非校准网络中超宽带层析成像-Beck_UWB_RTI_uncalibrated_networks_FINALpublished

Abstract— This paper considers the problem of tomographic area mapping using radio frequency measurements gathered by a network of mobile nodes. Termed radio tomographic imaging, the technique has shown potential for object tracking, imaging static obstacles, and even through-wall imaging. Our approach addresses substantial issues for the practical implementation of such a system, namely, the mitigation of multipath signal effects and the characterization of a large number of uncalibrated network links. We propose a system that utilizes ultrawideband direct path signal strength measurements as a means of reducing the effects of the multipath fading. Furthermore, we address the estimation of unknown path loss and link bias parameters online through the framework of a linear mixed effects model. This permits the estimation of a static area map without a prohibitive calibration of these parameters prior to deployment, which is crucial in a network that may contain hundreds of links. Our model is posed as a convex optimization problem using the elastic net for regularization. Bayesian performance bounds are derived and our method shows positive results in simulation. We then demonstrate the efficacy of our solution on real tomographic data gathered from our cognitive spectrum operations testbed.

下载论文：非校准网络中超宽带层析成像-Beck_UWB_RTI_uncalibrated_networks_FINALpublished.pdf