A self-adaptive probabilistic packet filtering scheme against entropy attacks in network coding

In this paper, based on a novel self-adaptive probabilistic subset linear-dependency detection (S-PSLD) algorithm, we propose an efficient packet filtering scheme against entropy attacks in network coding. The scheme verifies the received packets probabilistically instead of exactly, and thus it can rapidly filter out the resultant packets from entropy attacks. Moreover, to minimize the packet detection cost at forwarder while keeping the false positive rate at an expected low level, a self-adaptive algorithm is introduced such that each forwarder can dynamically tune the system security parameters according to the available bandwidth or the number of the received packets in buffer. Theoretical analysis and performance evaluation are given to demonstrate the validity and efficiency of the proposed scheme.