Need suggestions for literature: Radar/Sonar pulse design

One recent reference that treats this as a Bayesian experimental
design problem:

Optimal pulses

C. Ray Smith and Paul M. Goggans 
Department of Electrical Engineering, University of Mississippi,
University, Mississippi 38677

This paper is concerned with a problem of designing optimal wave-forms.
Specifically, the problem is to control the data-generation process so
that data will be optimal toward answering a question, in the sense
that the probabilities for answers to the question are maximally
dispersed. The method presented here results in a variational principle
in which the functional to be extremized is based upon the evidence for
one hypothesis relative to another. As an example, the radar waveform
that will be optimal toward discrimination between two targets is
determined; the optimal waveform's dependence on the target-impulse
response functions is quite revealing, but in accord with intuition.
2001 American Institute of Physics.

Citation here (make sure it's one line):
http://scitation.aip.org/dbt/dbt.jsp?KEY=APCPCS&Volume=567&Issue=1#P000029000001

You could probably get a reprint from Paul Goggans at Ole Miss:

http://home.olemiss.edu/~goggans/

-Tom

-- 

To respond by email, replace "somewhere" with "astro" in the
return address.

"Stan Pawlukiewicz" <spam@spam.mitre.org> wrote in message 
news:d14k87$1g0$1@newslocal.mitre.org...
> Fred Marshall wrote:
>> "Stan Pawlukiewicz" <spam@spam.mitre.org> wrote in message
>>>I think that Cook and Bernfield is out of print.  I think that Riachek's 
>>>book was reprinted by Peninsula Publishers a while back.  I've heard many 
>>>people recomend Skolnick.
>>
>>
>> It's Cook and BernFELD.  It appears to be in print by Artech House.  The 
>> original was from Academic Press.  Amazon for $146.00
>> Still not what I'd recommend for directly applying to sonar except as 
>> background and interest.
>>
>>
>> Fred
> Well Fred, what would you recommend?

Stan,

Good question.  Considering it's Rune who's asking, I don't think I should 
recommend Urick's Principles of Underwater Sound - and it won't deal with 
the question anyway.  I don't know Waite's book and there's not enough 
detail on Amazon to get a sense for it.  Bottom line: there aren't that many 
books on the subject and most are introductory. After that, a lot of the 
stuff was classified.  There are more books on radar but my feeling is that 
they can easily lead one down a dead-end path unless the following are 
considered:

I seriously answered his question about waveform design in an earlier 
response.  One way of looking at the sonar / radar comparison and overlap is 
to look at fractional bandwidth - the receiver bandwidth as a fraction of 
center frequency.  Typically sonars have a large fractional bandwidth and 
the radars have a small fractional bandwidth.  A quick comment on this means 
that radar designers can more readily consider increasing their signal 
bandwidth than sonars designers can.  This means that sonars can't take as 
much advantage of wideband waveforms because it would mean increasing the 
receiver bandwidth too much (as in below DC is too much).  Also, because 
sonar has considerable variation in attenuation vs. frequency, really 
wideband waveforms are distorted / lowpass filtered and lose some of their 
advantage in trying to be wideband.

As a matter of interest: radar initially went to FM pulses because their 
transmitters were peak power limited.  In order to get higher total pulse 
energy they needed longer pulses.  Longer CW pulses would mess up range 
resolution.  In order to get better range resolution they went to FM so 
there could be pulse compression.  Also, at radar frequencies and target 
speeds Doppler is a smaller effect.  Not so for sonar.  An airliner travels 
at 500 knots or 277 yards/sec or 250 meters/sec out of 300M meters/sec or 
0.0000833% frequency shift one-way.  In sonar a ship may travel at 10 knots 
or 15 fps out of 5000 fps (in water) which is 15/5000 0.3% one-way.  So, 
radar range is easier to measure than Doppler on a single pulse basis I 
believe.  But Doppler is about equally easy to measure as range in sonar on 
a single pulse.  Anyway, if one is not so much peak power limited then this 
isn't the consideration.

