The Design of Dynamic Probabilistic Caching with Time-Varying Content Popularity

In this paper, we design dynamic probabilistic caching for the scenario when instantaneous content popularity may vary with time while it is possible to predict average over a window. Based on popularity, optimal probabilities can be found, e.g., from solving optimization problems, and existing results in literature implement via static placement. The objective of work that: i) converge (in distribution) under time-invariant ii) adapt time-varying popularity. Achieving above requires novel replacement because cannot varying classic policies, such as LRU, target (as they do not exploit any information). We model policy problem finding state transition probability matrix Markov chain propose method generate refine matrix. Extensive numerical are provided validate effectiveness proposed design.