We consider a dynamic content caching problem wherein the contents get updated at a central server, and local copies of a subset of contents are cached at a local cache associated with a Base station (BS). When a content request arrives, based on the content’s availability and age in the local cache, the BS decides whether to fetch a fresh version from the central server or serve the cached version from the local cache. Fetching a content incurs a fixed fetching cost, and serving the cached version incurs an ageing cost proportional to the age-of-version (AoV) of the content. We formulate an optimal content fetching and caching problem to minimize the average cost subject to cache capacity constraints. We frame it as a continuous time restless multi-armed bandit process (RMAB), where each arm represents a content. We prove the indexability of the single content problem and provide a Whittle index based solution to the joint problem. We subsequently allow a “wait” option for the BS whereby it can also wait for additional requests before fetching and serving a fresh content. The “wait” option introduces queue lengths to the system’s state. It also substantially changes the state dynamics, the analysis and the caching policy. We again derive a Whittle index based policy. We compare the performance of the proposed policies to recent works and also show that the proposed policies are “almost” optimal via simulations.
Ankita Koley is a PhD student at the Department of Electronic Systems Engineering, Indian Institute of Science. She obtained her B.Tech in Electrical Engineering from the National Institute of Technology, Durgapur, in 2018. Prior to joining IISc, she was a commercial executive at Calcutta Electric Supply Corporation from 2018 to 2020. Her research interests include communication networks, edge caching, edge computing, and optimization. She is currently working on caching and delivery of dynamic content.