Price and Risk Awareness for Data Offloading Decision-Making in Edge Computing Systems

Abstract

The proliferation of multiaccess edge computing (MEC) paradigm has created a challenging multiuser–multiserver–multiaccess edge computing competitive environment, which brings the problem of data offloading decision-making to the forefront of research. In this article, we address this issue while jointly studying the impact of the user behavioral characteristics and the MEC servers pricing policies on determining the optimal user data offloading strategies. Prospect theory is exploited to reflect the user satisfaction and subjectivity from the data offloading, while the MEC servers’ probability of failure owing to the potential overexploitation by the users, is modeled via the theory of tragedy of the commons. A multileader multifollower Stackelberg game is formulated among the MEC servers (leaders) and the users (followers), to determine the servers’ optimal pricing policies and the users’ optimal data offloading strategies. The users’ data offloading decision-making is formulated as a noncooperative game among them and a Nash equilibrium is determined, while the MEC servers’ optimal computing service prices are obtained either through a semiautonomous game-theoretic approach, or through a fully autonomous reinforcement learning-based approach. The performance evaluation and demonstration of the superiority of the proposed framework against other benchmarking alternatives is achieved via modeling and simulation.