Content Pushing With Request Delay Information

Abstract

Pushing along with proactive caching, in which content items are transmitted prior to being requested by a user, holds the promise of trading scalable storage resources for substantial throughput increase in content-centric networks. In this paper, a joint pushing and caching (JPC) method is studied, which determines when and which content to be pushed to, and to be removed from, the receiver buffer based on content request delay information (RDI) that predicts a user’s request time for certain content items. Both offline and online JPC policies with noncausal, statistical, and causal RDI are proposed based on a greedy algorithm and dynamic programming. The effective throughputs of the policies are evaluated, and they are seen to increase with both the receiver buffer size and the pushing limited channel capacity. The analysis also reveals the fundamental tradeoff between communication and storage resources. Furthermore, RDI feedback is shown to significantly enhance the performance of the online JPC without incurring much signaling overhead or complexity, especially in small buffer scenarios.