Examples:  A 3kHz sonar might have a 300Hz bandwidth but not much more.  The 
transducers normally operate with some Q.  So, a 2kHz->6kHz signal design 
(bandwidth) would not be in the works.  As the center frequency increases 
then the bandwidth can increase.  So, if one can think of operating at 
200kHz where the attentuation is pretty high and ranges are pretty short 
then *maybe* a signal bandwidth higher than 20kHz would be OK.  The 
selection of optimum center frequency for a sonar is a bit interesting.  As 
you can see, it has a lot to do with the application.

Then there is another speed of sound vs. speed of light difference.  Radar 
has so much more capability to gather data on a target over multiple pulses. 
Sonar generally doesn't.  That is, until you get into ultrasound imaging 
sorts of applications where the frequencies are in the 100's of kHz and the 
ranges are in the centimeters.  In fact, many of the Doppler measuring 
methods in radar depend on multiple pulse time differences rather than on 
frequency shift - really the frequency shift of a pulse train.

Rune, what is the center frequency and range you're considering?  Is Doppler 
involved? These would shed some light on this question.

Fred

Fred Marshall wrote:
> "Stan Pawlukiewicz" <spam@spam.mitre.org> wrote in message
> news:d14k87$1g0$1@newslocal.mitre.org...
> > Fred Marshall wrote:
> >> "Stan Pawlukiewicz" <spam@spam.mitre.org> wrote in message
> >>>I think that Cook and Bernfield is out of print.  I think that
Riachek's
> >>>book was reprinted by Peninsula Publishers a while back.  I've
heard many
> >>>people recomend Skolnick.
> >>
> >>
> >> It's Cook and BernFELD.  It appears to be in print by Artech
House.  The
> >> original was from Academic Press.  Amazon for $146.00
> >> Still not what I'd recommend for directly applying to sonar except
as
> >> background and interest.
> >>
> >>
> >> Fred
> > Well Fred, what would you recommend?
>
> Stan,
>
> Good question.  Considering it's Rune who's asking, I don't think I
should
> recommend Urick's Principles of Underwater Sound - and it won't deal
with
> the question anyway.  I don't know Waite's book and there's not
enough
> detail on Amazon to get a sense for it.  Bottom line: there aren't
that many
> books on the subject and most are introductory. After that, a lot of
the
> stuff was classified.

I have the following books on sonar or sonar-related material,
in my shelf:

Waite: Sonar for Practising Engineers.
R.O. Jensen: Sonar Signal Processing
Urick: Principles of Underwater Sound
Urick: Sound Propagation in the Sea
F.B. Jensen & al: Computational Ocean Acoustics
Burdic: Underwater Acoustic System Analysis
van Trees: Detection, Estimation and Modulation Theory vols I-IV.

I have nothing (except for van Trees) on radar.

With the probable exception of van Trees, these at best mention the
matched filter in the context of pulse compression/increasing range
resolution. Most of them mention the ROC, but none that I am aware of
study how the matched filter increases the ability to detect weak
targets, which is what I want to do. But as you say, that's the kind
of stuff that would be classified. I think there may be something in
van Trees' volume I, though, but it takes time to read it and figure
out exactly what is there and what is not.

[... snip...]

> Then there is another speed of sound vs. speed of light difference.
Radar
> has so much more capability to gather data on a target over multiple
pulses.
> Sonar generally doesn't.

True. A few years ago I did a literature survey on synthetic aperture
sonar. It turned out all the theory was available in the literature
as early as 1976. I don't remember his name, but the author of this
paper had apparently been involved with the development of synthetic
aperture radar in the 1960ies, and had taken all the theory and
rewritten it once and for all in terms of the sonar problem, in one
huge article.

In the mid 1990ies, the ambition in the underwater acoustics community
was to design a synthetic aperture sonar that imaged the sea floor at
1 cm resolution at ranges ~1 km. For a 100 kHz carrier frequency, the
wavelength is on the order of 1.5 cm. The synthetic aperture imaging
principle requires the position of the platform to be controlled
to within a fraction of a wavelength, in this case on the order of
millimeters. The slow propagation speed of sound in water means that
at 750 m range, the two-way travel time is 1 s. Which means that
the sonar platform needs to hang virtually dead in the water. Which
isn't very easy to achieve in the real-world sea.

So while all the theory was available and served on a silver plate
in 1976, the technology needed to actually demonstrate the SAS
principle in the field, was not available until some 25 years later,
in 2000-2001. I don't know if the SAS ever evolved past the field
demonstrator stage, though.

