Wireless Location Technologies Nobuo Kawaguchi Graduate School of Eng. Nagoya University 1 About me Nobuo Kawaguchi Associate Professor Dept. Engineering, Nagoya University Research Topics Wireless Location Systems ( http://locky.jp ) Scalable Adaptive Multicast (SAM), XCAST6 UbiCompEnvironment and Communication Middleware named cogma Mobile agent based system 2 Location Information and the Internet (1) 1
Content of my lectures(4days) Wireless Location Technologies (Today) Location Information Processing P2P Computing Plug and Play Technologies and Ubiquitous Computing 3 Today s Topic Wireless Location Technologies What is WiFi based Positioning History Characteristics WiFi Location Database Wireless AP positioning Using directional antenna 4 Location Information and the Internet (1) 2
Wireless Location Technologies Tech. for both Indoor / Outdoor is required Outdoors GPS Cell Phone Environment Wireless LAN PHS Ultra Sonic Indoors RFID UWB 1m 10m 100m Estimation Accuracy 5 WiFi Everywhere Wide spread of wireless LAN Restaurant Airport Home Company Station University RSSI BSSID, Latitude, Longitude WiFi location Wireless LAN has become popular infrastructure A large number of position estimation system using wireless LAN have been developed 6 Location Information and the Internet (1) 3
WiFi Location Technology Every WiFi Access Point(AP) has followings ESSID (String) BSSID 6 byte vendor assigned unique address (MAC) Construct a database with BSSID and Position One can estimate the position just receive a WiFi BSSID. Important Points A lot of WiFi APs are already exist in the Wild. One can easily add new APs by oneself. Can increase accuracy by adding APs. Each AP do not requires network access for positioning. 7 History of WLAN positioning Start around 2000 (MS RADER) now Technology 2000:RADAR (Microsoft) Products 2003:AirLocation 2000:Ekahau (Hitachi) Service/Activities 2003:PlaceLab (Intel) 2005/7:Locky.jp (Nagoya Univ) 2006/3:Loki (Skyhook Wireless) 2006/7:PlaceEngine (Sony CSL) 2007/9:Digial Camera (Sony) 2008/1:iPhone (Apple) 8 Location Information and the Internet (1) 4
Positioning methods using WLAN Existing methods are classified into following three types. Triangulation Lateration (RSSI or TDOA) RADAR (MS Research), WiPS (Kyusyu Univ. Japan) AirLocation (TDOA)(Hitachi) Angulation (AOA) Proximity GUIDE Project (Keith et al) Scene Analysis RADAR(MS Research), Ekahau (Ekahau corp.) Place Lab (Intel Research) 9 Positioning Access Point Triangulation Using more than 3 reference point s distance A (Lateration) a E c b B (Angulation) Using more than 2 reference points direction A α E B β C Reference Direction Reference Point Estimated Point10 Location Information and the Internet (1) 5
Positioning Access Point Proximity Consider the position of most powerful AP as a current position. Transmission range of Wireless LAN is about 100 meters in the open air Not good accuracy but simple. Scene Analysis Consider Difficult to construct learning data as target area is spread Inappropriate for our objective 11 Proximity This method consider communication area of AP as user s position. Terminal s location AP n (x n,y n ) AP 1 (x 1,y 1 ) AP 2 (x 2,y 2 ) Reference Point Communication Area of Reference Point 12 Location Information and the Internet (1) 6
Scene Analysis This method use pre-observation wireless information called radio map. Most system use Monte Carlo family method (ex. Bayesian filter, particle filter) and radio map. Lecture Room A Lecture Room B Lecture Room C Lecture Room D Lecture Room E 13 GPS vs WiFi Location From, Delivering Real-World Ubiquitous Location Systems, C. ACM 2006. 14 Location Information and the Internet (1) 7
WiFi Location: Characteristics WiFi characteristics Difference of WLAN Adapter/Antennas Difference on Orientation WiFi Signal Strength Distribtion WiFi Positioning Acitivities Locky.jp Project WiFi AP positioning 15 Difference on WLAN Adapter 16 Location Information and the Internet (1) 8
