Neighborhood Overlap-based Stable Data Gathering Trees for Mobile Sensor Networks

Abstract

The hypothesis in this research is that the end nodes of a short distance link (the distance between the end nodes is significantly smaller than the transmission range per node) in a mobile sensor network (MSN) are more likely to share a significant fraction of their neighbors and such links are more likely to be stable. The author proposes to use Neighborhood Overlap (NOVER), a graph-theoretic metric used in complex network analysis, to effectively quantify the extent of shared neighborhood between the end nodes of a link and thereby the stability of the link. The author's claim is that links with larger NOVER score are more likely to be stable and could be preferred for inclusion while determining stable data gathering trees for MSNs. Through extensive simulations, the author shows that the NOVER-based DG trees are significantly more stable and energy-efficient compared to the DG trees determined using the predicted link expiration time (LET). Unlike the LET approach (currently the best known strategy), the NOVER-based approach could be applied without knowledge about the location and mobility of the nodes.