A Potential Game-Based Tasks Offloading Approach for Edge Computing in Industrial Internet of Things

With the acceleration of industrial digitization, integrating multi-access edge computing (MEC) into the Industrial Internet of Things (IIoT) effectively addresses copious latency-sensitive and computationally intensive tasks. It has been explored to organize IIoT devices into multiple units, but many studies overlook inter-unit interference, leading to overestimate uplink transmission rates and get suboptimal offloading decisions. To address this problem, a central cloud-assisted edge computing architecture is proposed to minimize the execution cost of tasks by optimizing the offloading decisions while considering the inter-unit interference, thereby achieving the joint optimization of latency and energy consumption. Due to the coupling of task costs, the optimization problem is formulated as a noncooperative game among users. By proving that the game is a potential game and thus has a pure-strategy Nash equilibrium (NE), based on the finite improvement of potential game, a best response-based task offloading algorithm (BRTOA) is developed to implement NE, complemented by a differential evolutionbased improving task set offloading policy algorithm (DETPA) to enhance the convergence. Simulation results indicate that the proposed algorithm outperforms existing methods in terms of system cost.