2. OUTLINE
1. LOCATION TECHNOLOGIES
• Introduction
• Location Representation
• Infrastructure&Client-Based Location
Systems
• Approaches to Determining Location
• Error Sources in Location Systems
2. LOCATION SYSTEMS
• Global Positioning System, Active Badge,
Active Bat, Cricket, UbiSense, RADAR,
Place Lab, PowerLine Positioning,
ActiveFloor, Airbus, Tracking with
12.12.2012 Cameras
3. Introduction
Examples
Determining location
Entertainment,
• Specific location Navigation,
• Context information Asset tracking,
• Context aware Healthcare monitoring,
applications Emergency response
Trade-offs: accuracy, range and cost
12/12/2012
5. Infrastructure&Client-Based
Location Systems
THREE CLASSES of LS • LOCATION
• client-based PRIVACY
Gps
• network-based
X
Active Badge • Battery Life
• network-assisted • Processing and
aGps Store Capability
12.12.2012
6. Approaches to Determining
Location
reference points>1
• Proximity GPS satellite
• Trilateration WiFi access point
Cellular Tower
• Time of Flight
• Signal Strength Attenuation
• Hyperbolic Lateration
• Triangulation
• Dead Reckoning
12/12/2012
7. Proximity
• device vs reference point • NFC in cms
• closeness of a device examples • Bluetooth 10ms
• more RP -> more accuracy • WiFi 100ms
• Cellular phone kms
12/12/2012
8. Trilateration
• distance between a device
• intersections of reference
and a number of
point circles
reference points
• types:
time of flight of signal
attenuation of the
strength of the signal
12.12.2012
9. Trilateration /
Time of Flight
KNOWN NEEDS
• Speed of Sound : • precise clock
344 meters per second in 21 C
synchronization
• Speed of Light :
299,792,458 meters per second • instead round trip delay
EASY TO CALCULATE!
X = V.t
EXAMPLES:radio or light signal for light,
12.12.2012
ultrasonic pulse for sound
10. Trilateration /
Signal Strength Attenuation
-decrease of the signal’s Challenges
strength by factor of 1/r² -signal propagation
-r:distance from source medium
-reflaction,
diffraction,changin
g direction
11. Hyperbolic Lateration
CALCULATION
• time difference between
signal arrival times to
more 3 rp.
12. Triangulation
the angle of arrival
(AOA) of signals to
reference points
!angle
measurement
errors.
12.12.2012
13. Dead Reckoning
USES DEPENDS ON
• previously known accuracy of speed and
direction,
location use of accelemators for
• elapsed time acceleration, odometers
for distance, gyroscope for
• direction direction
• average speed
12.12.2012
14. Error Sources
Sources of Errors
AIM •Incorrect reference point
• produce accurate coordinates
location estimates
•Delay in signal
•Clock synchronization
•Multipath
•Geometry
12.12.2012
15. LOCATION SYSTEMS
• Based on the general
concepts discussed
• Commercial & research
systems
• Historically important
and current systems http://www.toasystems.com/
• Differing characteristics
among the solutions
12.12.2012
16. Characteristics
Metrics for
Evaluation
http://www.army-
• Scalability
• Resolution
technology.com/features/feature12
1877/feature121877-1.html
• Active vs. Passive
• Centralization
• Infrastructure
http://www.pixavi.com/systems-wireless-
telemetry.html
12.12.2012
18. Global Positioning System
GPS
• Most popular outdoor location tracking
system
• Indoor tracking problematic
building occlusions
• Started as 24 satellites
http://www.nist.gov/pml/div688/grp40
/gpsarchive.cfm
orbiting the Earth, Now 30
12.12.2012
19. Global Positioning System
(cont’d)
• Satellite transmission
location and the current time
At least 4
various frequencies satellite needed!
