Content caching in small base stations or wireless infostations is consideredto be a suitable approach to improve the efficiency in wireless contentdelivery. Placing the optimal content into local caches is crucial due tostorage limitations, but it requires knowledge about the content popularitydistribution, which is often not available in advance. Moreover, local contentpopularity is subject to fluctuations since mobile users with differentinterests connect to the caching entity over time. Which content a user prefersmay depend on the user's context. In this paper, we propose a novel algorithmfor context-aware proactive caching. The algorithm learns context-specificcontent popularity online by regularly observing context information ofconnected users, updating the cache content and observing cache hitssubsequently. We derive a sublinear regret bound, which characterizes thelearning speed and proves that our algorithm converges to the optimal cachecontent placement strategy in terms of maximizing the number of cache hits.Furthermore, our algorithm supports service differentiation by allowingoperators of caching entities to prioritize customer groups. Our numericalresults confirm that our algorithm outperforms state-of-the-art algorithms in areal world data set, with an increase in the number of cache hits of at least14%.
展开▼