Nerimi: WiFi-based Subway Navigation System. Inje Lee1, Giwan Yoon2, and Dongsoo Han3. Department of Computer Science1,3 and Electrical Engineering2.
Nerimi: WiFi-based Subway Navigation System Inje Lee1, Giwan Yoon2, and Dongsoo Han3 Department of Computer Science1,3 and Electrical Engineering2 Korea Advanced Institute of Science and Technology (KAIST) Yuseong-gu, Daejeon, 305-701, Korea 1,3 2 E-mail: {lightmail,dshan}@kaist.ac.kr, {gwyoon}@ee.kaist.ac.kr Abstract — In this paper, we propose a WiFi-based stop notification system. The users of the system inform their destination stops to the system, then the system provides subway navigation services to the users in real-time. To construct the system, we have collected WiFi fingerprints at 15 lines for 536 Seoul subway stops. Also, we have implemented a WiFi-based subway stop notification system for Seoul subways in Korea. We have tested the accuracy of our system, Nerimi, at Seoul subways. Among the 536 stops, the system could detect the subway stops with over 90% accuracy. Thus, if we can capture the WiFi signals at every stop of the subways, using the WiFi signals will be highly eligible for building a subway navigation system. Index Terms — Subway, Wi-Fi, stop notification, navigation.
I. INTRODUCTION The subways are one of the most representative public transportations in large cities like Seoul, New York, Tokyo, London, Paris, Peking, etc. Every day, millions of people get on the subways to commute or travel. One common thing that subway passengers should be aware of will be not to miss their destination stops. In this reason, each passenger needs to check their current and next stops from time to time. When the subways are crowded with passengers and the trains will be in a noisy situation, the checking may not be easy. This may make the passengers feel uncomfortable and uneasy whenever they ride on the subways. Thus, it would be desirable if we can have a subway navigation service, i.e. the current and next subway stops are displayed on our hand-held devices such as cellular phones or smartphones in real-time. In this paper, we propose a WiFi-based stop notification system. Users of the system inform their destination stops to the system, then the system provides subway navigation services to the users in real-time. It finds a path from the current stop to the destination stop, and displays the traces of the train on the path. When the train approaches the destination stops, the system notifies it to the users such that they can prepare in advance to get off the train. The remainder of this paper is organized as follows. In Section II, we survey the related work of the subway navigation systems. In Section III, we explain the system overview. In Section IV, we introduce our mobile application, named the ‘Nerimi’. In Section V, we present the experimental results of the system. Finally, Section VI concludes this paper.
II. RELATED WORK The RADAR [1] is a system suitably used for locating and tracking any users inside a building. Its basic localization method is triangulation. The system must recognize the locations of the access points (APs) for using the triangulation method. Also, the system uses the empirical method. To use the empirical method, the system must have recorded the information about the signal strength (SS) of the APs. Thus, the data collection is positively necessary in the empirical method. It calculates a user’s current location using the Euclidean distance measure. This method performs quite well. The system described in [2] is a subway information system based on the WiFi localization technology. They developed the software named ‘Subway Stumbler’ in order to locate the APs themselves and also to record the WiFi environments. Also, they developed the subway information system for Nagoya city as a mobile application on iPod Touch. III. SYSTEM OVERVIEW There are two phases in WiFi-based subway stop notification system; off-line phase and on-line phase. In this section, we explain these two phases and the basic principle of positioning. A. Off-line Phase Off-line phase is the important part of the system. We can say that off-line phase is data collecting phase in other words. In the off-line phase, we collect WiFi fingerprints of every stop of Seoul subways. So, we developed smartphone application for collecting WiFi fingerprints. The distribution of the average number of AP at each subway line is shown in Fig 1. After collecting the WiFi fingerprints, we construct a subway stop-WiFi fingerprint association table for the recognition of subway stops in on-line phase using the association table generator. The association table generator removes both APs installed in trains and APs appeared temporarily because these APs can interfere with detecting current station. The subway stop-WiFi fingerprint association information enables the system to identify the corresponding subway stops for a given WiFi fingerprint.
