5946 IEEE SENSORS JOURNAL, VOL. 15, NO. 10, OCTOBER 2015 Data Fusion of Time Stamps and Transmitted Data for Unsynchronized Beacons Alexander Traub-Ens, Joan Bordoy, Johannes Wendeberg, Leonhard M. Reindl, Member, IEEE, and Christian Schindelhauer Abstract— Unsynchronized localization systems based on the measurement of time (difference) of arrival require reliable time stamps of the received signal. Noise, frequency shifts, and echoes disturb the signal and induce measurement errors of the time stamp, which leads to localization errors. Furthermore, the line of sight (LOS) signal has to be distinguished from the echoes to avoid false signal tracking. The proposed method combines the information of an ultrasound transmission with the measured time stamp and estimates the identifier. In our approach, the ultrasound transmission system uses phase-shift keying to modulate the signal. The received symbols and the time stamps are tracked and fused by the Kalman filter to increase the signal-to-noise ratio of the fused symbols and improve the validity of the decoding. Hence, the bias of the received symbols is tracked and the tracking allows to distinguish between the LOS signal and the echoes. As a result, the data fusion reduces the packet error rate from 70% at a distance of 21 m to 4.5%. Moreover, the median error of the localization is reduced from 7 to 4.6 cm. Index Terms— Ultrasound, localization, communication, data fusion, beacon, time stamp, TDOA, unsynchronized. I. I NTRODUCTION B EACONS are often used in emergency or navigation systems to locate the position. They are installed on known positions and transmits pulses in constant intervals. The simplest beacon is a lighthouse, which provides the seaman the important information about the mainland or dangerous regions. The light from the lighthouses has a characteristic blinking scheme which provides information about the posi- tion of the beacon [1] and thus, the navigator can estimates his own position. Modern radio frequency based beacons transmits an unique identifier to distinguish between different nodes and improve the localization accuracy [2]. The identification could be the media access control (MAC) address in a wireless local area network (WLAN) [3] or a special signal sequence to Manuscript received March 25, 2015; revised June 23, 2015; accepted June 25, 2015. Date of publication July 1, 2015; date of current version August 19, 2015. This work was supported in part by the Bundesministerium für Bildung und Forschung through the German Federal Ministry of Education and Research under Grant 16SV5988 and in part by the Spitzencluster MicroTec Suedwest. The associate editor coordinating the review of this paper and approving it for publication was Dr. Amitava Chatterjee. The authors are with the University of Freiburg, Freiburg im Breisgau 79110, Germany (e-mail: alexander.ens@imtek.uni-freiburg.de; bordoy@informatik.uni-freiburg.de; wendeber@informatik.uni-freiburg.de; reindl@imtek.de; schindel@informatik.uni-freiburg.de). Color versions of one or more of the figures in this paper are available online at http://ieeexplore.ieee.org. Digital Object Identifier 10.1109/JSEN.2015.2452227 distinguish between the beacons [4], [5]. Measuring the signal strength [3] and having a map of the beacons, the position can be calculated. If the beacon provides the time information (used in global positioning systems) [6] then the position can be estimated by triangulation algorithms. Ultrasound based localization systems without synchronization use modulation techniques to provide information about the sender [7]. Therefore, the data is mapped to symbols and the ultrasound signal is modulated by phase shift keying (PSK). The receivers demodulate the signal, decode the data and are able to distinguish between the senders. The time difference of arrival (TDOA) is then used to calculate the position [8]. Modern algorithms are able to detect the time difference between the senders and can perform a self calibration [9]. However, echoes, noise and distortions cause transmission errors and induce false decoding of the signal. Especially walls are good reflectors for sound and induce echoes [10]. Decoding the echo as the correct signal leads to outliers in the position estimation and hence, an imprecise localization with high standard deviation. Thus, localization systems without synchronization are sensitive to echoes and rely on the correct ID decoding. Consequently, the modulated signal requires additional information to improve the signal-to-noise ratio (SNR) and the validity of the decoding at the receiver and to distinguish between the echoes and the line of sight (LOS) signal. Saad et al. [11] showed an unsynchronized localization system, which uses the angle of arrival to increase the validity of the signal and improve the localization accuracy. However, the receiver requires multiple transducers and the accuracy depends on the alignment of the receiver. As a result, this approach uses the transmission intervals (blinking scheme) of the senders as additional information about the signal origin and hence, the sender identification. Data fusion gain more attention in the research due to the benefit of higher measuring accuracy by combining indepen- dent data sources. Luo et al. [12] showed a brief overview of the possible data fusion algorithms and methods and compare the complexity. As a result, the Kalman filter [13] has low complexity and high performance for linear data fusion applications. An extensive research of multi sensor fusion with Bayes filter and Kalman filter is described by Heizmann et al. [14]. Therefore, the same sensor type is placed at different positions to increase the entropy and the data is fused with a Kalman filter (e.g. increasing the contrast of an image recognition system). 1530-437X © 2015 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.