Network coding, first studied by Yeung and Zhang and Ahlswede et al., reveals that if coding is applied at the nodes in a network, rather than routing alone, the network capacity can be increased. Li et al. and subsequently Koetter and Medard proved that linear network coding is sufficient to achieve the maximum capacity in a single-source finite acyclic network.
Consequently, linear network coding for single-source finite acyclic networks has been a subject of much research interest. We refer the reader to (see also) for a tutorial on the subject. In this work, they classify linear network codes for single-source finite acyclic networks into four classes: generic; linear dispersion; linear broadcast; linear multicast. These four classes of linear network codes possess properties of decreasing strength.