978-1-4577-0963-0/11/$26.00 ©2011 IEEE
Fig. 2.
Android application of subway navigation system.
V. EXPERIMENTAL RESULTS Fig. 1. The distribution of the average number of AP at each subway line.
B. On-line Phase In on-line phase, user’s smartphone captures a WiFi fingerprint and detects the subway stops by comparing the captured WiFi fingerprint with the collected WiFi fingerprints in off-line phase. We describe the principle of positioning in the next subsection. C. Basic Principle of Positioning The system finds user’s current stop or location using a captured fingerprint in the on-line phase. After capturing a WiFi fingerprint, the system extracts the associated stop of each AP in the WiFi fingerprint from the association table which was constructed in the off-line phase. Once the list of the stops is prepared, the system decides the stop with the highest appearance frequencies in the list as the current stop of the user. Note that duplications of the stops are allowed in building the list. IV. SMARTPHONE APPLICATION Actually, we implemented a WiFi-based subway stop notification system for Seoul subways in Korea. The WiFi fingerprints were collected at 15 lines for 536 Seoul subway stops, and the subway stop-WiFi fingerprint association tables were constructed. Although there were some fluctuations in the numbers of the access points (APs) in the captured WiFi fingerprints of each subway stop, around 20 access points were detected in the WiFi fingerprints. This indirectly indicates that the WiFi signal is eligible for the detection of the subway stops at least in Seoul subway. The system, named ‘Nerimi’ was developed as an App for smart phones running “Google Android OS”. One very unique feature of the Nerimi is its high capability of making use of the user’s feedback obtained after its navigation. Based on the user’s feedback, it can monitor the situations of every stop of Seoul subways and moreover it even updates the subway stopWiFi fingerprint association tables. When we consider that the WiFi environments, ever changing, the adaptation capability of the Nerimi will drastically reduce the maintenance cost of the Nerimi.
We tested the accuracy of the Nerimi at Seoul subway. Among the 536 stops, the system detected the subway stops with over 90% accuracy. More specifically, the system couldn’t detect only 1 stop at line 7 and line 9. So, the accuracy of line 7 and line 9 is 97.6% and 95.8%, respectively. However, at line 2, the system couldn’t detect 4 stops and the accuracy of line 2 is 92.1%. Most of the incorrectly detected subway stops were revealed to have a relatively small number (2 or 3) of APs or large number (more than 150) of APs. For the stops with a relatively small number of the APs, we can easily improve the accuracy by installing some additional APs. For the stops with a relatively large number of the APs, we need to develop the methods to select the critical APs among the APs. The location of the APs and strength of the WiFi signals will be two primary criteria for the selection. In addition, we can use the time tables of the subways to compensate for the incorrectly detected subway stops. Although the accuracy of the detection of stops is severely dependent on the WiFi environments, the WiFi-based subway navigation system was revealed to show the better performance against the other GPS, 3G, or time table based approaches. VI. CONCLUSION In conclusion, if we can capture WiFi signals at every stop of subway, using WiFi signal is eligible for building subway navigation system. And how to cope with the changes of WiFi environments will be the key to the success of WiFi-based subway navigation system. REFERENCES [1] P. Bahl and V.N. Padmanabhan, "RADAR: An In-Building RFbased User Location and Tracking System," Proc. IEEE INFOCOM2000 Conf., pp. 775-784, 2000. [2] Nobuo Kawaguchi et al., "Underground Positioning: Subway Information System Using WiFi Location Technology," mdm, pp.371-372, 2009 Tenth International Conference on Mobile Data Management: Systems, Services and Middleware, 2009. [3] T. King, T. Haenselmann, and W. Effelsberg, "On-Demand Fingerprint Selection for 802.11-based Positioning Systems," World of Wireless, Mobile and Multimedia Networks, pp.1-8, 23-26, 2008.