Collaborative Self-Localization Techniques for Wireless Image Sensor Networks
Huang Lee, Hamid Aghajan · 2006
We introduce a novel localization technique that can jointly estimate the locations of a moving target and the sensor nodes in a wireless image sensor network. The proposed method is based on in-node image processing and can be implemented in a decentralized or clustered fashion. In our approach, two image sensors are used to define a relative coordinate system. In order to synchronize the observations, the node defined as the origin broadcasts packets that trigger image capture at other nodes. In the decentralized version of the technique, each one of the two reference nodes broadcasts its image plane position of the moving target at a few time instances. Each of the other nodes in the network that can detect the target in its image plane upon receiving a number of triggering broadcasts, calculates its own relative coordinates and orientation as well as the coordinates of the observed target. In the clustered version of the proposed technique, observations gathered by the nodes within a neighborhood cluster are sent to a cluster-head, which can be the reference node at the origin. The cluster-head combines the data and calculates the coordinates of the target and all the nodes that contributed observations. Experimental results are provided to verify the performance of both versions of the proposed algorithm.