• Increase font size
  • Default font size
  • Decrease font size
Home Projects Past Projects Multimodal Sensor Networks

Multimodal Sensor Networks

E-mail Print PDF


We investigate the benefits that multiple modalities bring to the problems of target and self localization in sensor networks. Specifically,our work focuses on coupling motes equipped with magnetometers with cameras.

We first looked at how feedback from cameras can be used to calibrate the magnetometers. In our model the magnetic field signal strength follows a power law with unknown exponent beta. The goal of the calibration process is to estimate the value of beta. The process works as follows: Motes use trilateration to estimate the position of the target. This information is then passed to the cameras which locate the actual position of the target. We then solve a non-linear optimization problem to minimize the distance between the target's actual location and the location estimated using the magnetometer measurements. Our simulation results show that the estimation process converges in a small number of iterations.


Next, we designed an integrated system for target tracking using cameras and magnetometers. The proposed system initially uses two pre-calibrated, PTZ cameras to localize the network's motes. To do so, each mote is equipped with a LED and the cameras scan the scene until they both have the LED at their center of view. Then, the camera angles are used to estimate the location of the mote through a generalized triangulation formulation. Note that we do not require that the locations of the cameras are known. Once the mote locations are known, cameras go to sleep to conserve energy. The motes then collaborate to estimate the location of the target using their non-imaging sensor (i.e. magnetometers). Once a target is located, the cameras are cued to capture an image of the target. We have evaluated the performance of the self- and target-localization mechanisms through simulation and implementation on a small testbed using MicaZ motes. Details of these results can be found on our ISSNIP paper.


  • Andreas Terzis, JHU
  • Roberto Garcia, JHU
  • Ryan Farrell, UMD
  • I-Jeng Wang, JHU/APL
  • Dennis Lucarreli, JHU/APL
  • Min Ding, GWU


  • Ryan Farrell, Roberto Garcia, Dennis Lucarellu, Andreas Terzis, I-Jeng Wang. Localization in Multi-Modal Sensor Networks. In the Proceedings of the third International Conference on Intelligent Sensors, Sensor Networks and Information Processing (ISSNIP) 2007. PDF
  • M. Ding, A. Terzis, I-J. Wang, D. Lucarreli, Multi-modal calibration of surveillance sensor networks, Appeared in the Proceedings of MILCOM 2006. PDF