Localization

2

Location

• Source of wireless signals–Wireless emitter

• Location of a mobile device– Some devices, e.g., cell phones, are a proxy of a

person’s location

• Used to help derive the context and activity information– Location based services

– Privacy problems

3

Location

• Well studied topic (3,000+ PhD theses??)

• Application dependent

• Research areas– Technology

– Algorithms and data analysis

– Visualization

– Evaluation

4

Representing Location Information

• Absolute– Geographic coordinates (Lat: 33.98333, Long: -86.22444)

• Relative– 1 block north of the main building

• Symbolic– High-level description

– Home, bedroom, work

5

Some outdoor applications

Car Navigation Child tracking

Bus view

E-911

6

Some indoor applications

Elder care

7

No one size fits all!

• Accurate

• Low-cost

• Easy-to-deploy

• Ubiquitous

• Application needs determine technology

Lots of technologies!

8

Ultrasonic time of flight

E-911

Stereo camera

Ad hoc signal strength

GPS

Physical contact

WiFi Beacons

Infrared proximity

Laser range-finding

VHF Omni Ranging

Array microphone

Floor pressureUltrasound

Wireless Technologies for Localization

Name Effective Range Pros Cons

GSM 35km Long range Very low accuracy

LTE 30km-100km

Wi-Fi 50m-100m Readily available; Medium range

Low accuracy

Ultra Wideband 70m High accuracy High cost

Bluetooth 10m Readily Available; Medium accuracy

Short range

Ultrasound 6-9m High accuracy High cost, not scalable

RFID & IR 1m Moderate to high accuracy

Short range, Line-Of-Sight (LOS)

NFC <4cm High accuracy Very short range

9

Localization Techniques

• Range-based algorithms

• Range-free algorithms

• Fingerprinting

10

Range Based Algorithms

• Rely on the distance (angle) measurement between nodes to estimate the target location

• Approaches– Proximity

– Lateration

– Hyperbolic Lateration

– Angulation

• Distance estimates– Time of Flight

– Signal Strength Attenuation

11

Approach: Proximity

• Simplest positioning technique

• Closeness to a reference point

• Based on loudness, physical contact, etc

12

Approach: Lateration

• Measure distance between device and reference points

• 3 reference points needed for 2D and 4 for 3D

13

Approach: Hyperbolic Lateration

• Time difference of arrival (TDOA)

• Signal restricted to a hyperbola

14

Approach: Angulation

• Angle of the signals

• Directional antennas are usually needed

15

Distance Estimation

• Multiple the radio signal velocity and the travel time– Time of arrival (TOA)

– Time difference of arrival (TDOA)

• Compute the attenuation of the emitted signal strength– RSSI

• Problem: Multipath fading

16

Distance Estimation: TOA

• Distance– Based on one signal’s travelling time from target

to measuring unit

– d = vradio * tradio

• Requirement– Transmitters and receivers should be precisely

synchronized– Timestamp must be labeled in the transmitting

signal

17

Distance Estimation: TDOA

• Distance– Based on time signals’ travelling time from target to

measuring unit

– d = vradio * vsound * (tradio- tsound) / (vradio – vsound))

• Requirement– Transmitters and receivers should be precisely

synchronized

– Timestamp must be labeled in the transmitting signal

– Line-Of-Sight (LOS) channel

18

Distance Estimation: RSSI

• Distance– Based on radio propagation model

–

• Requirement– Path loss exponent η for a given environment is

known

19

Range Free Algorithms

• Rely on target object’s proximity to anchor beacons with known positions– Neighborhood: single/multiple closest BS

– Hop-count: anchor broadcast beacons containing its location and hop-count

– Area estimation:

20

Fingerprinting

• Mapping solution

• Address problems with multipath

• Better than modeling complex RF propagation pattern

21

Fingerprinting: Steps

• Step1– Use war-driving to build up location fingerprints (i.e.

location coordinates + respective RSSI from nearby base stations)

• Step2– Match online measurements with the closest a priori

location fingerprints

22

Fingerprinting: Example

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

23

Fingerprinting: Features

• Easier than modeling

• Requires a dense site survey

• Usually better for symbolic localization

• Spatial differentiability

• Temporal stability

24

Summary of Localization Techniques

Measurement Scheme

Accuracy Special Requirement

Range-based TOA Moderate Synchronization, dense beacons

TDOA High Synchronization, LOS, dense beacons

AOA High Directional antenna

RSSI Moderate No

Range-free Neighborhood Low No

Area estimation Moderate Dense Beacons

Hop count Moderate Dense Beacons

Fingerprinting RSSI High No

25

Localization Systems

• Distinguished by their underlying signaling system– IR, RF, Ultrasonic, Vision, Audio, etc [13]

26

GPS

• Use 24 satellites

• TDOA

• Hyperbolic lateration

• Civilian GPS– L1 (1575 MHZ)• 10 meter acc.

