A positioning method based on RSSI and power spectrum waveform distinction

In this paper, we propose a positioning method based on the dual-complex fingerprint, which consists of the Received Signal Strength Indication (RSSI) and the Power Spectrum Waveform (PSW), including three stages. First, generate fingerprint library by data collected offline. For each reference point, RSSI and PSW are both stored in the library. Then make pre-positioning by RSSI fingerprint and the location of reference points. These points will be selected twice to remove the single points away from the others. Final positions are estimated by taking PSW Distinction (PSWD) and RSSI into consideration. In addition, we introduce an idea of evaluating PSWD by the Kullback-Leibler Distance (KLD). The MATLAB simulation results show that, comparing to other algorithms such as KNN and WKNN, the proposed method leads to lower number of observable misestimated points, and approximately 5% improvement in cumulative distribution function (CDF) of position error within 1.3 m.