A network cost function for clustered ad hoc networks: Application to group-based systems

Many wireless public safety and military networks are ad hoc networks, for which clustering is a well-known strategy to improve scalability. In addition these networks are structured according to a hierarchical organization, i.e. nodes belong to specific operational groups which implies that the traffic is mainly intra-group. So far, clustering algorithms are built using metrics such as node identifiers, node mobility, etc., and thus do not take into account the network hierarchical structure. The goal of this paper is therefore twofold: i) specify a way to benchmark clustering solutions from a system point of view, and ii) thanks to this benchmark determine the importance of using operational group information to build clusters. Therefore we define a novel network cost function based on additive metrics (e.g. delay) incorporating the traffic structure and the inter-clusters communications costs. Thanks to this function we show that the clustering solutions providing the best QoS to the end user depend on the group structure.