On the Eavesdrop Vulnerability of Random Network Coding over Wireless Networks

When random network coding is combined with path diversity, source data packets are mixed and distributed spatially across different parts of the network, providing some inherent protection against eavesdropping. In a wireless setting, the links may be of poor quality and are subject to intercept by a sophisticated eavesdropper at potentially significant distances from the network. Thus, this inherent protection against eavesdropping is likely to be fragile. In this paper, we propose a quantitative metric for eavesdrop vulnerability based on radio link budget considerations. Using the famous butterfly network as a simple example, we confirm that the wireless network is increasingly vulnerable as the eavesdropper´s capabilities increase or the average link quality of the network degrades. Under critical conditions, the eavesdropper is able to decode session traffic with high reliability by observing the transmissions of any relay node - i.e., no security is provided at all.