Why Localization

1
Why Localization?

• Spatial coordinates are fundamental to tasks that
  are embedded in the physical world
• Beamforming for collaborative signal processing
• Energy-efficient geographic routing across a
  terrain
• Representation of queries, results, tasks in
  terms of spatial location
• Spatially aware user interfaces, applications

Collaborating sensors compute relative phase
offsets for beamforming
Geographic routing algorithm routes around a
hole
Which camera can see the frog?
Photograph the frog in the tree over there
(pointing)
2
Multimodal Localization

• Individual ranging and positioning mechanisms
  suffer from unique sources of error
• GPS Obstructing foliage, RF jamming
• Acoustic NLOS case, detection of reflected paths
• UWB similar to acoustics more resilient to
  solid obstructions but more scattering from
  foliage
• Solution combine data from several modalities
• Conditions suffering error in one mode may not in
  another.
• Two modes in agreement can resolve ambiguities
• Two modes in disagreement can detect errors
• A low-resolution mode can corroborate or
  contradict a high-resolution mode

3
Early Progress

• Experiments with acoustic ranging
• COTS hardware, DSP in software,
  synchronization through RF channel
• 10 KHz PN pulse, sliding correlator detection
• Quite accurate when LOS, reliably detecting NLOS
  hard
• Experiments with image based ranging
• Nodes emit coded IR pulses from LEDs
• Camera images field image processing algorithms
  detect individual node LEDs, reject reflections
• Visible LEDs are known LOS angular displacement
  used to approximate positions of visible nodes

4
Future Directions

• Individual modalities
• In-lab testbed (array of acoustic receivers,
  tag platform includes IR LED, PC-104 camera
  unit)
• Rigorous analysis, development of error models
• Combination of modalities
• Form local coordinate system by fusing and
  comparing data from multiple sensor sources
• Federation of adjacent coordinate systems
• Analysis of error propagation
• Statistical techniques (least squares)
• Multi-tier coordinates reference to local frame
  coordinates within local frame

5
UWB vs Acoustic

• Ultra Wide Band RF
• Spreads across 2-3 GHz band, with notches for
  GPS, 2.4 GHz
• ? 0.3m 0.1m severe scattering from foliage
  sized objects
• Low power makes signal difficult to detect
• Hard to implement must solve synchronization
  problem
• Fast clocks hard to reduce power consumption
• Acoustics
• Spreads across audible spectrum (100 Hz 20 kHz)
• ? 3m 0.01m corresponds to wide range of
  feature sizes
• Low power signals can be used, but detection is
  generally easier
• Relatively easy to implement synchronization by
  radio.
• Synchronization pulse may be vulnerable to
  jamming
• Environmental dependencies (temp, humidity, wind)