Learning-Aided Content Placement in Caching-Enabled fog Computing Systems Using Thompson Sampling

摘要

In this paper, we focus on the problem of online content placement with unknown content popularity in caching-enabled fog computing systems, i.e., how to decide and update cached content on resourcelimited edge fog nodes to maximize cache hit rate and minimize switching costs of content update. Faced with such uncertainties, the placement procedure must be well integrated with effective online learning while ensuring minimum performance loss (a.k.a. regret) due to improper content updates. To overcome such difficulties, we formulate the problem as a multi-play multi-armed bandit problem. By adopting Thompson sampling methods, we propose LACP, a learning-aided content placement scheme which continuously improves its online decision-making by proactively learning with hit-or-miss feedback information. Our theoretical and simulation results demonstrate the effectiveness of LACP against baseline schemes with an O(logT) regret over time horizon T.