Network Coding-based Content Retrieval based on Bloom Filter-based Content Discovery for ICN

This paper presents a complete framework for content discovery and retrieval in Information-Centric Networks. For content discovery, we implement a method similar to our previously developed pull-based BFR [1], which uses Bloom filter-based signaling to inform servers about the name prefixes of available requests. For content retrieval, we propose in this paper a feedback-based cooperative protocol implementing network coding-based forwarding. The proposed network coding-based protocol provides a distributed solution to control the multi-session codeblock size, i.e., the number of variables that are combined into network coded packets, by setting a capacity constraint on each node and by piggybacking the available capacity as feedback on messages sent to neighbors. The network codes are decided using linear programming. We compare the proposed network coding-based protocol with push-based BFR [2] and pull-based BFR [1]. The results show that the proposed protocol outperforms both push-based BFR and pull-based BFR in terms of content discovery overhead and average content block retrieval delay.