Construction of convolutional network coding for cyclic multicast networks

In this paper, we present a practical algorithm to construct the convolutional multicast network coding over any finite directed cyclic network. The dual line graph of a directed cyclic graph is considered as a system. By regarding the global encoding kernels in the original graph as the corresponding inputs or outputs of some subsystem and the local encoding kernels in original graph as gains of channels, we can construct the convolutional network code through randomly choosing the local encoding kernels of the directed cycles in networks. By using Mason formula, the construction becomes very efficient.