A Distributed Algorithm for Waking-up in Heterogeneous Sensor Networks

Abstract

In this paper we present a distributed, application-morphable, algorithm for waking up appropriate sensor nodes in a heterogeneous sensor network. We assume a sensor field consisting of a large number of low power, limited functionality, tripwire nodes and a smaller number of powerful, energyhungry, tracker nodes. Our problem is that when an event is detected by a set of tripwire nodes a specific number of appropriate tracker nodes needs to be woken up. These tracker nodes will subsequently collaborate to perform the sensing task required by the application. Waking up non-suitable tracker nodes or employing more trackers than necessary for a specific task, can lead to significant waste of network resources (e.g. energy). The application indicates the number of nodes that are needed for a sensing task, as well as an optimization function to be used by the algorithm. Therefore, our algorithm is isolated from most application details and is simple and general enough to accommodate a wide range of sensing applications. We prove that our algorithm converges to a uniform optimal global decision for specific classes of optimization functions. Furthermore, we show that it is fast enough (<100ms) to be practical for most sensing applications and exhibits good performance in terms of total messages exchanged. Finally, we demonstrate that our algorithm is very robust, managing to retain its correct and efficient behavior for a wide range of scenarios, even under hostile environmental conditions (e.g. link loss probabilities up to 35%).