Diversity-inspired clustering for self-healing MANETs: Motivation, protocol, and performability evaluation

Swarm systems, which typically comprise a large number of lightweight mobile components, must be capable of self-healing. In this paper, we propose a self-organizing, self-healing framework called “superimposed” clustering for such systems. The framework makes a significant departure from traditional clustering algorithms that apply a single policy to form clusters through iterations. Specifically, our superimposed clustering protocol (SCP) selects a pair of diversified clustering polices to simultaneously build two sets of clusters, which we view as two cluster layers with one on top of the other. Via redundancy shadowing, SCP is able to extract and combine the complementary portions of the two layers to form a clustered network such that the vast majority of nodes can be organized through a single round. Moreover, SCP exploits shadow redundancy to enable gracefully degradable clustering coverage to mitigate cluster damage caused by node failure, death, or migration. We present the notion of superimposed clustering by devising a protocol and conducting a performability evaluation.