Basic Idea

1
Basic Idea
vairflow v12-v21/2

• Assuming the airflow is uniform between 1 and 2.
• Given distance (d), if v12 and v21 can be
  measured
• Vairflow (with an accuracy b/w 5cm/s to 5m/s)
  and also direction of flow can be determined.

2
Velocity measurements

• To begin with, lets assume we have the same
  sampling clock at origin and destination
• The transmitted and received signals have
  different phase.
• This phase shift is proportional to 2? f tflight
  2? f (d/v) 2? (f/v) d 2? d/?

3
Differential phase measurements
0
f1
d
f2
distance
time
Sampling clk _at_ rx

• The transmitted and received signals have
  different phases.
• This phase shift is proportional to 2? f tflight
  2? f (d/v) 2? (f/v) d 2? d/?
• We see that phase shift is proportional to f.!!
• Use two carriers ?1 2? f1 (d/v) ?2 2? f2
  (d/v)
• ??real ?1-?2 2? ?f (d/v)
• We can obtain velocity as v 2? ?f (d/ ??real
  )
• Differential measurements increase the
  periodicity significantly

4
How to measure phase
??measured2
f1
f2

• Xreceived(t)Acos(2?f1t ??real )
• Multiply this received signal by a reference
  cosine
• Y(t) Acos(2?f1t ??real ) cos(2?f1t)
• A/2 cos(??real ) cos(2?(2f1)t
  ??real)
• YDC1(t) A/2 cos(??real) After low pass
  filtering the double frequency term
• YDC2(t) A/2 sin(??real)

5
What about noise?!

• We wish to estimate the phase of a sinusoid
  embedded in white Gaussian noise.
• xn Acos(2?f0n ?) wn
• Knowing f0 and A we can get the maximum
  likelihood estimate for the phase.(from Kay,
  Statistical Signal Processing vol1., pp.167-168.)

6
Blocks to be added
YDC2 n2
MA
LPF
LPF
YDC2
sin(f1)
900
tan-1
Xreceived(t)
cos(f1)
??measured
YDC1
LPF
MA
LPF
YDC1n1

• As seen, statistical analysis calls for the
  averaging of low pass filter outputs.
• Lets use 80 tap moving average filters to smooth
  out the outputs.

7
Performance in Noise
2cm/s meas. error

• d 1m
• vacoustic 335m/s
• At 30dB SNR
• Background noise in a quite room is 45 dBA
• So we need a signal of 75dBA at receiver
• Still need to interpret this acoustic levels as
  electric signals

8
Simple protocol for implementation
D/A
uP
uP
v
v
A/D
B
A

• Initiate contact.(using radio)
• A transmits acoustic signal first. (DAC-gt
  speaker)600ksps
• B extracts phase approx 45ms. (Microphone-gtADC.)1
  20ksps
• B acknowledges and sends its measurement.(radio)
• B transmits its acoustic signal.
• A extracts phase and calculates velocity.

9
Agenda for the coming months.

• Using celoxica Handel C, program the FPGA.
• Perform fixed point simulations.
• Digital oscillator is ready and Acoustic sensor
  board is almost ready.(Both implemented by
  undergrad students.)
• also
• Try including multipath effects in models