• Receiver
distance to satellite calculated
• Signal
ID code, ephemeris data, almanac data
Which Status, date,
satellite? Orbital data time
12.12.2012
20. Global Positioning System
(cont’d)
• Signals’ travel time
the time difference of arrival (TDOA)
• Location Negative
effect
hyperbolic lateration in 3-D Multipath,
TDOA values Atmospheri
c delays
• Fourth satellite is required to correct any
Minimizing Errors
synchronization errors • Predicting atmospheric
delays
• Increase the number of
channels
12.12.2012 • Correction codes
21. Global Positioning System
(cont’d)
http://www.iranmap.com/2010/04/10/gps-signal-and-errors
12.12.2012
22. Active Badge
Properties
•Indoor, Worn
density and
badges placement of
the sensors
• Resolution
• Active
• central database
• networked
sensors deployed
throughout a
12.12.2012 building
23. Active Badge (cont’d)
Metrics
• Scalability – difficult deployment
• Resolution – high if well deployed
• Active vs. Passive – needs active tagging
• Centralization – keeps a centralized db
and a lookup table
• Infrastructure – low cost IR, room
specific sensors
12.12.2012
24. Active Bat
The Dark Knight Rises
The bat
in 2012
vs. 1997
12.12.2012
25. Active Bat (cont’d)
Properties
• Ultrasound pulse’s travel time and
location
trilateration
initiating with RF signal
Vlight > Vsound
• Multiple tags must coordinate their
pulses so as not to interfere with each
other’s time-of-flight calculations.
12.12.2012
26. Active Bat (cont’d)
Metrics
• Scalability – more tags cause
interference, activeness decreases
scalability
• Resolution – 90% at 3cm
• Active vs. Passive – needs active tagging,
if passive RF signalling independant of
#tags
• Centralization – central server, managing
use of ultrasound bandwith, lack of
12.12.2012
27. Cricket
Transmitter
(beacon)
Tag
RF transmitter/receiver,
Properties
Ultrasonic signal • No centralized architecture!
receiver, microcontroller
• Tags compute their own location
• Method similar to Active Bat
12.12.2012
28. Cricket (cont’d)
Metrics
• Scalability – independant of #tags
• Resolution – 90% at 3cm
• Active vs. Passive – passive
• Centralization – decentralized, preserves
privacy by local calculations
• Infrastructure – no networking between
beacons, difficult to deploy because of line-
of-sight operation
12.12.2012
29. UbiSense
• Ultrawideband (UWB)
signal for localization
• Each Ubitag
incorporates a
conventional RF radio
(2.4 GHz) and a UWB
radio (6–8 GHz).
12.12.2012
30. UbiSense (cont’d)
• Time and Location
the time difference of arrival
(TDOA)
angle of arrival (AOA)
triangulation At least two
UbiSensors
• Advantage of using UWB pulses is that it
is
easier to filter multipath signals and can
endure some occlusion
12.12.2012
31. UbiSense (cont’d)
Metrics
• Scalability – dependant of #tags, separate
coordination channel in favor
• Resolution – 90% at 15cm
• Active vs. Passive – active
• Centralization – centralized
• Infrastructure – physical timing cable,
difficult to deploy because of line-of-sight
operation
12.12.2012
32. Radar
Properties
• RF signal strength as indicator of the distance between an
AP and a receiver. Makes use of 802.11 WiFi network.
• Consumer does not have to purchase any specialized
equipment (WiFi-enabled mobile phones, PDAs can be
handled as a receiver or tag.)
• Problems with multipath led researchers to use a mapping
approach for localization
• Receiver measures signal strength and compares it with
the offline signal map
• Subject to environment change
12.12.2012
33. Radar (cont’d)
Metrics
• Scalability – dependant of #tags
• Resolution – 90% at 6m
• Active vs. Passive – active
• Centralization – decentralized
• Infrastructure – reuse of existing
infrastructure
12.12.2012
34. PlaceLab
Properties
• software-based indoor and outdoor
localization system.
• Makes use of 802.11 WiFi network. GSM towers,
Bluetooth
• detecting multiple unique IDs from these existing radio
beacons and referring to a map of these devices
So far localization similar to RADAR...
• location tracking at a larger scale outdoor
• Less dense calibration data, no need for an individual to
populate a signal map no surveying
12.12.2012
35. PlaceLab
War Driving
• War driving is the process of driving around with a
mobile device equipped with a GPS receiver and an 802.11,
GSM, and/or Bluetooth radio to collect traces of wireless
base stations.
time-stamped recordings containing GPS coordinates
the associated signal strength of the beacons
Location
• Position of the device is a weighted average of positions of
the overheard beacons millions of beacon estimates
already determined
12.12.2012
36. PlaceLab (cont’d)
Metrics
• Scalability – makes use of already
determined estimations, still dependant on
existance of tags
• Resolution – 90% at 20m
• Active vs. Passive – active
• Centralization – no central provider, clients
can determine their location privately
• Infrastructure – reuse of existing
infrastructure
12.12.2012
38. PowerLine Positioning
(cont’d)
Properties
• drawbacks to relying on public infrastructure
• indoor localization to work in nearly every building
use the power line as the signaling infrastructure!
