1. Calamari’s Design Decisions Kamin Whitehouse June 18, 2003

2. Project Summary  Collecting large amounts of data 10,000’s data points 10,000’s data points Connectivity, RSSI, acoustic, ultrasound Connectivity, RSSI, acoustic, ultrasound  Calibration and auto-calibration techniques  Matlab simulation of algorithms using real data  Implementation in NesC on pc, mica, and dot3  Designing new hardware http://www.cs.berkeley.edu/~kamin/calamari

3. Outline  Design Requirements  Radio Ranging  Acoustic Ranging  Algorithms  TinyOS code and demo  Evaluation

4. A Motivating Application

5. Design Principles  Node-level Resolution  Scalable Deployment  Event-driven  Simple and Approximate Operation

6. Existing Systems  GPS  Cricket  AHLoS  Millibots

7. Radio Ranging – Connectivity Data courtesy Alec Woo, Ganesan, et al

8. Radio Ranging – Connectivity Data courtesy Alec Woo, Ganesan, et al

9. Radio Ranging – Signal Strength

13.  Error equation: error (cm) ≈ noise (dB). Attenuation rate (dB) cm

14. Radio Ranging – Signal Strength

15. Acoustic Ranging

16. Acoustic Ranging – 4.3KHz Analog  Simultaneously send acoustic and RF  Time stamp RF; turn on acoustic circuit  Time stamp tone-detector interrupt  Subtract timestamps  Multiply by speed of sound  Filter

17. Acoustic Ranging – 4.3KHz Analog

19. Acoustic Ranging – 4.3KHz Digital  Digital sampling and filtering  Better range and accuracy  Slow, costly process  Scheduling needed

20. Acoustic Ranging – Ultrasound

23. Localization 1 6 11 12 7 3 4 8 9 14 15 13 10 5 2

24. Localization Accuracy

28. NesC Implementation  Mica platform being integrated with VU  Dot3 being integrated with ultrasound  Simulated ranging estimates for PC

29. Evaluation  Node-level Resolution  Scalable Deployment  Event-driven  Simple and Approximate Operation