC) are used for task processing of IoT devices. Although this approach can offload tasks to edge servers timely, it increases the computational expenses and wireless bandwidth expenses at edge networks. Therefore, we need to study computation offloading strategies more carefully to reduce the system overhead and ease the burden on NB-IoT devices. This paper investigates task offloading and resource allocation policy for IoT devices in smart cities. The aim is maximizing utility of the green MEC system while keeping stability of the edge network. This is a stochastic optimization problem. We transform this problem into a deterministic optimization problem by Lyapunov optimization technique and propose an utility maximization online algorithm (LUMOA). Theoretical analysis and experimental results demonstrate that LUMOA algorithm can approximately satisfy the system utility maximization while ensuring stability of the MEC system. Compared with equal allocation method and queue-weighted algorithm, the LUMOA algorithm increases the system revenue by 20.5% and 18.7%, respectively.