• modules continually emit their respective signals
over the power line, tags sense these signals in a building,
relay them wirelessly to a receiver
• site surveying needed
12.12.2012
39. PowerLine Positioning
(cont’d)
Metrics
• Scalability – dependant of #tags
• Resolution – 90% at 1m
• Active vs. Passive – needs active tagging
• Centralization – local or central
• Infrastructure – lower deployment costs
12.12.2012
40. Active Floor
Footstep signature
No tags! Also by ground reaction
Load sensors force
Tiles
12.12.2012
41. Active Floor (cont’d)
Metrics
• Scalability – independant of clients,
assuming only one individual on a single tile
• Resolution – 91% at 1m
• Active vs. Passive – passive
• Centralization – central
• Infrastructure – custom tiles makes
deployment difficult
12.12.2012
42. Airbus
• detecting gross • presence of a
human movement person
and room transitions • mass rather than
by sensing individual
differential air
pressure central heating,
ventilation, and
air conditioning
(HVAC)
• less obtrusive
than installing
motion detectors
12.12.2012
43. Airbus (cont’d)
Metrics
• Scalability – scalable in the installed
environment
• Resolution – 88% at room level
• Active vs. Passive – passive
• Centralization – central, HVAC is the single
monitoring point
• Infrastructure – less additional
infrastructure for deployment
12.12.2012
44. Tracking with Cameras
Properties
• cameras and computer vision techniques
• no specialized tag and possible to leverage existing
cameras
• stereo camera images for locating the position, color
images
for inferring identities
• face recognition
On The Other Hand;
• occlusions
• dependant on the field of view of cameras, difficult
coordination, small close space tracking not possible
• privacy concerns
12.12.2012
45. Tracking with Cameras
(cont’d)
Metrics
• Scalability – scalable, independant of
#people
• Resolution – 50% to 80% at 1m
• Active vs. Passive – passive
• Centralization – central
• Infrastructure – reuse of existing
infrastructure is possible
12.12.2012
47. Summary
• basic concepts of location technologies
• current and historical location systems
• client-based vs. network-based positioning
• major sources of error
• challenges and opportunities
No single location technology today that is
ubiquitous, accurate, low-cost and easy to deploy.
Road to integration!
12.12.2012
49. Referenced from Article
Location in Ubiquitous
Computing
Alexander Varshavsky and
Shwetak Patel
12.12.2012
Notes de l'éditeur
Inclient-basedlocationsystem, device computes it ownlocationbyusingsignalsthatcomesfromthesatelliteor a network. GPS can be given as an example. A navigation device calculatesitspositionbyusingreceivedsignalsfrom at least 4 GPS satellites.In network-basedsystems, network infrastructurewillcalculatethedevice’slocation. It is commonlyused in thebuildingstodetectwhere is thuser. The device which is carriedorwornby a person as badgesendsinfraredsignalstothedeployed sensor networks in thebuilding.In network assistedlocationsystems, both device andtheinfrastructuretakeplace in calculatingtheposition. As an example, assisted GPS can be given.The device calculatesitsownlocationbyusingagain GPS signalsandalsoadditionalinformation is takenfromthecellular network infrasturture(usually network operator) tospeedupthefindingthesatellites.Themainadvantage of theclientbasedlocationsystem is that it forbidsthetransmittingany data tooutsidefromthe device. Sotheprivacy is protected.But the as a disadvantage, our device has touseitsownprocessingandstoragecapabilities, whichcausestoshortage of thebattery life.Inothersystems, yourplace is knownfromtheinfrastructure.
Wewill talk aboutsixfundementalapprocahesfordeterminingthelocation of a device. Thesearelistedhere.Some of thesetechniquesrequiresoneormorereferencepointswhosepreciselocation is known in advance. Examplerefencepointsare GPS, wifiaccesspointsor a cellulartower.