Distribution Difference Wireless LAN Adapter A Wireless LAN Adapter B 17 Difference on Orientation 0º Wireless LAN Card A Wireless LAN Card B Wireless LAN Card C 315º 45º 270º 90º 225º 135º 180º 18 Location Information and the Internet (1) 9
Distribution Pattern of WiFi RSSI 0.25 0.2 Probability Density 0.15 0.1 0.05 0-60 -55-50 -45-40 Received Signal Strength (dbm) 19 WiFi Location Database Required for wide area location system. Acquisition Method Accuracy Efficiency How to construct a large Database Locky.jp 20 Location Information and the Internet (1) 10
Acquisition Methods Variety of acquisition methods with different transportation Walking 5km/h Bicycle 15km/h Near to the buildings and houses Car 30km/h Influenced by traffic jams and regulations 21 Estimation Accuracy 100% 90% Cumulative rate 80% 70% 60% 50% 40% Learning Data ( Walking ) Learning Data ( Bicycle ) Learning Data ( Car ) 30% 20% ~10 ~20 ~30 ~40 Cumulative estimation accuracy of evaluative data by walking ~50 ~60 ~70 ~80 ~90 ~100 100~ Cannot Estimate Accuracy(m) The position estimation using the learning data acquired by walking can make estimates at less than 30m in 63% of the area The estimation data by car is worse by about 5% in all ranges Beyond 40 m, the bicycle data can estimate almost as well as that obtained by walking 22 Location Information and the Internet (1) 11
Accuracy of WiFi location 0.4 Car Bicycle Motor Bike Bicycle Car 0.3 Probability Density 0.2 0.1 0 0-20 21-40 41-60 61-80 81-100 101- unknown Accuracy (m) 23 Result: Bicycle is more suitable than Others Table: Position Estimation of Experimental data (Walking, Proximity) All data First-round data Strong signal APs (1 st round) (RSSI > - 90) No. of APs Accuracy (m) Coverage (%) No. of APs Accuracy (m) Coverage (%) No. of APs Accuracy (m) Coverage (%) Walking (5km/h) 247 24.3 86.8 214 31.3 72.2 155 30.1 68.1 Bicycle (15km/h) 269 26.7 87.8 209 23.2 74.6 178 23.5 73.2 Car (30km/h) Bicycle has better efficiency than others 183 29.6 83.3 122 26.0 68.2 49 32.9 53.8 24 Location Information and the Internet (1) 12
WarDriving using bycicle GPS WiFi Antenna / Card Stumbler Note PC Battery 25 Locky.jp Project 26 Location Information and the Internet (1) 13
Locky.jp Project WiFi Location Portal for Japan Data Collection of WiFi AP(BSSID) + Location Place Information Begins July 2005 Currently, we have collected 369,045 APs currently 27 Locky.jp Environment full of wireless LAN Users Information Service WiFi Location Locky.jp WiFi Location Database Effect on Service area Estimation accuracy Goals of Locky.jp Wide-area WiFi Location acquisition by collaboration with user Creation of position estimation system using WiFi Location database Producing Low-Cost positioning system for everywhere everybody Activity Research on positioning system Public relations of positioning system Creation of acquisition assistance tools 28 Location Information and the Internet (1) 14
Collecting AP s by Collaborators 200 over members collecting AP Mainly for Tokyo/Osaka/Nagoya - Ranking 29 Trend of No. of AP in Locky.jp Over 360,000 WiFi locations in Japan No. of AP 250000 200000 150000 WiFi User 120 100 80 No. of collaborator 60 100000 40 50000 20 0 Oct-05 Nov-05 Dec-05 Jan-06 Feb-06 Mar-06 Apr-06 May-06 Jun-06 Jul-06 Aug-06 Sep-06 0 Month 30 Location Information and the Internet (1) 15
Tokyo area ( over 50000 AP) 31 Tokyo APs plotted on GoogleEarth 32 Location Information and the Internet (1) 16
Advanced Internet Technology IV January 21st 2008 View from Tokyo-Castle 33 Nagoya Area, Over 40000APs 34 Location Information and the Internet (1) 17
Access Points in Nagoya City Area 1km Square 878 APs 1km Center of Nagoya JAPAN Access Point Total 878 8.8 unit / 100m 100m 35 APs in Residential Area 1km Residential area Height is restricted 278 APs in 1km square area Residential Area Access Point Total 278 3.5 unit /100m 100m 36 Location Information and the Internet (1) 18
