Abstract
A wireless sensor network (WSN) is useful for personal communication, sensor data collection, emergency services, and event monitoring (such as radio transmissions, earthquakes and tsunamis) in an environment, including the geo-location of the source of the event. WSNs can cover short as well as long ranges. Determining the location of the source event can be undertaken in different ways in a wireless network. Traditionally, geo-location is conducted using a small number of high-accuracy sensors, and measurements from individual sensors are transmitted to a central processing facility to determine the target location. There are three key problems with this type of approach: high accuracy in individual sensors
translates to high cost in construction and calibration, the locations of the sensors on the baseline must be carefully planned based on an anticipated target location, and the information must be transmitted to a central node for processing. Here, in the context of a large network of low-accuracy sensors, the individual locations of which do not need to be carefully planned, we address the latter issue and investigate the centralized and decentralized application of centroid localization of a target at an unknown location. In particular, we develop a decentralized processing procedure,
based on the Gossip algorithm, and show that it performs as well as the centralized procedure for pair-wise comparison. By performance and sensitivity analysis, we show the robustness of this novel decentralized algorithm and that it works with any kind of wireless network.