Complexity Dependencies-Based Optimal Tasks Offloading and Computing Resource Allocation Scheme in Industrial Internet of Things

The evolution of the Industrial Internet of Things (IIoT) is essential for fostering the growth of smart manufacturing. In the realm of smart manufacturing, the interconnectivity of production devices (PDs) has emerged as a new production paradigm. In this paradigm, each PD is required to manage various production subtasks that possess interdependencies. Precisely representing the dependencies among these subtasks is critical for reflecting the production status in the smart factories. However, existing studies typically concentrate on the internal dependencies (IDs) among subtasks, overlooking the significant aspect of inter-connectivity. As a result, there is an absence of comprehensive descriptions regarding the dependencies among subtasks across different devices. This paper introduces a dependency model for complex subtasks sets, integrating both IDs and external dependencies (EDs) to provide a precise representation of the interdependencies among subtasks. Additionally, considering the constraints posed by limited computing resources on Industrial Edge Servers (IESs), we formulate an optimization problem aimed at the optimal allocation of resources to minimize the cost of system. To address this optimization problem, we propose the proximal policy optimization (PPO) based dependent complex tasks for offloading and computing resources allocation (DOCRA) algorithm, which significantly reduces the cost of system in comparison with alternative solutions.