Security overhead of random linear network coding in multicast relay networks

The relays in random linear network coding (RLNC) approach recode and transmit the coded packets to boost the performance against packet loss. However, this can endanger the system to pollution attacks, which can be eliminated by Homomorphic message authentication code (HMAC). Cache coding limits RLNC by coding only the decoded packets in the relays. In the other words, the relay node in cache coding first decodes the coded packets, then recodes and protects them by a MAC-based scheme that allows the next hop to discover the pollution in the network. This study provides analytical models for cache coding and RLNC to compare their communication overhead. First, we approximate the number of transmissions in a network consisting several relays. The approximation models are for the cases of a few and large number of relays. Then, we validate our analysis using simulation results and illustrate that cache coding can eliminate the pollution attack issue of RLNC with only a slight increase in the bandwidth usage when there are no malicious nodes in the network. Moreover, we conclude that it is more beneficial to employ the cache coding approach at the presence of malicious nodes in the network.