Performance Analysis of Probabilistic Caching Scheme Using Markov Chains

This paper presents a new analytical model to analyze the performance of a probabilistic caching scheme with various cache replacement policies in content-centric networks. The cache replacement policies include Random Replacement (RR), First In First Out (FIFO), and Least Recently Used (LRU). This analytical model is based on Markov chains under Independent Reference Model (IRM) and Zero Download Delay (ZDD) assumption. A closed-form expression of the stationary distribution of cache state is derived and is used to compute the hit rates of caching systems. Moreover, we use this model to establish several important properties of the probabilistic caching scheme as well as the guidelines on effectively using it. Results of computer simulations show that the proposed analytical solution can model the probabilistic caching scheme very accurately.