Towards effective defense against pollution attacks on network coding

Abstract

Network coding provides a promising alternative to the traditional store-and-forward transmission paradigm. For the system using random linear network coding, the adversary could inject corrupted messages into the networks by compromising the network nodes, which is known as the pollution attack. Corrupted messages injected by the adversary, if undetected, could cause a devastating impact to the network performance. To address this issue, a number of pollution attack defense schemes for network coding have been developed in the recent years. The overhead caused by defensive techniques against pollution attacks should be low especially in wireless networks with limited resources. However, there is lack of a systematical strategy for evaluating those schemes and establishing a foundation for designing attack pollution defense schemes for network coding in wireless networks. Towards this end, we first develop the taxonomy of existing network coding authentication schemes. To fairly compare the effectiveness of those schemes, we conduct theoretical analysis and implement different schemes within the same security level. Our extensive simulation and implementation results validate our findings well. Our research summarizes the state-of-art research development and lay out future directions in this area.