Location Sensing Techniques

1
Location Sensing Techniques

• Maria Papadopouli
• Mobile Computing Group

2
Radio Propagation

• Blocking
• Reflection signal is reflected and it is not as
  strong as the original
• Scattering an incoming signal is scattered into
  several weaker outgoing signals
• Reflected energy is diffused in all directions
• Diffraction as in scattering, radio waves will
  be deflected at an edge and propagate in
  different directions

3
Radio Wave Propagation

• Due to multipath reflections from various
  objects, electromagnetic waves travel along
  different paths of varying lengths
• The original signal is spread due to different
  delays of parts of the signal
• Interaction between these waves cause multipath
  fading at a specific location
• Strength of waves ? as distance between
  transmitter receiver ?

4
Propagation Models

• Large-scale predict mean signal strength for T-R
  over large distances
• Small-scale (fading) characterize rapid
  fluctuations of received signal strength over
  very short travel distances or short times
• Path loss signal attenuation (dB)

5
Radio Propagation

• Received power at distance d from transmitter
• Pr K d-n
• Free space loss, n2 (clear line-of-sight T-R)
• Pr ( d ) ( Pt Gt Gr l2 ) / ( (4p)2 d2 L )
• Gt, Gr T/R antenna gain
• l wavelength in meters (carrier frequency)
• L system loss factor
• Typically, 4 ? n ? 3

6
Ground Reflection (Two-Ray Model)

• Reasonable for large-scale signal strength over
  distances of several kilometers for
  line-of-sight microcell channels in urban
  environment
• Pr ( d ) ( Pt Gt Gr ht2 hr2 ) / d4
• Gt, Gr T/R antenna gain
• ht,hr height of the T/R antenna

7
Indoor Propagation

• Environmental factors
• People, movement,furniture, walls, metallic
  surfaces, floors, other devices
• Metallic tinted windows yield greater attenuation
• Opening a door change of 10dBm in signal
  strength
• Other handheld devices present in test area up to
  5dBm
• Rotating the handheld device or placing a hand
  over the antenna
• Signal decays much faster

8
Indoor Propagation Example of Path Loss

• Path loss unit loss 10 n log (d) k F I W
• unit loss power loss (dB) at 1 m distance
• n power-delay index
• d distance between transmitter receiver
• k number of floors the signal traverse
• Floss per floor
• Inumber of walls the signal traverse
• Wloss per wall

9
Path Loss Exponents for Different Environments
10
WLAN deployment

• CAD measurement and prediction software such as
  SitePlanner
• Load blueprint of building within computer
• Propagation modeling techniques can predict user
  data throughput based on radio signal strength
  and interference prediction algorithms

11
Related work

• Hentys MS thesis Throughput measurements and
  empirical prediction models for 802.11b wireless
  LAN installations, Virginia Tech, 8/2001

12
Application Requirements

• Environment
• Outdoors
• Indoors
• Coverage/Range
• Infrastructure
• Deployment
• Power
• Weight

• Software
• Cost
• Training overhead
• Performance (scalability)
• Parameters that affect the accuracy
• Confidence

13
Mechanisms for location sensing

• Signal characteristics
• Different modalities of communication
• Radio frequency
• Infrared
• Ultrasound
• Video
• Combination of the above
• Signal propagation model

14
Signal characteristics

• Time-of-arrival of the signal
• Strength
• Structure (multi-path pattern)

15
Ultrasound systems

• Single transmission frequency (40kHz)
• Outdoors ? high probability of interference from
  other ultrasound sources

16
GPS-based

• Requires a GPS receiver
• Consumes power
• Adds weight
• Poor indoor coverage
• Outdoors
• 6-7 meters
• Not great for apps with high accuracy
  requirements

17
Software

• Do the wireless devices provide
  software-accessible signal strength or signature
  readings ?

18
Satellites

• 24 satellites
• L1 at 1575.42 MHz and L2 at 1227.6 MHz.
• Satellite signal
• Local FM radio
• 88MHz-108Mhz
• 100,000 Watts

19
Time-of-Arrival Measurementvia GPS-receiver

• Each satellite transmits a unique code
• Receivers gradually time shift internal clock
  till it corresponds to the received code
  (lock-on)
• Once locked-on to a satellite, the receiver
  determines the exact timing of the received
  signal in reference to its own internal clock
• If receiver satellite clock synchronized
• Distance travel-time speed-of-light
• GPS receivers have inaccurate clocks
• A mere microsecond corresponds to a 300-meter
  error

20
Quick Quiz

• How can you use timing information but without
  the need for time synchronization ?

21
Client-side Location SensingUsing Signal Strength
Signal strength measurements from APs
location, average signal strength
Store
Target
AP
Database
AP
Triangulation
AP
AP Access Point
22
Client-side Location SensingUsing Signal Strength
Signal strength measurements from APs
location, average signal strength
Store
Database
Target
AP
Response
AP
Query
Triangulation
AP
Client
AP Access Point
23
Pattern-matching algorithm

• Measure the signal strength from a computer to
  all available wireless access points
• Compare measurements to a table containing unique
  reading of signal strengths for each location
• Average signal strength samples and stores them
  in a table that it can reuse
• During use, the algorithm compares measure values
  to those in the table computes the differences
• Return the location in the entry with the
  smallest difference

24
Signature structure of signal
Signal is bouncing off of buildings and other
obstacles, reaching their destination (the base
station) via multiple paths
25
Client-side Location SensingUsing Signal
Structure
Signal strength measurements from APs
location, signal structure
Store
Target
AP
Database
AP
Triangulation
AP
AP Access Point
26
Client-side Location SensingUsing Signal
Structure
Signal strength measurements from APs
location, average signal structure
Store
Pattern matching
Target
Database
AP
Response
AP
Query
Triangulation
AP
Client
AP Access Point
27
Signal signature approach

• Radio frequency pattern multipath phase and
  amplitude characteristics of an operating cell
  phone or other wireless device
• Compares the RF pattern signature to a database
  of previously identified
• RF signatures and their corresponding geographic
  locations within the calibrated network
• By matching the signature pattern of the callers
  signal with the database of known signature
  patterns, the callers geographic location is
  identified

28
Application Requirements

• Environment
• Outdoors
• Indoors
• Coverage/Range
• Infrastructure
• Deployment
• Power
• Weight

• Software
• Cost
• Training overhead
• Performance (scalability)
• Accuracy
• Parameters, Confidence

29
Accuracy of measurements

• Density of access points needed
• Number of training points
• Line-of-sight requirement
• Impact of environmental factors
• Confidence range

30
Pattern-matching algorithm accuracy

• CMU-approach clients gather the data (1,500
  training points for a building)
• RADAR-approach access points gather the data

31
Energy consumption using CMU approach

• Force the wireless NI to scan nearby access
  points
• Perform location calculations once every 10 sec
• Decrease battery life by 6-8

32
Performance

• Scalability
• Number of training points required
• Number of users
• Can you use caching ?
• Location prediction heuristics ?

33
Related work

• Aura project
• RADAR
• GPS-less localization for small devices UNC-CH
  Tracker
• RadioCamera
• Hentys MS thesis Throughput measurements and
  empirical prediction models for 802.11b wireless
  LAN installations, Virginia Tech, 8/2001