Joint Task Offloading and Power Allocation for Delay-Sensitive Tasks in IIoT

Fueled by technological advancements and the rising demand for intelligent manufacturing, the Industrial Internet of Things (IIoT) has rapidly evolved into a key driver of modern industrial transformation. In smart factories, heterogeneous devices generate a wide range of tasks, each with differing sensitivity to delay. This variability presents substantial challenges in task offloading and resource allocation. Many existing studies fail to consider these variations, resulting in suboptimal resource allocation. To tackle this issue, this paper introduces satisfaction of task (SOT) functions, which are designed to quantify the specific offloading requirements based on delay sensitivity of each task. Furthermore, a joint optimization framework is proposed, integrating task offloading and power allocation to maximize the SOT. A soft actor-critic (SAC) based algorithm is developed to solve the optimization problem and improve offloading efficiency. Simulation results show that the proposed approach better handles delay-sensitive tasks and demonstrates enhanced adaptability to diverse task types with varying delay sensitivities, outperforming current methods.