Robust localization of multiple events in sensor networks

Localization of events is a canonical problem for sensor networks, especially for applications such as intrusion detection or neighborhood monitoring. We employ a general sensor model which models various types of sensor nodes. However, traditional triangulation method will be unsolvable when the sensor nodes are positioned on a line. To overcome such ill-conditioned cases, we can use MAP or MMSE methods to estimate event locations. Moreover, to locate multiple simultaneous events and overcome the ill-conditioned case when the sensor nodes are collinear, we need to solve a system of nonlinear equations. We propose a novel method which solves this system of nonlinear equations and multiple-event locations can be found. Experiments and error analysis of the method demonstrate the performance of our methods