The system is divided into three parts to implement which are mobile user, posi-tioning server, and localization environment. The mobile user carries an Android smart phone, called HTC magic and equipped with a WiFi interface, and an IMU, called 3DM-GX1 made by MicroStrain company. The dimension of HTC magic is 113 mm × 55.56 mm × 13.65 mm, and its weight including battery is 118 g.
The version of the android operating system is 1.5. It supports 2G (GSM), 2.5G (GPRS), 3G (WCDMA) and WiFi 802.11 b/g networks. And, the GPS, G-sensor and electronic compass are embedded. 3DM-GX1 is composed of one triaxial ac-celerometer, one triaxial magnetometer, and one triaxial angular rate gyroscope.
The dimension of 3DM-GX1 is64 mm × 90 mm × 25 mm, and its weight is 75 grams. The IMU can provides 3D g-values in the range of ±5 g, 3D magnetic field in the range of±1.2 Gauss, and the rate of rotation in the range of 300◦ per
second. The sampling rate of these readings is350 Hz at most. In addition, it can provide its orientation in Euler angle (pitch, roll, yaw) but at most in the rate of 100 Hz. The 3DM-GX1 communicates with the handheld device via an RS-232 or RS-485 interface. The optional communication speeds are19.2, 38.4 and 115.2 kBaud.
The 3DM-GX1 is equipped by user with the belt as Figure 2.1. We gain the prompt mobility and behaviors of user by it‘s measurements. For instance, we adopt a proposed walking judged algorithm using accelerations to simulate a pe-dometer. Therefore, it can determine that the user is walking or stopping, and cumulate the strides. And, we use the yaw of euler angle directly to simulate the electronic compass to track the orientations of user.
The g-sensor communicates with the mobile device via a UART interface or Zigbee protocol. The mobile device has the ability to sense Wi-Fi radio signals sent from surrounding access points. In the localization, the user are positioned accordance with the positioning interval. We use a program to collect the Wi-Fi radio signals and IMU sensor measurements regularly. And then, the sensor measurements will be imported into the Behavior Predicting Module. When po-sitioning interval expired, a popo-sitioning pattern including RSS, strides, heading, and user’s behaviors will be packaged and send to the positioning server via the WLAN. Secondly, once the positioning server receives a positioning pattern, the positioning pattern will be inputted to the positioning algorithm. Upon the
lo-calization result is estimated, it will be forwarded to the mobile user for demon-strating. In the last part, the environment of localization is a multi-storey building deploying WLAN infrastructure. The floor layout maps of the building are con-structed in 2.5-dimensional (2.5D) description including hallways, rooms, walls, stairways and elevators. In the localization, the positioning server uses it as the positioning reference.
The performances are evaluated by the errors of localization in the building (the4th, 5th and 6th floors) of the Computer Science Department at the National Chiao Tung University, Engineering Building III. The dimensions of the floors are74.4 meters by 37.2 meters. The map of the positioning area is represented in the 2.5D format to assume that the feet of the user are constrained to lie on the floor during the stance phase. It is consisted of the floor plane maps including walls and rooms. And, they are connected by the stairways and elevators. In the positioning, this map will be imported into the localization system. There are153 fingerprints deployed randomly on the hallways and other public areas of the map.
Each of them is trained by 100 RSS patterns. The performances are contrasted with the nearest neighbor in signal space (NNSS), the mobility free particle filter (P F MCL) and our sensor enhanced particle filter (P F MM) by moving around in the positioning area. The number of particles of these particle filter approaches is500. The user is positioned with the period per second, and the average amount of access points which can be detected of each pattern is11. The errors are 7.33,
4.92 and 3.01 meters individually. The P F MM improves NNSS by 59% and P F MCL by 39%.
Chapter 6
Conclusions
In this paper, we developed an application level location tracking system. We proposed a sensor-enhanced particle filter scheme to assist the RF-based pattern matching localization system by user‘s immediate information. The user equips the IMU sensors (called G-Sensor) and electronic compass (called M-Sensor)with the belt. And, the sensor measurements will be converted to a mobility model.