Based on the general concepts and techniquessignaling tech-nology (e.g., IR, RF, load sensing, computer vision, or audition), line-of-sight requirements, accuracy, and cost of scaling
An important consideration is theperformance or accuracy of the system and its resolution (e.g., low resolution for weather forecasts and high resolution for indoor navigation).At the same time, one must consider the infrastructure requirements toevaluate the ease of deployment, cost and installation, and maintenanceburden. (simultaneously deploying, cost for ordinary home-owners vs companies)Spectral requirements (in hospitals against RF IR preferred)
An important consideration is theperformance or accuracy of the system and its resolution (e.g., low resolution for weather forecasts and high resolution for indoor navigation).At the same time, one must consider the infrastructure requirements toevaluate the ease of deployment, cost and installation, and maintenanceburden. (simultaneously deploying, cost for ordinary home-owners vs companies)Spectral requirements (in hospitals against RF IR preferred)
fully operational in 1994. started at 1973.GPS first originated for military applications, but today, GPS-based solutions permeate throughout many civilian and consumer applications, such as in-car navigation systems, marine navigation, and fleet management services.Civilian GPS has a median accuracy of 10 metersoutdoor, but areas with substantial occlusions, such as tall buildings andlarge mountains can reduce the accuracy of the system.30 healthy, 2 old satellite
These GPS satellites transmit data over various radio frequencies, designated as L1, L2, etc. Civilian GPS uses the L1 frequency of 1575.42 MHz inthe ultrahigh frequency band.
Unlike the GPS satellites, GPS receivers do not have atomic clocks andare not synchronized with the GPS satellites. therefore, a GPS receivercalculates the time difference of arrival (TDOA) using the timing slackrequired to synchronize the GPS receiver’s generation of a pseudorandomID code with those being transmitted by the satellite to determine the sig-nals’ travel time. To determine its location, the receiver applies hyperboliclateration in 3-D using the estimated TDOA values. In addition, a fourthsatellite is required to correct any synchronization errors.some factors the can degradethe quality of the GPS signal originating from the satellites:Multipath—occurs when the GPS signal is reflected off tall build-ings, thus increasing the time-of-flight of the signal.Too few satellites visible—occurs when there are major obstructions(e.g., GPS does not work well indoors or underground). Atmospheric delays—signals can slow as they pass through theatmosphere.predict and model the atmospheric delays and apply a constant correctionfactor to the received signal. !e other strategy is to increase the numberof channels in the receiver to allow for more satellite signals to be seen. Arecent system, called differential GPS, uses a collection of terrestrial beaconsto emit correction codes (using long wave radio between 285 and 325 kHz) inmultipath-prone areas. the accuracy of differential GPS has been shown tobe 1.8 meters at least 95% of the time (LaMarca and de Lara, 2008). Anotherapproach called Real-Time Kinematic GPS uses phase measurements fromexisting GPS signals to provide receivers with real-time corrections.
1992The badge transmits a unique code via a pulse-width modulated IR signal tonetworked sensors/receivers deployed throughout a building. !e ActiveBadge uses 48-bit ID codes and is capable of two-way communication.
1997Like push vs pullPassive betterBat vs cricketMultiple tags must coordinate their pulses so as not to interfere with each other’s time-of-flight calculations.The system supports 75 tags being tracked in a 1000 square meters space consisting of 720 receivers.
Given three or more measurements to the receivers, the 3-D position of the tagcan be determined using trilateration.negligible RF trip
Multiple tags must coordinate their pulses so as not to interfere with each other’s time-of-flight calculations.The system supports 75 tags being tracked in a 1000 square meters space consisting of 720 receivers.instrumentation to space,
2000
at leasttwo Ubisensors
networking between sensors
2000RADAR system implements a location service using the information obtained from an already existing 802.11 WiFi network.RADAR uses the RF signal strength [also knownas the received signal strength indicator (RSSI)] as an indicator of the dis-tance between an AP and a receiver. the major advantage of this approachis that a consumer does not have to purchase any specialized equipmentand can still benefit from a location-aware application. For example, exist-ing devices, such as WiFi-enabled mobile phones, PDAs, or laptops, can berepurposed as a receiver or tag.
