A location and mobility independent scheme to quantify the neighbourhood stability of a node in mobile sensor networks

Abstract

We propose to quantify the neighbourhood stability of a node without using the location and mobility information of the nodes in a mobile sensor network (MSN). We make use of the notion of Jaccard similarity to quantify the similarity in the sets of neighbours of a node between any two successive sampling time instants and compute the weighted average of the Jaccard neighbourhood similarity (WJNS) over a period of time. Our hypothesis is that nodes with larger WJNS scores have a stable neighbourhood and could be preferred for inclusion as intermediate nodes for stable paths. We validate our hypothesis through extensive simulations: The lifetime of the WJNS-based paths is significantly larger than that of the minimum hop paths and the predicted link expiration time (LET)-based paths; the trade-off being a marginal increase in the hop count compared to that of the minimum hop paths.