Using tapped delay line to improve the precision of an ensemble of classifiers in device-free localization

Abstract

In this paper, we adopt a tapped delay line (TDL) equalizer as the arbiter of ensemble learning for device-free localization over IEEE 802.11 wireless local area network (WLAN). The proposed model is a two-stage decision process. While an input signal along with a delay line is given, a trained support vector machine (SVM) and Bayesian classifier performs the first stage prediction. Then, the second stage decision selects the most frequent one among the intermediate outputs of stage one as the final output. Experimental results show the proposed method can not only to be as the arbiter of ensemble learning, but also significantly improve the precision to achieve 99.02%.