Passive localization using array sensor with support vector machine

Abstract

A new method for passive localization using an array sensor system based on spatial smoothing processing (SSP) with support vector machine (SVM) is proposed. The array sensor uses only one array antenna as the receiver to observe the signal subspace spanned by eigenvector through array signal processing. The signal subspace represents the radio wave propagation of interest. Based on the eigenvector, it can detect and classify simple human activities: entering a room, standing, and moving. The advantages of the system are as follows: it guarantees privacy of users; it eliminates installation difficulties; it also offers a wide detection range. Although the conventional method can detect simple human activities, it cannot determine the position of the human being in detail. The proposed method uses multiple transmitters emitting different frequency signals to extend the dimension of the signal subspace. In addition, we separate coherent signals by using the SSP to obtain more features of radio wave propagation than the number of transmitters. The features are used as inputs to SVM to localize human position. The experimental results show that the proposed method improves the localization accuracy and the root mean square error (RMSE) compared to the previous method.