Efficient Weakly-Secure Network Coding Schemes against Wiretapping Attacks

Network coding is promising for improving network throughput in various networking systems. However, the systems adopting network coding are vulnerable to wiretapping attacks, in which adversaries wiretap on a subset of links and solve for information symbols that are intended to be decoded only by receivers. A number of solutions have been proposed to prevent information leakage to wiretappers, however, they either enlarge the finite field over which the coding is done and thus consume more communication bandwidth, or reduce the multicast rate due to inserting random numbers into the source. In this paper, we propose two efficient coding schemes against wiretapping attacks. Our schemes are weakly-secure, i.e., the wiretapper can obtain some linear combination of the information symbols, but he cannot solve for any single one. Both schemes utilize a permutation function to randomize the message vector sent by the source. The first scheme inserts only one random symbol into the source; the second scheme inserts no random symbols at all and thus achieves the maximum multicast capacity. Moreover, the second scheme retains the same finite field size as insecure coding schemes, thus does not consume more bandwidth to transmit symbols and encoding coefficients. Our schemes are lightweight: the permutation function can be implemented using Linear Feedback Shift Register (LFSR) or its variants, which can be efficiently implemented in hardware.