A performance comparison of protocols for clustering and backbone formation in large scale ad hoc networks

Abstract

This work concerns the comparative performance evaluation of protocols for clustering and backbone formation in ad hoc networks characterized by a large number of resource-constrained nodes. A typical example of these networks are wireless sensor networks. The DCA protocol represents those protocols whose backbone construction method is based on selecting nodes as clusterheads and then joining them to form a connected backbone. The algorithm proposed by Wu and Li has been chosen to exemplify those algorithms that build a connected backbone and then prune away redundant nodes. Finally, the algorithm by Wan et al. has been considered here for its more theoretical properties of producing a backbone with a constant approximation factor, linear time complexity and optimal message complexity. In order to obtain a backbone reasonably small at reasonable cost we propose an enhancement of the DCA algorithm, termed DCA-S, which enriches the DCA backbone construction with a recently proposed and resource effective sparsification rule. DCA-S leads to a robust backbone close in size to that generated by the Wan et al. protocol without significantly degrading the performance in terms of all the other relevant metrics.