On Routing-Optimal Network for Multiple Unicasts

In this paper, we consider networks with multiple unicast sessions. Generally, non-linear network coding is needed to achieve the whole rate region of network coding. Yet, there exist networks for which routing is sufficient to achieve the whole rate region, and we refer to them as routing-optimal networks. We identify a class of routing-optimal networks, which we refer to as information-distributive networks, defined by three topological features. Due to these features, for each rate vector achieved by network coding, there is always a routing scheme such that it achieves the same rate vector, and the traffic transmitted through the network is exactly the information transmitted over the cut-sets between the sources and the sinks in the corresponding network coding scheme. We present more examples of information-distributive networks, including some examples from index coding and single unicast with hard deadline constraint.

[1]  Zhen Zhang,et al.  The Capacity Region for Multi-source Multi-sink Network Coding , 2007, 2007 IEEE International Symposium on Information Theory.

[2]  Robert D. Kleinberg,et al.  On the capacity of information networks , 2006, IEEE Transactions on Information Theory.

[3]  Randall Dougherty,et al.  Insufficiency of linear coding in network information flow , 2005, IEEE Transactions on Information Theory.

[4]  Ziv Bar-Yossef,et al.  Index Coding With Side Information , 2011, IEEE Trans. Inf. Theory.