환경적인 바이어스를 이용한 효과적인 실내 측위 알고리즘

Abstract

In this thesis, the new effective localization algorithms using environmental bias in indoor environments are proposed. Wireless localization systems cannot estimate an accurate location because of the errors induced by clock offset and non-line-of-sight (NLOS) environments. The detection error caused by the clock offset of devices greatly influences timing synchronization, which is required to estimate the distance between a transmitter and a receiver. In indoor environments, the NLOS errors block the signal propagation between two wireless devices. In particular, the NLOS errors significantly affect the distance estimation between two wireless devices. In this thesis, effective time offset estimation and indoor localization algorithms that minimize these effects of the errors are proposed. First, an effective time offset estimation algorithm for accurate timing synchronization is proposed. The effective time offset estimation algorithm is applied to a generalized chirp signal sequence that consists of up and down chirp signals in a real-time locating system (RTLS). The sub-autocorrelation functions use the conjugate property between up and down chirp signals. The proposed time offset estimation algorithm based on sub-autocorrelation functions is used to estimate accurate timing synchronization. Secondly, an effective indoor localization algorithm that minimizes the effect of NLOS errors in NLOS environments is proposed. In this localization algorithm, a base station (BS) selection method is used to select three measured distances that contain fewer NLOS errors between the BSs and a mobile station (MS) in three or more BSs for the localization system. The proposed localization algorithm with the BS selection method is utilized to accurately estimate the location of the wireless device. Various simulations and experiments show that the proposed localization algorithms are more accurate than the conventional algorithms. An effective time offset estimation algorithm based on sub-autocorrelation functions eliminates the timing metric plateau, which causes uncertainty in arrival time of signals and accurately estimates the timing synchronization. The proposed time offset estimation algorithm shows a much smaller root mean square error (RMSE) than conventional algorithms. In NLOS environments, the proposed localization algorithm with BS selection shows higher localization accuracy than conventional localization algorithms. In real environments, the average estimation error of the proposed localization algorithm is reduced by about 50% compared to that of conventional algorithms.