Transcript Document

Rice Atmospheric Information Network (RAIN)
Joe Halbouty, Clay McPheeters, Genaro Picazo, Ed Rodriguez, Daniel Wu
Project Motivation
The Transmitter
Assessment
- Floods in Houston, other cities costly
- Current systems are expensive and
don’t produce data in real-time
- RAIN sensor network: robust, realtime, inexpensive, non-invasive,
scalable design
- Designed to measure accurate meteorological data, predict flooding, early
warnings for residents, businesses
- Employs binary frequency shift key (FSK) transmission scheme at 9 kHz and 11 kHz; others possible
- Measurement and communication in same path
- Power supply of 6V battery regulated to 5V
- Comparator converts input of 0V or 5V to .8V or 1V
- AD654 voltage-to-frequency chip produces FSK bits
Assembled Transmitter Circuits
System Design & Principles
- Network of independent nodes: gather
local data, all compiled by central server
- Dual laser use for optical detection and
communication: low power, overhead vs.
radio communication
- Equipment cheap, standard:
laser-pointer, DSP vs. current RADAR
Receiver TMS320F2812 DSP
Transmitter Circuit Schematics
The Receiver
- TI TMS320F2812 DSP: real-time calculations
- Rainfall → scintillations in signal: sample @ 22 kHz,
calculate signal variance to measure rain rate
- Bandpass filter around 1 kHz: remove low frequency
turbulence variations, well-defined data relationship
System operation block diagram: transmitter and receiver
- Each node has transmitter, receiver
- Rain disturbs transmitted signal
- Received signal hits photodiode, output
to DSP for rain rate calculation
- Rain data transmitted via network to
server access points
- Transmitter design is well-suited for data
collection and transmission (FSK)
- Compact circuits keep node size small
- Receiver DSP progress slow, but results
are accurate real-time calculations
- Processing local data at nodes will
mitigate overhead of central processing
- Low-power equipment for efficient nodes
- Single node cost: probably < $400
DSP real-time calculated spectrum of
-----10 kHz input square wave
Real-time spectra of filtered output on DSP
- Constant K: relates laser data to a tipping bucket’s
data, calibrates computation of laser data
- Spherical domain equation gives rain rate directly:
2
2
 : variance,L : path length
  K  4L  f  h
f : filter bandw idth,h : rain rate
10 cm x 6 cm
13.5 cm x 7.5 cm
Looking Ahead
- Use DSP to solve for actual rain rate
- Equip DSP with D/A: data transmission
- Transmitter control: MSP430, GNOMES
- Complete design of independent nodes
- Networking: redundant, dense, efficient
Acknowledgements
- Profs. Young and Baraniuk, our advisors
- Stephen So and Patrick Frantz, for their
--extensive support with the F2812 board