A Per-Face Popularity Scheme to Increase Cache Robustness in Information-Centric Networks

Abstract

In-network storage is an essential element of nearly all Information-Centric Networking (ICN) architectures. Thus, protecting the cache of popular items in the network storage devices is crucial. With cache pollution attacks, an attacker seeks to push legitimately popular items out of the cache in favor of less popular items. This paper presents a per-face popularity (PFP) scheme to mitigate effects of cache pollution attacks and to increase overall cache robustness in Information-Centric Networks. The scheme normalizes the contributions from each face so that Interests arriving over any one face has no more effect on the popularity than any other face. As a result, the items that remain in the cache are those that have legitimate popularity, represented across all faces. The experimental results confirm its effectiveness against cache pollution attacks over one or a relatively small percentage of the faces, and demonstrate its overall effectiveness in preserving popular items and increasing hit rate.