> That is, until you get into ultrasound imaging
> sorts of applications where the frequencies are in the 100's of kHz
and the
> ranges are in the centimeters.  In fact, many of the Doppler
measuring
> methods in radar depend on multiple pulse time differences rather
than on
> frequency shift - really the frequency shift of a pulse train.
>
> Rune, what is the center frequency and range you're considering?  Is
Doppler
> involved? These would shed some light on this question.

I don't want to get too much into details about the application,
but center frequencies remain to be decided, depending on what field
equipment is available or possible to design. We used ~100 kHz in
the lab, but expect to end up a bit lower in a field system. At the
higher frequencies, we used low-Q transducers and were just able to
detect the target we wanted but with reasonable range resolution.
At the lower frequencies there were no problems detecting the target,
but due to high-Q transducers we couldn't achieve a range resolution
anywhere near what would be required in a field system. So unless we
can get low-Q low-frequency tranducers (which I suspect could be
difficult), we have to stick to the higher frequency ranges in order
to both detect the target and achieve the range resolution we need.
Hence the need to study alternative source signals. As you suggested
in your first post, we *could* try with CWs for longer durations than
we did this time.

Doppler is not involved. We can use multiple-pulse imaging.
Noise is expected to become a severe problem in the field. 

Rune

I havent heard anyone mentioning UWB (Ultra wide Band), or ground
penetrating radar, as a possible consideration- im thinking more in terms of
using the latest developments in processing. I think this can become
beneficial over other more popular methods when dealing with mediums having
NON uniform propagation ( such as the ground). Coupling energy efficiently
across a wide-band can be difficult, and the traditional "blast" (as in
geological imaging) is the poor-mans UWB sonar pulse... Of course, your
problem may not be relevant to this, but it seems you need a low-ish Q,
hence some BW.

Remember, you dont have to accurately create a pre-determined waveform, just
know what waveform has actually been sent.
( IE even a UWB "blast" is very useful, if you accurately record what it
looked like as it left the TX... but to use this to optimise image
construction requires some heavy number crunching).

Good luck!

