Game-Theoretic Cloud-Edge Resource Allocation for Video Analytics in the Factory of the Future
Yiyun Li, Ta-Sheng Lin, Hung‐Yu Wei · 2021
Factory of the Future (FoF) is a vision for how manufacturers should enhance management and production. In such a factory, real-time video analytic tasks are critical and can be provided by modern cameras and the server system behind them. With the 5G wireless technology and the developing deep learning models, the cloud-edge computing architecture can be applied to meet the application requirements of low latency and high accuracy. In this paper, we propose an efficient, near-optimal, and truthful mechanism to deal with the incentive-compatible resource allocation problem of video analytic service in FoF. To provide latency- and accuracy- aware service instantly, we relax the optimality and propose an efficient allocation algorithm that also helps truthful pricing in the mechanism design. With the theoretical analysis and the numerical simulations, we show the mechanism guarantees desired properties- computational efficiency, individual rationality, truthfulness, and weakly budget-balance.