Framework towards developing a stability heuristic for cluster computation in MANETs

Hierarchical routing schemes in an ad-hoc environment outperform the flat routing schemes. Several algorithms like Lowest ID, LCC, Highest in-degree, WCA, IWCA, neural network based etc. have been proposed for clustering of nodes but none of them take into account the environment specific dynamic nature of a heterogenous ad-hoc network. They do not examine the combined effect of parameters like battery power, degree of node and mobility on cluster formation. Although these factors can be considered as inputs to a neural network, training the network and choosing the training algorithm is a computationally intensive hence time consuming step. In this letter we address this issue by computing a Stability factor for deciding cluster-heads. This factor is independent of the underlying environment, computationally un-intensive and takes into account environmental changes. It involves no GPS like schemes to measure mobility which clearly assumes a pre-existing facility in every computing device acting as a member of an ad-hoc network to measure the position of another node or relies on an external device to convey the position. The Stability factor also takes care of the interference anomaly - which we define as a false alarm resulting in change of a cluster-head due to a decrease in the received power levels at a node. This change in cluster-head is not due to any relative motion between them. The stability factor calculation could easily be built into a software and can be deployed for cluster-head calculation in any ad-hoc environment with no underlying assumptions.