Convolutional network codes (CNC) for cyclic graphs are analyzed. A CNC requires a smaller memory for each coding coefficient at edge e. For CNC the overhead is the initial delay before the sinks start receiving symbols. It was shown that the CNC with the sequential decoder above achieves good performance for some networks. It was found that networks in which there is worst case delay is also improved by CNC.
|Number of pages
|IEEE International Symposium on Information Theory - Proceedings
|Published - 2004
|Proceedings - 2004 IEEE International Symposium on Information Theory - Chicago, IL, United States
Duration: 27 Jun 2004 → 2 Jul 2004