Secure Caching with Non-Uniform Demands

Caching helps in reducing network load at times of heavy traffic by utilizing content duplication, local file storage and multicast content delivery. The multicast delivery could be prone to information leakage if a user becomes malicious or if the multicast link is tapped by an external adversary. In this paper, we introduce the concept of information-theoretically secure file delivery when the files have a non-uniform popularity distribution. For this scenario, the cache memory of each user is split into two parts: one part is used to store data (some function of the files), whereas the other part is used to store keys. A novel algorithm for storing data and keys in the users' cache while accounting for the non-uniform file popularities is presented. This design is then exploited to simultaneously reduce the delivery rate while keeping the multicast communication secure. We show that the proposed scheme can guarantee information theoretic security at a negligible cost compared to insecure schemes, especially for large number of files and users. The secure-rate-vs-memory trade-off of the proposed scheme is analyzed and shown to be close to the information theoretic optimal for all feasible values of problem parameters.