Digital-twin-enabled task offloading for industrial internet of things based on prospect theory framework

Abstract Digital twin (DT) bridge the gap between the real and virtual worlds, enhancing decision-making efficiency by facilitating task offloading in the industrial Internet of Things through comprehensive real-world status information. However, the often-overlooked discrepancies between DT and their real-world counterparts introduce high uncertainty into offloading decisions, potentially leading to unexpected outcomes. To address this issue, we adopt prospect theory to formulate a task offloading decision problem that integrates the behavioral tendencies of system participants, thereby maximizing participants’ utility and providing a realistic approach to managing offloading decisions. We reformulate this NP-hard problem as a potential game and demonstrate the existence of a Nash Equilibrium (NE). The finite improvement property is used to implement a decentralized algorithm that identifies the NE in the potential game as a solution to the offloading problem. Furthermore, we theoretically derive an upper bound on the algorithm’s convergence time. The superiority of our proposed scheme over existing schemes in terms of performance and scalability is evaluated and demonstrated through extensive simulations.