I'm going to go out on a limb here. The following is based on what I understand
about discrete-time filters, and I admit I am not fluent in their design, but
may understand the theory enough to post this....
12/9/06 10:06 PM
Dr. Gerald N. Johnson wrote:
But an FIR or IIR filter can't compare coherent signal to incoherent
noise and enhance multiple signals while suppressing the noise between
them.
Correct, but that's where autocorrelation function is applied. The
autocorrelation results are used to adjust the coefficients of the FIR/IIR
filter in a way which increases the next autocorrelation result (provided there
is some coherency in the band pass). So the filter transfer function converges
on a solution to maximize coherentness, whatever that may look like in the
frequency domain. If there were 5 discrete tones in a passband, the filter would
take a form that would look (in the frequency domain) like 5 bandpass filters,
one centered on each tone.
A FIR/IIR filter can be made to have any arbitrary transfer function, although
it is mostly used for the standards (low pass, high pass, bandpass and bandstop
(notch)). This allows adaption techniques to change it to fit around an said
arbitrary set of signals within a passband.
12/9/06 12:13 AM
Dr. Gerald N. Johnson wrote:
I propose a test to compare how a narrow filter might work and how
autocorrelation might work. Look for some weak multiple tone signals,
perhaps a digital multiple tone signal or many weak signals in a DX pile
up spread over the receiver bandwidth. Attenuate those signals to be in
the noise. Apply the NR. Autocorrelation has a chance of enhancing all
the CW signals, while a single adaptive filter can only do ONE. It takes
As stated above, a single filter using autocorrelation adaption, as mentioned
above, can enhance any arbitrary number of signals only limited by its
resolution which is defined by the number of coefficients (taps) used.
The same coefficients used to attenuate signals of frequency x and pass
frequency y can be change to pass both signals. But since it is working in the
time domain, all coefficients may need to be adjusted to make this change.
as many parallel adaptive filters as you have separate signals to go the
filter approach and I don't think there's enough DSP in your receiver to
do that while there can be an autocorrelation time delay that enhances
all the tones. It may be long because the tones are not coherent with
each other and the autocorrelation delay has to find the delay that
amounts to full cycles for all the tones at the same time. This test
might work better with two or three than with 50 signals.
These filters are applied not to the baseband signal, but to the final IF
(centered on 14.5kHz for the Orion(s), if I'm not mistaken), therefore, the
total latency is manageable.
You might remember some saying that the early Orion V1.xxx NR caused SSB to be
"muddy" sounding. I don't know personally know how it sounded, but if one
considers that coherency is what was being allowed to tranfer through the
filter, then all unvoiced speech sibilants would be filtered out. These are the
s, sh, ch and f sounds of speech. I'm not sure, but this loss of sibilants may
cause a voice to sound muddy.
Lin
WB1AIW
|