Location in Pervasive Computing Shwetak N. Patel University of Washington More info: shwetak.com Special thanks to Alex Varshavsky and Gaetano Borriello for their contribution to this content design: use: build: university of washington ubicomp lab university of washington Computer Science & Engineering Electrical Engineering Location A form of contextual information Person’s physical position Location of a device Device is a proxy of a person’s location Used to help derive activity information 2 Location Well studied topic (3,000+ PhD theses??) Application dependent Research areas Technology Algorithms and data analysis Visualization Evaluation 3 Location Tracking 4 Representing Location Information Absolute Relative Geographic coordinates (Lat: 33.98333, Long: -86.22444) 1 block north of the main building Symbolic High-level description Home, bedroom, work 5 No one size fits all! Accurate Low-cost Easy-to-deploy Ubiquitous Application needs determine technology 6 Consider for example… Motion capture Car navigation system Finding a lost object Weather information Printing a document 7 Others aspects of location information Indoor vs. outdoor Absolute vs. relative Representation of uncertainty Privacy model 8 Lots of technologies! GPS WiFi Beacons VHF Omni Ranging Ultrasound Ad hoc signal strength Floor pressure Laser range-finding Stereo camera Array microphone Ultrasonic time of flight Infrared proximity E-911 Physical contact 9 Some outdoor applications E-911 Bus view Car Navigation Child tracking 10 Some indoor applications Elder care 11 Outline Defining location Methods for determining location Ex. Triangulation, trilateration, etc. Systems Challenges and Design Decisions Considerations Approaches for determining location Localization algorithms Proximity Lateration Hyperbolic Lateration Angulation Fingerprinting Distance estimates Time of Flight Signal Strength Attenuation 13 Proximity Simplest positioning technique Closeness to a reference point Based on loudness, physical contact, etc 14 Lateration Measure distance between device and reference points 3 reference points needed for 2D and 4 for 3D 15 Hyperbolic Lateration Time difference of arrival (TDOA) Signal restricted to a hyperbola 16 Angulation Angle of the signals Directional antennas are usually needed 17 Determining Distance Time of flight Signal strength Speed of light or sound Known drop off characteristics 1/r^2-1/r^6 Problems: Multipath 18 Fingerprinting Mapping solution Address problems with multipath Better than modeling complex RF propagation pattern 19 Fingerprinting SSID (Name) BSSID (MAC address) Signal Strength (RSSI) linksys 00:0F:66:2A:61:00 18 starbucks 00:0F:C8:00:15:13 15 newark wifi 00:06:25:98:7A:0C 23 20 Fingerprinting Easier than modeling Requires a dense site survey Usually better for symbolic localization Spatial differentiability Temporal stability 21 Reporting Error Precision vs. Accuracy 22 Reporting Error Cumulative distribution function (CDF) Absolute location tracking systems CDF of Localization error 1 0.9 Percentage 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0 0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 Error (m) Accuracy value and/or confusion matrix Symbolic systems 23 Location Systems Distinguished by their underlying signaling system IR, RF, Ultrasonic, Vision, Audio, etc 24 GPS Use 24 satellites TDOA Hyperbolic lateration Civilian GPS L1 (1575 MHZ) 10 meter acc. 25 Active Badge IR-based Proximity 26 Active Bat Ultrasonic Time of flight of ultrasonic pings 3cm resolution 27 Cricket Similar to Active Bat Decentralized compared to Active Bat 28 Cricket vs Active Bat Privacy preserving Scaling Client costs Active Bat Cricket 29 Ubisense Ultra-wideband (UWB) 6-8 GHz Time difference of arrival (TDOA) and Angle of arrival (AOA) 15-30 cm 30 RADAR WiFi-based localization Reduce need for new infrastructure Fingerprinting 31 Place Lab “Beacons in the wild” WiFi, Bluetooth, GSM, etc Community authored databases API for a variety of platforms RightSPOT (MSR) – FM towers 32 ROSUM Digital TV signals Much stronger signals, well-placed cell towers, coverage over large range Requires TV signal receiver in each device Trilateration, 10-20m (worse where there are fewer transmitters) 33 Comparing Approaches Many types of solutions (both research and commercial) Install custom beacons in the environment Ultra-wideband (Ubisense), Ultrasonic (MIT Cricket, Active Bat), Bluetooth Use existing infrastructure GSM (Intel, Toronto), WiFi (RADAR, Ekahau, Place Lab), FM (MSR) 34 Limitations Beacon-based solutions Requires the deployment of many devices (typically at least one per room) Maintenance Using existing infrastructure WiFi and GSM Not always dense near some residential areas Little control over infrastructure (especially GSM) 35 Beacon-based localization 36 Wifi localization (ex. Ekahau) 37 GSM localization Tower IDs and signals change Coverage? over time! 38 PowerLine Positioning Indoor localization using standard household power lines 39 Signal Detection A tag detects these signals radiating from the electrical wiring at a given location 40 Signal Map 1st Floor 2nd Floor 41 2 d ( x, y ) ( ( xi yi ) 2 ) i 1 Example 42 2 d ( x, y ) ( ( xi yi ) 2 ) i 1 Passive location tracking No need to carry a tag or device Hard to determine the identity of the person Requires more infrastructure (potentially) 43 2 d ( x, y ) ( ( xi yi ) 2 ) i 1 Active Floor Instrument floor with load sensors Footsteps and gait detection 44 2 d ( x, y ) ( ( xi yi ) 2 ) i 1 Motion Detectors Low-cost Low-resolution 45 2 d ( x, y ) ( ( xi yi ) 2 ) i 1 Computer Vision Leverage existing infrastructure Requires significant communication and computational resources CCTV 46 2 d ( x, y ) ( ( xi yi ) 2 ) i 1 Other systems? Inertial sensing HVACs Ambient RF etc. 47 Considerations Location type Resolution/Accuracy Infrastructure requirements Data storage (local or central) System type (active, passive) Signaling system 48