Based on these models, we enhanced the sampling stage of the traditional particle filter. The particles will be propagated by horizontal displacement vector and vertical displacement vector to close the user. And, we added a particle filtering module between the particle sampling module and particle weighting module. The incredible particles will be filtered out by speed filtering and passing wall filtering with maps information. We examined the performance by errors of localization in the simulation and experiment. According to the statistics in the simulation,
our performances are outstanding in spite of any environment. And, our engine is robust regardless the increasing of sensor errors. In the experiment, we improved NNSS by 59% and P F MCL by 39%.
Bibliography
[1] M. Addlesee, R. Curwen, S. Hodges, J. Newman, P. Steggles, A. Ward, and A. Hopper. Implementing a Sentient Computing System. Computer, 34(8):50–56, 2001.
[2] P. Bahl and V. N. Padmanabhan. RADAR: An In-building RF-based User Location and Tracking System. In IEEE INFOCOM, volume 2, pages 775–
784, 2000.
[3] S. Beauregard, Widyawan, and M. Klepal. Indoor pdr performance enhance-ment using minimal map information and particle filters. In Position, Lo-cation and Navigation Symposium, 2008 IEEE/ION, pages 141–147, May 2008.
[4] E. Brassart, C. Pegard, and M. Mouaddib. Localization using infrared bea-cons. Robotica, 18(2):153–161, 2000.
[5] B. Ferris, D. Hhnel, and D. Fox. Gaussian processes for signal strength-based location estimation. In In Proc. of Robotics Science and Systems, 2006.
[6] D. Fox, J. Hightower, L. Liao, D. Schulz, and G. Borriello. Bayesian Filter-ing for Location Estimation. IEEE Pervasive ComputFilter-ing, 2(3):24–33, 2003.
[7] S.-P. Kuo and Y.-C. Tseng. A Scrambling Method for Fingerprint Position-ing Based on Temporal Diversity and Spatial Dependency. IEEE Trans. on Knowledge and Data Engineering, 20(5):678–684, 2008.
[8] D. Niculescu and B. Nath. Ad Hoc Positioning System (APS) Using AOA.
In IEEE INFOCOM, volume 3, pages 1734–1743, 2003.
[9] J. J. Pan, J. T. Kwok, Q. Yang, and Y. Chen. Multidimensional Vector Regression for Accurate and Low-Cost Location Estimation in Pervasive Computing. IEEE Trans. on Knowledge and Data Engineering, 18(9):1181–
1193, 2006.
[10] N. B. Priyantha, A. Chakraborty, and H. Balakrishnan. The Cricket Location-Support System. In ACM/IEEE MOBICOM, pages 32–43, 2000.
[11] A. Savvides, C.-C. Han, and M. B. Strivastava. Dynamic Fine-Grained Lo-calization in Ad-Hoc Networks of Sensors. In ACM/IEEE MOBICOM, pages 166–179, 2001.
[12] S. Thrun, D. Fox, and W. Burgard. Monte carlo localization with mixture proposal distribution. In Proc. of the National Conference on Artificial In-telligence, 2000.
[13] T.-C. Tsai, C.-L. Li, and T.-M. Lin. Reducing Calibration Effort for WLAN Location and Tracking System using Segment Technique. In IEEE Int’l Conf. on Sensor Networks, Ubiquitous, and Trustworthy Computing (SUTC), volume 2, pages 46–51, 2006.
[14] Widyawan, M. Klepal, and S. Beauregard. A novel backtracking particle filter for pattern matching indoor localization. In MELT ’08: Proceed-ings of the first ACM international workshop on Mobile entity localization and tracking in GPS-less environments, pages 79–84, New York, NY, USA, 2008. ACM.
[15] O. Woodman and R. Harle. Pedestrian localisation for indoor environments.
In UbiComp ’08: Proceedings of the 10th international conference on Ubiq-uitous computing, pages 114–123, New York, NY, USA, 2008. ACM.
[16] L.-W. Yeh, M.-S. Hsu, Y.-F. Lee, and Y.-C. Tseng. Indoor localization: Au-tomatically constructing today’s radio map by irobot and rfids. In IEEE Sensors Conference, 2009.
[17] J. Yin, Q. Yang, and L. M. Ni. Learning adaptive temporal radio maps for signal-strength-based location estimation. IEEE Transactions on Mobile Computing, 7(7):869–883, 2008.
[18] G. Zhou, T. He, S. Krishnamurthy, and J. A. Stankovic. Impact of Radio Irregularity on Wireless Sensor Networks. In ACM MobiSys, pages 125–
138. ACM Press New York, NY, USA, 2004.