Pharos: Enable Physical Analytics through Visible Light based Indoor Localization (2013)

Published in

• Proceeding
  HotNets-XII Proceedings of the Twelfth ACM Workshop on Hot Topics in Networks

Online copies

Article / Slides

Synopsis

Localization through multilateration with LED bulbs. Each bulb uses BFSK to broadcast its own position to a receiver, distance from which to each bulb will be estimated by optical channel model.

Optical channel model

Given RSS, estimate distance between a receiver’s light sensor and the bulb.

Note: Overall RSS is estimated, given some components of RSS and the broadcasted duty cycle of PWM.

Assumptions of model

• Lambertian radiation pattern.
• All bulbs facing downward.
• Light sensor facing squarely upward.

Broadcast

Use BFSK for modulation to broadcast each LED bulb’s

• position, and
• duty cycle of PWM.

Mitigate collisions

Random channel hopping.

Channelization

Whole available spectrum is bounded by

• minimum frequency to prevent human perceivable flickering,
• minimum of LED bulb’s On/Off speed, and
• response speed of the light sensor on the receiver.

Multilateration

• Newton’s method, when \(n = 3\);
• LMS method, when \(n > 3\);

where \(n\) is number LED bulbs observed.

Evaluation

• Compared to LED based algorithms that locate receiver to
  • position of the light source with the highest RSS, or
  • weighted average of the locations of the observed bulbs, using their RSS as weights.
• Compared to Wi-Fi based systems in terms of
  • accuracy,
  • database usage, and
  • overhead.