Fred Marshall wrote:
> "Stan Pawlukiewicz" <spam@spam.mitre.org> wrote in message 
> news:d14k87$1g0$1@newslocal.mitre.org...
> 
>>Fred Marshall wrote:
>>
>>>"Stan Pawlukiewicz" <spam@spam.mitre.org> wrote in message
>>>
>>>>I think that Cook and Bernfield is out of print.  I think that Riachek's 
>>>>book was reprinted by Peninsula Publishers a while back.  I've heard many 
>>>>people recomend Skolnick.
>>>
>>>
>>>It's Cook and BernFELD.  It appears to be in print by Artech House.  The 
>>>original was from Academic Press.  Amazon for $146.00
>>>Still not what I'd recommend for directly applying to sonar except as 
>>>background and interest.
>>>
>>>
>>>Fred
>>
>>Well Fred, what would you recommend?
> 
> 
> Stan,
> 
> Good question.  Considering it's Rune who's asking, I don't think I should 
> recommend Urick's Principles of Underwater Sound - and it won't deal with 
> the question anyway.  I don't know Waite's book and there's not enough 
> detail on Amazon to get a sense for it.  Bottom line: there aren't that many 
> books on the subject and most are introductory. After that, a lot of the 
> stuff was classified.  There are more books on radar but my feeling is that 
> they can easily lead one down a dead-end path unless the following are 
> considered:
> 
> I seriously answered his question about waveform design in an earlier 
> response.  One way of looking at the sonar / radar comparison and overlap is 
> to look at fractional bandwidth - the receiver bandwidth as a fraction of 
> center frequency.  Typically sonars have a large fractional bandwidth and 
> the radars have a small fractional bandwidth.  A quick comment on this means 
> that radar designers can more readily consider increasing their signal 
> bandwidth than sonars designers can.  This means that sonars can't take as 
> much advantage of wideband waveforms because it would mean increasing the 
> receiver bandwidth too much (as in below DC is too much).  Also, because 
> sonar has considerable variation in attenuation vs. frequency, really 
> wideband waveforms are distorted / lowpass filtered and lose some of their 
> advantage in trying to be wideband.
> 
> As a matter of interest: radar initially went to FM pulses because their 
> transmitters were peak power limited.  In order to get higher total pulse 
> energy they needed longer pulses.  Longer CW pulses would mess up range 
> resolution.  In order to get better range resolution they went to FM so 
> there could be pulse compression.  Also, at radar frequencies and target 
> speeds Doppler is a smaller effect.  Not so for sonar.  An airliner travels 
> at 500 knots or 277 yards/sec or 250 meters/sec out of 300M meters/sec or 
> 0.0000833% frequency shift one-way.  In sonar a ship may travel at 10 knots 
> or 15 fps out of 5000 fps (in water) which is 15/5000 0.3% one-way.  So, 
> radar range is easier to measure than Doppler on a single pulse basis I 
> believe.  But Doppler is about equally easy to measure as range in sonar on 
> a single pulse.  Anyway, if one is not so much peak power limited then this 
> isn't the consideration.
> 
> Examples:  A 3kHz sonar might have a 300Hz bandwidth but not much more.  The 
> transducers normally operate with some Q.  So, a 2kHz->6kHz signal design 
> (bandwidth) would not be in the works.  As the center frequency increases 
> then the bandwidth can increase.  So, if one can think of operating at 
> 200kHz where the attentuation is pretty high and ranges are pretty short 
> then *maybe* a signal bandwidth higher than 20kHz would be OK.  The 
> selection of optimum center frequency for a sonar is a bit interesting.  As 
> you can see, it has a lot to do with the application.
> 
> Then there is another speed of sound vs. speed of light difference.  Radar 
> has so much more capability to gather data on a target over multiple pulses. 
> Sonar generally doesn't.  That is, until you get into ultrasound imaging 
> sorts of applications where the frequencies are in the 100's of kHz and the 
> ranges are in the centimeters.  In fact, many of the Doppler measuring 
> methods in radar depend on multiple pulse time differences rather than on 
> frequency shift - really the frequency shift of a pulse train.
> 
> Rune, what is the center frequency and range you're considering?  Is Doppler 
> involved? These would shed some light on this question.
> 
> Fred
> 
> 
> 
> 
> 
> 
> 
Nielson's "Sonar Signal Processing", goes through the issues that you 
mention.  The stuff is relatively straight forward in the noise limited 
case. The book that is needed, at least from my perspective, is one that 
goes into waveform design against reverberation.

Rune Allnor wrote:

>            ... At the
> higher frequencies, we used low-Q transducers and were just able to
> detect the target we wanted but with reasonable range resolution.
> At the lower frequencies there were no problems detecting the target,
> but due to high-Q transducers we couldn't achieve a range resolution
> anywhere near what would be required in a field system. So unless we
> can get low-Q low-frequency tranducers (which I suspect could be
> difficult), we have to stick to the higher frequency ranges in order
> to both detect the target and achieve the range resolution we need.
> Hence the need to study alternative source signals. As you suggested
> in your first post, we *could* try with CWs for longer durations than
> we did this time.

   ...

Why not use two frequencies, exploiting the best resources of each?

Jerry
-- 
Engineering is the art of making what you want from things you can get.
�����������������������������������������������������������������������

Simone Merrett wrote:
> I havent heard anyone mentioning UWB (Ultra wide Band), or ground
> penetrating radar, as a possible consideration- im thinking more in terms of
> using the latest developments in processing. I think this can become
> beneficial over other more popular methods when dealing with mediums having
> NON uniform propagation ( such as the ground). Coupling energy efficiently
> across a wide-band can be difficult, and the traditional "blast" (as in
> geological imaging) is the poor-mans UWB sonar pulse... Of course, your
> problem may not be relevant to this, but it seems you need a low-ish Q,
> hence some BW.
> 
> Remember, you dont have to accurately create a pre-determined waveform, just
> know what waveform has actually been sent.
> ( IE even a UWB "blast" is very useful, if you accurately record what it
> looked like as it left the TX... but to use this to optimise image
> construction requires some heavy number crunching).
> 
> Good luck!

I've never worked seriously on sonar systems. I'm just going by what 
others have told me. However, my understanding is wideband just falls 
apart. There is lots of refraction at layer boundaries in the ocean, and 
the refractive index is frequency dependant. Therefore the high 
frequency energy you get back is very often from a different place than 
the low frequency energy.