2000RADAR system implements a location service using the information obtained from an already existing 802.11 WiFi network.RADAR uses the RF signal strength [also knownas the received signal strength indicator (RSSI)] as an indicator of the dis-tance between an AP and a receiver. the major advantage of this approachis that a consumer does not have to purchase any specialized equipmentand can still benefit from a location-aware application. For example, exist-ing devices, such as WiFi-enabled mobile phones, PDAs, or laptops, can berepurposed as a receiver or tag.has a known median error of 5 meters and a 90 percentile resolu-tion of 15–20 meters
2005Place Lab runs on commodity devices such as notebooks, PDAs, and mobile phones, and determines their position using radio beacons, such as 802.11 APs, GSM cellphone towers, and fixed Bluetooth devices that are already deployed in the environment
2005Wigle.net and Worldwidewardrive.org are some examples of war driving repositoriesthat contain millions of known APs.A similar effort was also started in Japan by the Sony Computer ScienceLaboratory called PlaceEngineTM (http://www.placeengine.com/en). Place Engineprovides a mechanism for a community of users to update 802.11 beaconpositions and the ability to track the location of any WiFi-enabled device.Place Lab also inspired commercial products such as Skyhook (http://www.skyhookwireless.com/) and Navizon (http://www.navizon.com/).
Although this approach only infers the locationof the beacons, it has the added benefit that millions of beacon estimateshave already been determined. thus, this allows the ability to scale a loca-tion tracking system much more quickly despite the loss in accuracy. thisapproach has shown a median accuracy of 20–30 meters in large cities.
2006, 2008
drawbacks to relying on public infrastructure or thedeployment of many beaconsPowerLine Positioning is the first example of a whole-house or whole-building indoor localization systemthat repurposes the electrical system.Depending on the location of the tag, the detected signal levelsprovide a distinctive signature, or fingerprint, resulting from the densityof electrical wiring present at the given location and the distance from theplug-in module!ese modules inject a mid-frequency (300–1600 kiloHertz), attenuated signal throughout the electricalsystem of the home. Both modules continually emit their respective signalsover the power line, and location tags equipped with specially tuned tagssense these signals in a building and relay them wirelessly to a receiver inthe building. Depending on the location of the tag, the detected signal levelsprovide a distinctive signature, or "ngerprint, resulting from the densityof electrical wiring present at the given location and the distance from theplug-in module. PowerLine Positioning is capable of providing subroom-level positioning for multiple regions of a building. !e current PLP systemhas a median error of 0.75 meters and a 90 percentile accuracy of 1 meter
These modules inject a mid-frequency (300–1600 kiloHertz), attenuated signal throughout the electricalsystem of the home. Both modules continually emit their respective signalsover the power line, and location tags equipped with specially tuned tagssense these signals in a building and relay them wirelessly to a receiver inthe building. Depending on the location of the tag, the detected signal levelsprovide a distinctive signature, or fingerprint, resulting from the densityof electrical wiring present at the given location and the distance from theplug-in module. PowerLine Positioning is capable of providing subroom-level positioning for multiple regions of a building. !e current PLP systemhas a median error of 0.75 meters and a 90 percentile accuracy of 1 meter
1997This heel-to-toe transfer time and heel/toe GRF values canbe used to calculate a footstep signature. ActiveFloor uses these featuresfrom the footstep signatures to build a hidden Markov model in order toidentify the person. For further reading, a similar approach is also used bythe Smart Floor project (http://www.cc.gatech.edu/fce/smartfloor/).When an individual walks on the surface, the reac-tion that the load sensors produce in response to the weight and inertia of abody in contact with the *oor is called the ground reaction force (GRF) (inthis case, the person’s foot)
2008Disruptionsin home air*ow caused by human movement through the house, espe-cially those caused by the blockage of doorways and thresholds, result instatic pressure changes in the HVAC air handler unit. !e system detectsand records this pressure variation using di#erential sensors mounted onthe air "lter and classi"es where certain movement events are occurring,such as an adult walking through a particular doorway or the openingand closing of a door.
2000 oncameras and computer vision techniquesdistinctive textures in the pair of images to determine which points from the left cameraimage corresponds to a particular point in the right camera image.Blob detection and background subtraction techniques are used to infer the location of moving objects (usuallypeople) in the camera’s view.
Challenge of building and maintaining location-aware middleware and location-aware back-end services from limited existing solutions. making the informationavailable to third-party applications in a scalable and privacy-preserv-ing manner.Opportunity For instance, although the median error of GPS is 10 meters, the combined solu-tion of GPS and European’s global navigation satellite system called Galileo(as soon as it comes online) should yield a median accuracy of 1.5 meters.For example, a recent study showed that users want plausible deni-ability in a location system (Iachello et al., 2005). Another study showedthat people’s preferences for disclosing location information differs basedon many parameters, including the location of the user and the other per-son, the current user activity, and the relationship between the user andthe other person