Analysis of a transport protocol based on rateless erasure correcting codes

This paper presents a protocol designed as an alternative to classical TCP for channels which experience high loss-rates. The protocol is simplified with respect to TCP by eliminating the need of retransmissions and the associated buffers. This is achieved by applying a rateless erasure correcting code to the data that is going to be transferred. A modified version of the TCP congestion control algorithms is used in order to better differentiate between losses caused by errors on the channel and losses caused by congestion in the network. A statistical model for the steady-state throughput of the protocol is included in the paper as well. Using simulations performed in ns-2 the throughput model is validated. Other aspects of the behavior of the protocol are also investigated. The results are promising, mainly showing that the protocol does not experience any significant performance degradation in terms of steady-state throughput, even if the loss rate on the channel is high.

[1]  David R. Smith,et al.  Digital Transmission Systems , 1980 .

[2]  V. Jacobson,et al.  Congestion avoidance and control , 1988, SIGCOMM '88.

[3]  B. R. Badrinath,et al.  I-TCP: indirect TCP for mobile hosts , 1995, Proceedings of 15th International Conference on Distributed Computing Systems.

[4]  Satish K. Tripathi,et al.  Split TCP for mobile ad hoc networks , 2002, Global Telecommunications Conference, 2002. GLOBECOM '02. IEEE.

[5]  Michael Luby,et al.  LT codes , 2002, The 43rd Annual IEEE Symposium on Foundations of Computer Science, 2002. Proceedings..

[6]  David Mazières,et al.  Rateless Codes and Big Downloads , 2003, IPTPS.

[7]  Rajesh Krishnan,et al.  Explicit transport error notification (ETEN) for error-prone wireless and satellite networks , 2004, Comput. Networks.

[8]  Bruno Sericola,et al.  A Markov model of TCP throughput, goodput and slow start , 2004, Perform. Evaluation.

[9]  Radu Theodorescu,et al.  Neyman type A distribution revisited , 2005 .

[10]  Amin Shokrollahi,et al.  Raptor codes , 2011, IEEE Transactions on Information Theory.

[11]  Anirban Mahanti,et al.  An analytic throughput model for TCP NewReno , 2010, TNET.