Distributed clustering algorithm in dense group-based ad hoc networks
Raphaël Massin, Christophe J. Le Martret, Philippe Ciblat · 2016
For dense ad hoc networks, clustering is an appropriate strategy to efficiently organize the network. Moreover, public safety or military networks are structured through a hierarchical organization via operational groups. This organization has an impact on both the mobility of nodes which move in groups, and the data flow since the traffic is mainly intra-group. In this work we propose a novel distributed clustering algorithm suited to such networks, called Dynamic Clustering with Operational Groups (DCOG). This algorithm is designed in order to achieve the following properties: each cluster includes the highest possible number of members of some operational groups, and each cluster size is the closest possible to a given maximum. We first prove the theoretical convergence of DCOG and then compare by simulation its performance against five other clustering schemes from the literature. Our simulations show that DCOG leads to a lower end-to-end communication delay and offers a better stability to mobility.