Stability-Awareness Edge Task Offloading in Industrial Internet of Things

Edge computing technology has long been a popular paradigm for solving task offloading problems in the Industrial Internet of Things (IIoT), and favored by researchers. The most commonly used task offloading metrics are latency and energy consumption. However, in IIoT scenarios with potential critical risks in the process flow (e.g. in chemical production), the stability of the system is prioritized over energy consumption and other metrics. Therefore, for such IIoT scenarios with potential critical risks, this paper conducts scenario modeling and defines the task offloading optimization objective function as the weighted sum of latency and offloading system stability, to achieve a trade-off between task processing latency and offloading system stability. Therefore, a task offloading method based on deep reinforcement learning (DRL) is proposed. The experimental results show that, whether in terms of latency performance, system stability or workload balance, the proposed task offloading method can achieve better results compared with the baseline optimization methods.