Efficient information gathering from large wireless sensor networks

Abstract

Considering large WSNs and comparing data aggregation approaches in terms of scalability, robustness, completeness, and time/energy effectiveness, recent research has asserted the superiority of serial structure-free approaches over serial structure-based ones, and over both parallel structure-free and parallel structure-based techniques. But, in spite of the fact that serial structure-free approaches excel in large and medium-scale networks, their underlying path-construction algorithms are not optimal and can be improved by reducing the involved communications and further shortening the visiting path. To respond to this need, this paper presents Geometric Serial Search (GSS); a new serial structure-free algorithm specifically designed to efficiently gather information from large wireless resource-constrained networks. In addition to its completeness (i.e., visiting all nodes), collision-free nature, high scalability, energy/time efficiency, and robustness against topology changes (failures in links/nodes, …), the main advantage distinguishing GSS is that it considerably reduces communications and always approaches the optimal number of hops. More precisely, in GSS, no control packets or complex data structures are required, instead, one packet hops from node to node and explores the entire network. While gradually finding its way through the network, this packet interrogates nodes and collects their responses at the same time. The followed path is not established in advance, can stem from any node in the network, and requires only the one-hop neighborhood information of each traversed node to be gradually drawn. The obtained OMNeT++ simulation results presented in this paper demonstrate the efficiency of GSS and confirm all the previously cited claims.