Locky.jp upload page 37 Tools for Locky.jp Locky Stumbler KML converter Log KML (Google Earth) 38 Location Information and the Internet (1) 19
Positioning of WiFi AP Positioning of wireless access points is required. Construction of wireless LAN based positioning system Some systems use position of access point for positioning. (Ex. PlaceLab [Intel]/ Locky.jp uses position of access point) Detection of un-authorized access points for network management Communication interference Illegal connection to network 39 Directional Antenna Using distance property of wireless LAN Signal strength of WLAN varies according to a distance between sender and receiver Using directional antenna Signal strength of WLAN changes according to a change of receiver s antenna direction Access point s position estimation using asymmetric model based on distance and direction 40 Location Information and the Internet (1) 20
Utilizing Directional pattern Received Signal Strength Distribution of Directional Antenna (16m) 41 Example of Position Estimation -80dBm A B E -60dBm C -40dBm Existence Probability Peak Line Antenna s Position and Direction Estimated Point 42 Location Information and the Internet (1) 21
Estimation of Wireless Access Point (1/2) Set of Observation O = { o,..., } Each observation o = p, d, s consists of {User location p i, Direction of antenna d i, Signal strength s i } Candidate position C = c,..., Each c n is estimation candidate point i 1 Access point s location is assumed in candidate C o m ( ) { } 1 i i c n i 43 Estimation of Wireless Access Point (2/2) When a user get observation To use signal strength model, the system calculates posterior probability of signal strength. Each observation is considered independent and to use Bayesian inference, the posterior probability of each candidate point is ( o,..., o ) P c j 1 m ( s p, d c ) P, i = n i ( p, d, c ) m P s k = 1 k k k j m { P( sk pk, dk, cl ) } l= 1 k = 1 i j ( p, d s ) o =, j =1,...,n This posterior probability is existing probability of wireless access point. 44 i i i i j =1,...,n Location Information and the Internet (1) 22
Advantage of the Method -60dBm -40dBm -62dBm To use asymmetric model, It is possible to acuminate existence probability using only direction change State 1 State 2 State 3 total result Existence Probability Peak Line Antenna s Position and Direction Estimated Point 45 Access Point data Acquisition WSA : Wireless Search Assistant Laptop GPS Omni Directional Antenna Directional Antenna Direction Sensor Segway 46 Location Information and the Internet (1) 23
Screenshot of WSA Software Probability of estimation result Signal Strength Graph Access Point BSSID List 47 48 Location Information and the Internet (1) 24
Indoor Positioning Outdoor Using GPS Indoor Using Deadreknoing to measure wheel moving. Fiber Sensor Amp Units Fiber Sensor Fiber Units Fiber sensor for measuring of wheel s action. 49 Utilization of WiFi Location System Location systems are widely spread But not for WiFi Location System GPS is primary location source A lot of GPS application are public NMEA format We need integrate WLS with GPS 50 Location Information and the Internet (1) 25
Summary Wireless Location Technology With WiFi, one can easily estimate one s position Collection of WiFi Location Database Locky.jp WiFi AP position estimation Using directional antenna 51 Exercise Wireless Location Technology is a promised technology for positioning in the urban area. What is a key factor of WiFi positioning technology for deployment in your country? What kind of services/applications can be implement using WiFi location system? such as Web services Embedded devices Mobile devices Wearable devices (Answer more than 3 services/applications) 52 Location Information and the Internet (1) 26