Recommenadation aided Caching using Combinatorial Multi-armed Bandits
Pavamana K J, Chandramani Kishore Singh
TL;DR
This paper addresses cache optimization in wireless networks with recommendations by modeling joint caching and recommendation as a CMAB problem, aiming to maximize cache hits under partial observation. It develops a UCB-based algorithm that estimates per-item request probabilities from cached-content observations and selects the top $C$ contents to cache while recommending $R$ items among them; a confidence interval scales with the average recommendation acceptance $\bar{w}^{\text{rec}}$ and context parameter $\eta$. A second contribution tackles unknown user acceptance, introducing an estimator for $\bar{w}^{\text{rec}}(t)$ and adapting the UCB indices accordingly, with a dedicated algorithm and discussion of future regret analysis. Numerical results on synthetic settings (e.g., $N=50$, $C=20$, $U=20$) show improved cache-hit performance over baselines, robustness to Zipf-distributed recommendations, and effective learning of acceptance rates, highlighting practical gains for edge caching integration with recommendations.
Abstract
We study content caching with recommendations in a wireless network where the users are connected through a base station equipped with a finite-capacity cache. We assume a fixed set of contents with unknown user preferences and content popularities. The base station can cache a subset of the contents and can also recommend subsets of the contents to different users in order to encourage them to request the recommended contents. Recommendations, depending on their acceptability, can thus be used to increase cache hits. We first assume that the users' recommendation acceptabilities are known and formulate the cache hit optimization problem as a combinatorial multi-armed bandit (CMAB). We propose a UCB-based algorithm to decide which contents to cache and recommend and provide an upper bound on the regret of this algorithm. Subsequently, we consider a more general scenario where the users' recommendation acceptabilities are also unknown and propose another UCB-based algorithm that learns these as well. We numerically demonstrate the performance of our algorithms and compare these to state-of-the-art algorithms.