On the overhead and throughput performance of scent-based MANET routing

Biologically inspired networking paradigm introduces throughput gains in an adaptive and robust operation over the classical routing algorithms designed for mobile ad hoc networks (MANETs). However, this performance gain comes at the price of additional overhead as in the case of ant routing protocols. In this context, we present the Scented Node Protocol (SNP) to reduce the overhead requirement while maintaining the throughput improvement of biologically inspired routing solutions. In particular, we assume that sources send ant-based control packets for route discovery, which is assisted by the scent dissemination from destinations. The scent messages propagate over the network and reflect the popularity of the corresponding destinations. We show the throughput and overhead improvement of SNP compared to the existing solutions of ant colony optimization and MANET routing. In the asymptotic model where the node density grows, we analyze the maximum scent range and evaluate the overhead requirement that has the same asymptotic order for ant-based control packets and scent dissemination. Then, we consider the local dynamics of scent message exchange between neighboring nodes. The results show how the finite overhead constraint limits the scent range and how network coding can be applied locally to reduce the SNP overhead.