Convolutional network codes
Elona Erez, Meir Feder · 2004
Convolutional network codes are considered for cyclic graphs. In CNC each node receives several streams and generates output streams whose current symbols depend on the current input symbols and previous input symbols in the node memory. A multicast CNC can be constructed using an algorithm, in order to minimize the memory and overhead, coefficients of lower polynomial degree are drawn to consideration. For CNC the overhead is the initial delay before the sinks start receiving symbols. CNC with the sequential decoder achieves good performance for some networks.