Dissimilarity metrics for probabilistic broadcasting in wireless multi-hop networks

Broadcasting is the main mechanism to spread out information in all-to-all fashion in wireless multi-hop networks at various layers of the communication stack. It is widely employed in applications such as disseminating emergency information, the discovery phase of routing protocols, and the maintenance of routing information. Among the broadcast approaches, probabilistic methods present several advantages such as reduced overhead, resilience against failures and mobility of nodes, and better balance of power consumption. Among the parameters used to tune the forwarding probability in probabilistic broadcasting, the Euclidean distance is one of the most used. However, adjusting the forwarding probability based on the Euclidean distance presents some issues. First, nodes require a positioning system, which is not suitable for indoor scenarios. Second, some broadcasting approaches based on Euclidean distance require that nodes know their radio transmission range, which can be different from the nominal value depending on external factors like interferences and obstacles. We present several dissimilarity metrics based on the dissimilarity of the neighborhood of nodes and demonstrate that they are well-suited for probabilistic broadcasting protocols.