Robustness of Random Network Coding to Interfering Sources

Abstract

Random network codes are considered for multicast scenarios where not all of the sources are required by the sinks. These unwanted sources could model interfering or even adversarial sources. The objective of the paper is to determine the effect of the interfering sources on the probability of successful decoding. A lower bound on the success probability is determined. The bound is the product of the probability that the random network code is feasible without interference, multiplied by a robustness factor. The robustness factor uniquely depends on the in-degree of the sink and the size of the finite field. Analysis of the robustness factor shows that the effect of interference can be completely mitigated through a small increase in the field size.