Regards,
Steve

"Stan Pawlukiewicz" <spam@spam.mitre.org> wrote in message 
news:d16vkb$895$1@newslocal.mitre.org...
> Fred Marshall wrote:
>> "Stan Pawlukiewicz" <spam@spam.mitre.org> wrote in message 
>> news:d14k87$1g0$1@newslocal.mitre.org...
>>
>>>Fred Marshall wrote:
>>>
>>>>"Stan Pawlukiewicz" <spam@spam.mitre.org> wrote in message
>>>>
>>>>>I think that Cook and Bernfield is out of print.  I think that 
>>>>>Riachek's book was reprinted by Peninsula Publishers a while back. 
>>>>>I've heard many people recomend Skolnick.
>>>>
>>>>
>>>>It's Cook and BernFELD.  It appears to be in print by Artech House.  The 
>>>>original was from Academic Press.  Amazon for $146.00
>>>>Still not what I'd recommend for directly applying to sonar except as 
>>>>background and interest.
>>>>
>>>>
>>>>Fred
>>>
>>>Well Fred, what would you recommend?
>>
>>
>> Stan,
>>
>> Good question.  Considering it's Rune who's asking, I don't think I 
>> should recommend Urick's Principles of Underwater Sound - and it won't 
>> deal with the question anyway.  I don't know Waite's book and there's not 
>> enough detail on Amazon to get a sense for it.  Bottom line: there aren't 
>> that many books on the subject and most are introductory. After that, a 
>> lot of the stuff was classified.  There are more books on radar but my 
>> feeling is that they can easily lead one down a dead-end path unless the 
>> following are considered:
>>
>> I seriously answered his question about waveform design in an earlier 
>> response.  One way of looking at the sonar / radar comparison and overlap 
>> is to look at fractional bandwidth - the receiver bandwidth as a fraction 
>> of center frequency.  Typically sonars have a large fractional bandwidth 
>> and the radars have a small fractional bandwidth.  A quick comment on 
>> this means that radar designers can more readily consider increasing 
>> their signal bandwidth than sonars designers can.  This means that sonars 
>> can't take as much advantage of wideband waveforms because it would mean 
>> increasing the receiver bandwidth too much (as in below DC is too much). 
>> Also, because sonar has considerable variation in attenuation vs. 
>> frequency, really wideband waveforms are distorted / lowpass filtered and 
>> lose some of their advantage in trying to be wideband.
>>
>> As a matter of interest: radar initially went to FM pulses because their 
>> transmitters were peak power limited.  In order to get higher total pulse 
>> energy they needed longer pulses.  Longer CW pulses would mess up range 
>> resolution.  In order to get better range resolution they went to FM so 
>> there could be pulse compression.  Also, at radar frequencies and target 
>> speeds Doppler is a smaller effect.  Not so for sonar.  An airliner 
>> travels at 500 knots or 277 yards/sec or 250 meters/sec out of 300M 
>> meters/sec or 0.0000833% frequency shift one-way.  In sonar a ship may 
>> travel at 10 knots or 15 fps out of 5000 fps (in water) which is 15/5000 
>> 0.3% one-way.  So, radar range is easier to measure than Doppler on a 
>> single pulse basis I believe.  But Doppler is about equally easy to 
>> measure as range in sonar on a single pulse.  Anyway, if one is not so 
>> much peak power limited then this isn't the consideration.
>>
>> Examples:  A 3kHz sonar might have a 300Hz bandwidth but not much more. 
>> The transducers normally operate with some Q.  So, a 2kHz->6kHz signal 
>> design (bandwidth) would not be in the works.  As the center frequency 
>> increases then the bandwidth can increase.  So, if one can think of 
>> operating at 200kHz where the attentuation is pretty high and ranges are 
>> pretty short then *maybe* a signal bandwidth higher than 20kHz would be 
>> OK.  The selection of optimum center frequency for a sonar is a bit 
>> interesting.  As you can see, it has a lot to do with the application.
>>
>> Then there is another speed of sound vs. speed of light difference. 
>> Radar has so much more capability to gather data on a target over 
>> multiple pulses. Sonar generally doesn't.  That is, until you get into 
>> ultrasound imaging sorts of applications where the frequencies are in the 
>> 100's of kHz and the ranges are in the centimeters.  In fact, many of the 
>> Doppler measuring methods in radar depend on multiple pulse time 
>> differences rather than on frequency shift - really the frequency shift 
>> of a pulse train.
>>
>> Rune, what is the center frequency and range you're considering?  Is 
>> Doppler involved? These would shed some light on this question.
>>
>> Fred
>>
>>
>>
>>
>>
>>
>>
> Nielson's "Sonar Signal Processing", goes through the issues that you 
> mention.  The stuff is relatively straight forward in the noise limited 
> case. The book that is needed, at least from my perspective, is one that 
> goes into waveform design against reverberation.

