Management of Target-Tracking Sensor Networks -

Management of Target-Tracking Sensor Networks

Target tracking has emerged as an important application of sensor networks. There are two subproblems inherent to target tracking. The first is the initial location of the target as it enters the region being covered. The second is following its track once it has been discovered.

In this work, we outline an approach to target tracking. Our energy management schemes perform better in terms of track quality and have an energy consumption similar to other schemes.

We present a multitarget tracking algorithm; in connection with that, we also present a filtering algorithm that improves the quality of tracking. We also study adaptive approaches to manage the tracking process to the observed mobility characteristics of the target. Such adaptive approaches are shown to have noticeable performance advantages.

Most of our work involves sensors using a 0/1 detection model. In other words, a sensor reports that it has detected the target; it does not report the signal strength. Obviously, such information has the potential for improving our target position estimates. We provide results in this work to quantify the extent of such improvement.

Adaptive approaches, which allow the system to react to perceived parameter values in the operating environment, have obvious potential for improving system performance.

We study the advantages of learning the intruders’ mobility model parameters and then adapting the tracking parameters appropriately. We are considering problems associated with sensor miscalibration or drift when signal strength information is used for target esti- mation.

To counter the inaccuracies that then result, we are developing approaches which seek to correlate inputs from neighboring sensors and to carry out mutual recalibration.

Finally, we are working on extending our analytical model to handle non-uniform sen- sor distributions.

To read more of this external content, download the complete paper from the online archives at the University of Massachusetts.

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.