Research On Secure Network Coding Against Eavesdropping And Contamination

With comprehensive application of network coding, there has been increasing attention to thesecurity of network coding. There exist two kinds of adversaries in communication networks usingnetwork coding: eavesdropping and contamination. In this dissertation, two adversaries are studiedand efficient secure network coding algorithm is designed. All of the researches in this dissertationare in the multicast network with single source node and several sink nodes.Firstly, a weakly secure coding algorithm against wiretap is constructed, which is based onVandermonde matrix and hash function.At the source node, a Vandermonde matrix is structured byusing hash function and source information is transformed with this matrix. It’s proved that it canmeet the demand of weakly security using certain network coding while the ability of eavesdropperis not restricted. Using it in random network coding, when the eavesdropper doesn’t know the hashcode, the weakly secure condition can be achieved at probability of one. It can reach the networkmulticast capacity under having secret link between source and sink nodes. Aiming ateavesdropping attack, an information-theoretic secure network coding algorithm against wiretap isconstructed, which is based on cryptography. When using certain linear network coding, codessatisfy information-theoretic security. Under secret link model, this algorithm can be improved toreach the network multicast capacity. When the eavesdropper has limit ability of computation, thisalgorithm can be improved to use in random network coding. Aiming at contamination attack, anew efficient secure network coding algorithm against contamination is also provided, which isbased on homomorphism hash function and Diffie-Hellman key exchange protocol. It’s proved thatthis algorithm can resist pulution effectively. Sink nodes verify the Signature and intermediatenodes only compute and transmit Signature. The simulation results verify the performance of thisalgorithm.