27

Active Bat

• Ultrasonic

• Time of flight of ultrasonic pings

• 3cm resolution

28

Cricket

• Similar to Active Bat

• Decentralized compared to Active Bat

29

Cricket vs Active Bat

• Privacy preserving

• Scaling

• Client costs

Active Bat Cricket

30

31

RADAR• WiFi-based localization

• Reduce need for new infrastructure

• Fingerprinting

31

32

Place Lab• “Beacons in the wild”–WiFi, Bluetooth, GSM, etc

• Community authored databases

• API for a variety of platforms

• RightSPOT (MSR) – FM towers

33

Computer Vision• Leverage existing infrastructure

• Requires significant communication and computational resources

• CCTV

Performance Metrics

• Accuracy – Mean distance error (RMSE)

• Precision– Variation in accuracy over many trials (CDF of RMSE)

• Robustness– Performance when signals are incomplete

• Cost– Hardware, energy

34

Performance Evaluation

System/Solution

Wireless Technologies

Accuracy Precision Robustness Cost

Active Badge [1]

IR 3cm 90% Poor Low

Cricket[2] Ultrasound 5cm 90% Poor Medium

BeepBeep [3] Sound 4cm 95% Poor High

Virtual Compass [4]

Bluetooth+ WiFi RSSI

3.19m 90% Good Medium

APIT [5] WiFi RSSI 0.4 * radio range

Medium Low

DV-Hop [6] WiFi RSSI 3.5m 90% Medium Low

35

Performance Evaluation

System/Solution

Wireless Technologies

Accuracy Precision Robustness Cost

Centroid [7] WiFi RSSI 3.5m 90% Good Low

Amorphous [8] WiFi RSSI 0.2* radio range

Medium Low

RADAR [9] WiFi RSSI 5.9m 95% Good Low

Horus [10] Bluetooth+ WiFi RSSI

2.1m 90% Good Low

SurroudSense [11]

WiFi RSSI 90% N/A Good High

Ekahau [12] WiFi RSSI 2m 50% Good Low

36

E-V Loc: Goal

• Find a specific person’s accurate location based on his electronic identifier and visual image

- Publication:Boying Zhang, Jin Teng, Junda Zhu, Xinfeng Li, Dong Xuan, and Yuan F. Zheng, EV-Loc: Integrating

Electronic and Visual Signals for Accurate Localization, to appear in ACM MobiHoc’12.

37

E-V Loc: Problem Formulation

• Input: a target object’s electronic identifier EID*, a set (in a short time span) of E Frames with clear EIDs and the corresponding V Frames with possibly vague VIDs

• Output: the target object’s accurate position together with its visual appearance VID*

38

E-V Loc: Work Flow

39

Need more signal samples?

E-V Loc: Nature of Our Solution

• E-V matching – Uses electronic and visual signals as target object’s

location descriptors in E frames and V frames

– Matches the corresponding E and V location descriptors using Hungarian algorithm

40

41

E-V Loc:Localizing with Distinct VIDs• Best match problem between EIDs and VDs

EIDs VIDs

42

E-V Loc: Incremental Hungarian algorithm• Find the best match between the EIDs and VIDs in

each pair of E and V frame• Iteratively perform the matching until a threshold is

satisfied• The threshold is derived based on the variance model

of EIDs and VIDs

43

E-V Loc:Localizing with Indistinct VIDs• Multi-dimensional best match problem

Between EIDs and VIDs Among VIDs

44

E-V Loc: Two-dimensional Hungarian Algorithm• Finding correspondence between different VIDs in

neighboring frames• Based on the correspondence, generating a consistent

set of VIDs in all frames• Using incremental Hungarian algorithm to perform

the match

References

1. Roy Want, Andy Hopper, Veronica Falcao, and Jonathan Gibbons. The active badge location system. ACM Transactions on Information Systems, 10(1):91–102, 1992.

2. N. B. Priyantha, A. Chakraborty, and H. Balakrishnan. The cricket locationsupportsystem. In Proc. of ACM MobiCom, pages 32–43, 2000.

3. C. Peng, G. Shen, Y. Zhang, Y. Li, and K. Tan. BeepBeep: ahigh accuracy acoustic ranging system using COTS mobiledevices. In ACM SenSys, pages 1–14, 2007.

4. N. Banerjee, S. Agarwal, P. Bahl, R. Chandra, A. Wolman, and M. Corner.Virtual compass: relative positioning to sense mobile social interactions. InPervasive, 2010.

5. T. He, C. Huang, B. Blum, J. Stankovic, and T. Abdelzaher. Range-free localizationschemes for large scale sensor networks. In Proc. of ACM MobiCom,pages 81–95, 2003.

References

6. D. Niculescu and B. Nath. DV based positioning in ad hoc networks. Journalof Telecom. Systems, 2003.

7. N. Bulusu, J. Heidemann, and D. Estrin. Gps-less low cost outdoor localizationfor very small devices. IEEE Personal Communications Magazine, 7(5):28–34,October 2000.

8. R. Nagpal. Organizing a global coordinate system from local information on anamorphous computer. In A.I. Memo 1666. MIT A.I. Laboratory, August 1999.

9. P. Bahl and V. N. Padmanabhan. RADAR: an in-building rf-based user locationand tracking system. In Proc. of IEEE INFOCOM, March 2000.

10. M. Youssef and A. Agrawala. The Horus WLAN location determination system.In Proc. of ACM MobiSys, June 2005.

11. M. Azizyan, I. Constandache, and R. Roy Choudhury. Surroundsense: mobilephone localization via ambience fingerprinting. In Proc. of ACM MobiCom,2009.

References

12. http://www.ekahau.com/

13. Shwetak N. Patel , Location in Pervasive Computing, University of Washington.