Stan,

Is it a "book" or a "reasonable method" that's needed????  :-)

In my experience, active sonar is reverberation limited for the early part 
of the ping cycle and can be noise limited thereafter.  So, it's range 
dependent for some sonars/situations.

Obviously, Doppler can be a discriminator.  So can angle of arrival if it's 
being measured and if there's at least a reaonable SRR (signal to 
reverberation ratio).  If there is no Doppler and if the situation is 
reverberation-limited then detection is a tough problem.  I know of no 
waveform tricks that help out in that case.  Do you or Rune?

Fred 

Fred Marshall wrote:
> "Stan Pawlukiewicz" <spam@spam.mitre.org> wrote in message 
> news:d16vkb$895$1@newslocal.mitre.org...
> 
>>Fred Marshall wrote:
>>
>>>"Stan Pawlukiewicz" <spam@spam.mitre.org> wrote in message 
>>>news:d14k87$1g0$1@newslocal.mitre.org...
>>>
>>>
>>>>Fred Marshall wrote:
>>>>
>>>>
>>>>>"Stan Pawlukiewicz" <spam@spam.mitre.org> wrote in message
>>>>>
>>>>>
>>>>>>I think that Cook and Bernfield is out of print.  I think that 
>>>>>>Riachek's book was reprinted by Peninsula Publishers a while back. 
>>>>>>I've heard many people recomend Skolnick.
>>>>>
>>>>>
>>>>>It's Cook and BernFELD.  It appears to be in print by Artech House.  The 
>>>>>original was from Academic Press.  Amazon for $146.00
>>>>>Still not what I'd recommend for directly applying to sonar except as 
>>>>>background and interest.
>>>>>
>>>>>
>>>>>Fred
>>>>
>>>>Well Fred, what would you recommend?
>>>
>>>
>>>Stan,
>>>
>>>Good question.  Considering it's Rune who's asking, I don't think I 
>>>should recommend Urick's Principles of Underwater Sound - and it won't 
>>>deal with the question anyway.  I don't know Waite's book and there's not 
>>>enough detail on Amazon to get a sense for it.  Bottom line: there aren't 
>>>that many books on the subject and most are introductory. After that, a 
>>>lot of the stuff was classified.  There are more books on radar but my 
>>>feeling is that they can easily lead one down a dead-end path unless the 
>>>following are considered:
>>>
>>>I seriously answered his question about waveform design in an earlier 
>>>response.  One way of looking at the sonar / radar comparison and overlap 
>>>is to look at fractional bandwidth - the receiver bandwidth as a fraction 
>>>of center frequency.  Typically sonars have a large fractional bandwidth 
>>>and the radars have a small fractional bandwidth.  A quick comment on 
>>>this means that radar designers can more readily consider increasing 
>>>their signal bandwidth than sonars designers can.  This means that sonars 
>>>can't take as much advantage of wideband waveforms because it would mean 
>>>increasing the receiver bandwidth too much (as in below DC is too much). 
>>>Also, because sonar has considerable variation in attenuation vs. 
>>>frequency, really wideband waveforms are distorted / lowpass filtered and 
>>>lose some of their advantage in trying to be wideband.
>>>
>>>As a matter of interest: radar initially went to FM pulses because their 
>>>transmitters were peak power limited.  In order to get higher total pulse 
>>>energy they needed longer pulses.  Longer CW pulses would mess up range 
>>>resolution.  In order to get better range resolution they went to FM so 
>>>there could be pulse compression.  Also, at radar frequencies and target 
>>>speeds Doppler is a smaller effect.  Not so for sonar.  An airliner 
>>>travels at 500 knots or 277 yards/sec or 250 meters/sec out of 300M 
>>>meters/sec or 0.0000833% frequency shift one-way.  In sonar a ship may 
>>>travel at 10 knots or 15 fps out of 5000 fps (in water) which is 15/5000 
>>>0.3% one-way.  So, radar range is easier to measure than Doppler on a 
>>>single pulse basis I believe.  But Doppler is about equally easy to 
>>>measure as range in sonar on a single pulse.  Anyway, if one is not so 
