Computing and Informatics

Tracking moving targets is one of the important problems of wireless sensor networks. We have considered a sensor network where numerous sensor nodes are spread in a grid like manner. These sensor nodes are capable of storing data and thus act as a separate datasets. The entire network of these sensors act as a set of distributed datasets. Each of these datasets has its local temporal dataset along with spatial data and the geographical coordinates of a given object or target. In this paper an algorithm is introduced that mines global temporal patterns from these datasets and results in the discovery of linear or nonlinear trajectories of moving objects under supervision. The main objective here is to perform in-network aggregation between the data contained in the various datasets to discover global spatio-temporal patterns; the main constraint is that there should be minimal communication among the participating nodes. We present the algorithm and analyze it in terms of the communication costs.