Experimental Log: Roger Bland, November 25, 2001
Matched Filtering for RAFOS Sources
(continuing()
(
logbook
|
ifremer on RAFOS
|Garfield at
NPS
)
Today I hope to determine the correct function to fit the RAFOS
signals. I am working with one signal which is large enough to
determine some details of its shape.
The signal is SS1 from day 245, starting at about 4:34:58Z and ending at
4:36:17Z. The files involved are:
- r0124504.24m PSM data file (59,047,586 B)
- r0124504.24b beam-formed data: day0 (double) = 245.18291991,
idat(7,376,896) (int)
- r0124504.24w downsampled: day0 (double) = 245.18291991, lblk (long),
nblk (long), sig(57632) (float) (for lblk = 128)
- gdsig downsampled signal for this signal: npt (long),
yd(npt) (double), websig(npt) (float)
Now I will make a special file with the same 90-sec bite of data, but
without mixing or downsampling: s1245b.dat. The start time will be
4:34:50Z = 0.190856481 d, or point number 599319. The segment selected
is thus idat(599319:689318). The file contains: npt (long) = 90,000,
yd(npt) (double), websig(npt) (float).
I have finished a good attack on this problem. I found a good match
to the data with a linear sweep based on the following parameters (see
program):
- f1 = 259.375 Hz (the value suggested by IFREMER)
- W (freq. sweep) = 1.54 Hz (the IFREMER value was 1.523 Hz)
- f1 = 2.0 rad
- tshift = 8.65 x 10-5 days
- T (duration of sweep) = 80 sec.
The equation used to calculate the replica is
r(t) = sin f(t), t1 < t <
t1+T
r(t) = 0 otherwise
where
w(t) = w1 + W/T *
(t - t1)
and
f(t) = f1
+ w1 t + 1/2 (W/T) (t-t1)2
Plots
- Plot of replica beating against the
signal, for the best-fit parameters for the replica.