>>>much peak power limited then this isn't the consideration.
>>>
>>>Examples:  A 3kHz sonar might have a 300Hz bandwidth but not much more. 
>>>The transducers normally operate with some Q.  So, a 2kHz->6kHz signal 
>>>design (bandwidth) would not be in the works.  As the center frequency 
>>>increases then the bandwidth can increase.  So, if one can think of 
>>>operating at 200kHz where the attentuation is pretty high and ranges are 
>>>pretty short then *maybe* a signal bandwidth higher than 20kHz would be 
>>>OK.  The selection of optimum center frequency for a sonar is a bit 
>>>interesting.  As you can see, it has a lot to do with the application.
>>>
>>>Then there is another speed of sound vs. speed of light difference. 
>>>Radar has so much more capability to gather data on a target over 
>>>multiple pulses. Sonar generally doesn't.  That is, until you get into 
>>>ultrasound imaging sorts of applications where the frequencies are in the 
>>>100's of kHz and the ranges are in the centimeters.  In fact, many of the 
>>>Doppler measuring methods in radar depend on multiple pulse time 
>>>differences rather than on frequency shift - really the frequency shift 
>>>of a pulse train.
>>>
>>>Rune, what is the center frequency and range you're considering?  Is 
>>>Doppler involved? These would shed some light on this question.
>>>
>>>Fred
>>>
>>>
>>>
>>>
>>>
>>>
>>>
>>
>>Nielson's "Sonar Signal Processing", goes through the issues that you 
>>mention.  The stuff is relatively straight forward in the noise limited 
>>case. The book that is needed, at least from my perspective, is one that 
>>goes into waveform design against reverberation.
> 
> 
> Stan,
> 
> Is it a "book" or a "reasonable method" that's needed????  :-)
> 
> In my experience, active sonar is reverberation limited for the early part 
> of the ping cycle and can be noise limited thereafter.  So, it's range 
> dependent for some sonars/situations.
> 
> Obviously, Doppler can be a discriminator.  So can angle of arrival if it's 
> being measured and if there's at least a reaonable SRR (signal to 
> reverberation ratio).  If there is no Doppler and if the situation is 
> reverberation-limited then detection is a tough problem.  I know of no 
> waveform tricks that help out in that case.  Do you or Rune?
> 
> Fred 
> 
> 
Backscatter is a frequency dependent phenomena.  In some instances, it 
can be significant.  I can see some advantages of using a range of 
frequencies to probe a partially characterized environment. There are 
are other reasons why multiple frequencies make sense that you are 
probably aware of.

Jerry Avins wrote:
> Rune Allnor wrote:
>
> >            ... At the
> > higher frequencies, we used low-Q transducers and were just able to
> > detect the target we wanted but with reasonable range resolution.
> > At the lower frequencies there were no problems detecting the
target,
> > but due to high-Q transducers we couldn't achieve a range
resolution
> > anywhere near what would be required in a field system. So unless
we
> > can get low-Q low-frequency tranducers (which I suspect could be
> > difficult), we have to stick to the higher frequency ranges in
order
> > to both detect the target and achieve the range resolution we need.
> > Hence the need to study alternative source signals. As you
suggested
> > in your first post, we *could* try with CWs for longer durations
than
> > we did this time.
>
>    ...
>
> Why not use two frequencies, exploiting the best resources of each?
>
> Jerry

I did suggest that in my analysis report. The danger is that the
low frequency signal detects the target, while the high frequency
signal is hampered by noise, so that we use the wrong peak at the
detector output for localization. While using a low frequency
signal to confirm a detection is a viable idea in this application,
we still need the high frequency signal to work on its own.

Rune