Steps Towards Decentralized Deterministic Network Coding

Recently, Li and Pan proposed a deterministic network coding resource optimization method built to overcome the issue of severe performance degradation of network coding in the presence of erasures. Considering a multicast scenario, their method relied on the use of Ant Colony Optimization (ACO) to find shortest disjoint paths from a source to each sink and identified the coding nodes as the heads of the overlapping path segments among multiple sinks. Although their technique appears to be suitable for the small artificial topologies on which it has been exemplified, there is no clear study regarding the scalability of the proposed solution for larger Internet-like networks where properties such as clustering are observed. Throughout this paper we point out and provide solutions to several problematic aspects that were not previously brought up for discussion. We conduct extensive network measurements on scale-free networks, deriving a probability function for link erasures and construct a software simulation platform to assess and enhance the practicality of their solution.