R²T: A Rapid and Reliable Hop-by-Hop Transport Mechanism for Information-Centric Networking

Hop-by-hop transport is promising in information-centric networking because of its adaptability to in-network caching and lossy environments. Mobility first transport protocol (MFTP), as a classical example, transports content chunks and waits for the retransmissions of the lost packets in each hop. However, due to its chunk-level reliability, the MFTP hops transport chunks in series rather than in parallel, causing significant longer end-to-end latency than that in transmission control protocol (TCP). To address this issue, we propose a novel hop-by-hop transport mechanism, R²T. In each hop, R²T guarantees the content reliability in packets rather than in chunks. R²T routers immediately forward all the received packets downstream so that the end-to-end latency can be remarkably decreased. With separated packet name space in global and local scale, R²T decouples the hop-by-hop reliability control and the end-to-end content transport, obtaining both performance advantages from MFTP and TCP. The evaluation results show that R²T has as low transport latency as TCP, yet keeps as high bandwidth usage as MFTP. Meanwhile, it is demonstrated that R²T does not suffer significant memory and energy consumption, achieving promising feasibility in real implementations.