Task Offloading in Fog: A Matching-driven Multi-User Multi-Armed Bandit Approach

Abstract

Fog computing is a potential technology that can solve computationally intensive, latency-sensitive, and energy-intensive tasks. This paper studies the task offloading problem for decentralized multi-users with unknown system-side information. On the basis, we design the distributed task offloading bandit (DTOB) algorithm to balance task offloading exploration and exploitation in fog computing (MFC) systems. Finally, the effectiveness of the offloading scheme based on bidirectional matching is verified by experiments.