Storage Efficient Edge Caching with Time Domain Buffer Sharing at Base Stations

摘要

Edge caching has attracted great attention recently due to its potential for reducing the service delays and the peak rate demand, especially when the quality-of-service (QoS) and data rate requirements of mobile users are ever increasing. One of the key issues in edge caching is the storage efficiency. To achieve high storage efficiency, we present an edge caching policy with time domain buffer sharing. More particularly, our scheme allows a Base Station (BS) to determine whether and how long a content item should be cached at the buffer of the BS. To this end, we formulate a queue-theoretic model, in which the storage cost can be determined by the maximum caching time of content items via Little's Law. Based on this model, we present a probabilistic caching policy with random maximum caching time to strike the optimal tradeoff between the storage cost and the overall hit ratio of content items. For content items having different popularity, we further investigate how the storage resources should be allocated among these content items. An efficient two-layer iteration algorithm is presented to solve the storage allocation problem, which is a nonconvex optimization problem.