Active Popularity Learning with Cache Hit Ratio Guarantees using a Matrix Completion Committee

Edge caching is a promising technology to face the stringent latency requirements and back-haul traffic overloading in 5G wireless networks. However, acquiring the contents and modeling the optimal cache strategy is a challenging task. In this work, we use an active learning approach to learn the content popularities since it allows the system to leverage the trade-off between exploration and exploitation. Exploration refers to caching new files whereas exploitation use known files to cache, to achieve a good cache hit ratio. In this paper, we mainly focus to learn popularities as fast as possible while guaranteeing an operational cache hit ratio constraint. The effectiveness of proposed learning and caching policies are demonstrated via simulation results as a function of variance, cache hit ratio and used storage.