Forums

Simple SONAR Detection Question

Started by Randy Yates October 12, 2015
Consider the fundamental concept of a sonar (or radar) hich transmits a
pulse and is looking for the return.

Are there targets which significantly distort the pulse shape, so that
using a simple correlation (matched filter) for the detector results in
degraded performance?

In other words, it seems like what you REALLY want to use is a return,
or received, signal correlator, not a transmit signal correlator. Of course that
would require knowledge of the "target transfer function" H(w), so that
what you are correlating against is R(w) = T(w)H(w).

Other than a difference in the possible expected H(w)'s, are there any
differences in this problem for a radar versus a sonar?

Thoughts?
-- 
Randy Yates
Digital Signal Labs
http://www.digitalsignallabs.com
On Mon, 12 Oct 2015 13:49:27 -0400, Randy Yates
<yates@digitalsignallabs.com> wrote:

>Consider the fundamental concept of a sonar (or radar) hich transmits a >pulse and is looking for the return. > >Are there targets which significantly distort the pulse shape, so that >using a simple correlation (matched filter) for the detector results in >degraded performance? > >In other words, it seems like what you REALLY want to use is a return, >or received, signal correlator, not a transmit signal correlator. Of course that >would require knowledge of the "target transfer function" H(w), so that >what you are correlating against is R(w) = T(w)H(w). > >Other than a difference in the possible expected H(w)'s, are there any >differences in this problem for a radar versus a sonar? > >Thoughts?
It's been a long time since I worked on radar, but this was never an issue when I did, i.e., you always correlated against the Tx pulse. If you are expecting or detecting Doppler, that has to be accounted for, but I don't think that's what you're asking about. Sonar has other things to deal with, like dispersion due to thermoclines, etc., and I think there are countermeasures that are acoustically absorptive and maybe even frequency selective. Eric Jacobsen Anchor Hill Communications http://www.anchorhill.com
On 10/12/2015 1:49 PM, Randy Yates wrote:
> Consider the fundamental concept of a sonar (or radar) hich transmits a > pulse and is looking for the return. > > Are there targets which significantly distort the pulse shape, so that > using a simple correlation (matched filter) for the detector results in > degraded performance? > > In other words, it seems like what you REALLY want to use is a return, > or received, signal correlator, not a transmit signal correlator. Of course that > would require knowledge of the "target transfer function" H(w), so that > what you are correlating against is R(w) = T(w)H(w). > > Other than a difference in the possible expected H(w)'s, are there any > differences in this problem for a radar versus a sonar?
I was never well versed in SONAR, but I do remember there are dispersive effects that distort a pulse shape. One of the effects is just reverb where some of the transmitted signal bounces back from the medium itself. I would expect your filter might want to compensate for the distortion, but that would need to be function of distance which could be substituted for by time. I guess the question is, just how significant is the distortion? I don't recall any of the sub systems using such a filter, but like I said, I wasn't in that job long enough to learn much. -- Rick
eric.jacobsen@ieee.org (Eric Jacobsen) writes:

> On Mon, 12 Oct 2015 13:49:27 -0400, Randy Yates > <yates@digitalsignallabs.com> wrote: > >>Consider the fundamental concept of a sonar (or radar) hich transmits a >>pulse and is looking for the return. >> >>Are there targets which significantly distort the pulse shape, so that >>using a simple correlation (matched filter) for the detector results in >>degraded performance? >> >>In other words, it seems like what you REALLY want to use is a return, >>or received, signal correlator, not a transmit signal correlator. Of course that >>would require knowledge of the "target transfer function" H(w), so that >>what you are correlating against is R(w) = T(w)H(w). >> >>Other than a difference in the possible expected H(w)'s, are there any >>differences in this problem for a radar versus a sonar? >> >>Thoughts? > > It's been a long time since I worked on radar, but this was never an > issue when I did, i.e., you always correlated against the Tx pulse. > If you are expecting or detecting Doppler, that has to be accounted > for, but I don't think that's what you're asking about. > > Sonar has other things to deal with, like dispersion due to > thermoclines, etc., and I think there are countermeasures that are > acoustically absorptive and maybe even frequency selective.
Thanks for the datapoints, Eric. All good to know. -- Randy Yates Digital Signal Labs http://www.digitalsignallabs.com
rickman <gnuarm@gmail.com> writes:

> On 10/12/2015 1:49 PM, Randy Yates wrote: >> Consider the fundamental concept of a sonar (or radar) hich transmits a >> pulse and is looking for the return. >> >> Are there targets which significantly distort the pulse shape, so that >> using a simple correlation (matched filter) for the detector results in >> degraded performance? >> >> In other words, it seems like what you REALLY want to use is a return, >> or received, signal correlator, not a transmit signal correlator. Of course that >> would require knowledge of the "target transfer function" H(w), so that >> what you are correlating against is R(w) = T(w)H(w). >> >> Other than a difference in the possible expected H(w)'s, are there any >> differences in this problem for a radar versus a sonar?
Hey Rick, Thanks for your thoughts.
> I was never well versed in SONAR, but I do remember there are > dispersive effects that distort a pulse shape. One of the effects is > just reverb where some of the transmitted signal bounces back from the > medium itself. I would expect your filter might want to compensate > for the distortion, but that would need to be function of distance > which could be substituted for by time.
Yeah, but even if you knew time/distance, how would you know what to compensate?
> > I guess the question is, just how significant is the distortion? I > don't recall any of the sub systems using such a filter, but like I > said, I wasn't in that job long enough to learn much.
Was it a sonar or radar? -- Randy Yates Digital Signal Labs http://www.digitalsignallabs.com
On 10/12/2015 3:58 PM, Randy Yates wrote:
> rickman <gnuarm@gmail.com> writes: > >> On 10/12/2015 1:49 PM, Randy Yates wrote: >>> Consider the fundamental concept of a sonar (or radar) hich transmits a >>> pulse and is looking for the return. >>> >>> Are there targets which significantly distort the pulse shape, so that >>> using a simple correlation (matched filter) for the detector results in >>> degraded performance? >>> >>> In other words, it seems like what you REALLY want to use is a return, >>> or received, signal correlator, not a transmit signal correlator. Of course that >>> would require knowledge of the "target transfer function" H(w), so that >>> what you are correlating against is R(w) = T(w)H(w). >>> >>> Other than a difference in the possible expected H(w)'s, are there any >>> differences in this problem for a radar versus a sonar? > > Hey Rick, > > Thanks for your thoughts. > >> I was never well versed in SONAR, but I do remember there are >> dispersive effects that distort a pulse shape. One of the effects is >> just reverb where some of the transmitted signal bounces back from the >> medium itself. I would expect your filter might want to compensate >> for the distortion, but that would need to be function of distance >> which could be substituted for by time. > > Yeah, but even if you knew time/distance, how would you know what to > compensate? > >> >> I guess the question is, just how significant is the distortion? I >> don't recall any of the sub systems using such a filter, but like I >> said, I wasn't in that job long enough to learn much. > > Was it a sonar or radar?
SONAR. I was thinking active. Otherwise what pulse would you be working with? With active you know the time of flight and from that can get a good value for range. I would expect the pulse distortion to be fairly constant possibly only varying a bit with temperature. A lot of SONAR word is relatively constant temperature because the sounds bounce off thermal layers and travel almost like in a waveguide. Subs have their own assumptions about environments. What exactly is your environment? Rereading your question, you are asking about distortion from the target. I can't say I've ever considered that or seen where anyone else did. But as I said, my experience was short and some 20+ years ago. -- Rick
On Monday, October 12, 2015 at 4:05:22 PM UTC-4, rickman wrote:
> On 10/12/2015 3:58 PM, Randy Yates wrote: > > rickman <gnuarm@gmail.com> writes: > > > >> On 10/12/2015 1:49 PM, Randy Yates wrote: > >>> Consider the fundamental concept of a sonar (or radar) hich transmits a > >>> pulse and is looking for the return. > >>> > >>> Are there targets which significantly distort the pulse shape, so that > >>> using a simple correlation (matched filter) for the detector results in > >>> degraded performance? > >>> > >>> In other words, it seems like what you REALLY want to use is a return, > >>> or received, signal correlator, not a transmit signal correlator. Of course that > >>> would require knowledge of the "target transfer function" H(w), so that > >>> what you are correlating against is R(w) = T(w)H(w). > >>> > >>> Other than a difference in the possible expected H(w)'s, are there any > >>> differences in this problem for a radar versus a sonar? > > > > Hey Rick, > > > > Thanks for your thoughts. > > > >> I was never well versed in SONAR, but I do remember there are > >> dispersive effects that distort a pulse shape. One of the effects is > >> just reverb where some of the transmitted signal bounces back from the > >> medium itself. I would expect your filter might want to compensate > >> for the distortion, but that would need to be function of distance > >> which could be substituted for by time. > > > > Yeah, but even if you knew time/distance, how would you know what to > > compensate? > > > >> > >> I guess the question is, just how significant is the distortion? I > >> don't recall any of the sub systems using such a filter, but like I > >> said, I wasn't in that job long enough to learn much. > > > > Was it a sonar or radar? > > SONAR. I was thinking active. Otherwise what pulse would you be > working with? With active you know the time of flight and from that can > get a good value for range. I would expect the pulse distortion to be > fairly constant possibly only varying a bit with temperature. A lot of > SONAR word is relatively constant temperature because the sounds bounce > off thermal layers and travel almost like in a waveguide. > > Subs have their own assumptions about environments. What exactly is > your environment? > > Rereading your question, you are asking about distortion from the > target. I can't say I've ever considered that or seen where anyone else > did. But as I said, my experience was short and some 20+ years ago. > > -- > > Rick
Absorption/dispersion technology goes back a long way; e.g.: https://en.wikipedia.org/wiki/Anechoic_tile and pics: https://www.google.com/search?q=acoustic+tiles+on+russian+submarines&hl=en&gbv=2&prmd=ivns&tbm=isch&tbo=u&source=univ&sa=X&ved=0CBkQsARqFQoTCJXswJ3xvcgCFUh0PgodQz0Pow Kevin
On Mon, 12 Oct 2015 13:49:27 -0400, Randy Yates
<yates@digitalsignallabs.com> wrote:

>Consider the fundamental concept of a sonar (or radar) hich transmits a >pulse and is looking for the return. > >Are there targets which significantly distort the pulse shape, so that >using a simple correlation (matched filter) for the detector results in >degraded performance? > >In other words, it seems like what you REALLY want to use is a return, >or received, signal correlator, not a transmit signal correlator. Of course that >would require knowledge of the "target transfer function" H(w), so that >what you are correlating against is R(w) = T(w)H(w). > >Other than a difference in the possible expected H(w)'s, are there any >differences in this problem for a radar versus a sonar? > >Thoughts? >-- >Randy Yates >Digital Signal Labs >http://www.digitalsignallabs.com
Dunno, but I wonder if the field of medical ultrasound would be a good place to look for info. Seems like these issues must come up there all the time, and though I haven't looked I would guess there might be more publications available: Hot field, and no security restrictions. Best regards, Bob Masta DAQARTA v8.00 Data AcQuisition And Real-Time Analysis www.daqarta.com Scope, Spectrum, Spectrogram, Sound Level Meter Frequency Counter, Pitch Track, Pitch-to-MIDI FREE 8-channel Signal Generator, DaqMusiq generator Science with your sound card!
N0Spam@daqarta.com (Bob Masta) writes:

> On Mon, 12 Oct 2015 13:49:27 -0400, Randy Yates > <yates@digitalsignallabs.com> wrote: > >>Consider the fundamental concept of a sonar (or radar) hich transmits a >>pulse and is looking for the return. >> >>Are there targets which significantly distort the pulse shape, so that >>using a simple correlation (matched filter) for the detector results in >>degraded performance? >> >>In other words, it seems like what you REALLY want to use is a return, >>or received, signal correlator, not a transmit signal correlator. Of course that >>would require knowledge of the "target transfer function" H(w), so that >>what you are correlating against is R(w) = T(w)H(w). >> >>Other than a difference in the possible expected H(w)'s, are there any >>differences in this problem for a radar versus a sonar? >> >>Thoughts? >>-- >>Randy Yates >>Digital Signal Labs >>http://www.digitalsignallabs.com > > Dunno, but I wonder if the field of medical ultrasound would > be a good place to look for info. Seems like these issues > must come up there all the time, and though I haven't looked > I would guess there might be more publications available: > Hot field, and no security restrictions.
Thanks, Bob. As it turns out, this is an ultrasound application. I don't know much more - just looking at an abbreviated job description. -- Randy Yates Digital Signal Labs http://www.digitalsignallabs.com
On Tuesday, October 13, 2015 at 11:01:43 AM UTC-4, Randy Yates wrote:
> N0Spam@daqarta.com (Bob Masta) writes: > > > On Mon, 12 Oct 2015 13:49:27 -0400, Randy Yates > > <yates@digitalsignallabs.com> wrote: > > > >>Consider the fundamental concept of a sonar (or radar) hich transmits a > >>pulse and is looking for the return. > >> > >>Are there targets which significantly distort the pulse shape, so that > >>using a simple correlation (matched filter) for the detector results in > >>degraded performance? > >> > >>In other words, it seems like what you REALLY want to use is a return, > >>or received, signal correlator, not a transmit signal correlator. Of course that > >>would require knowledge of the "target transfer function" H(w), so that > >>what you are correlating against is R(w) = T(w)H(w). > >> > >>Other than a difference in the possible expected H(w)'s, are there any > >>differences in this problem for a radar versus a sonar? > >> > >>Thoughts? > >>-- > >>Randy Yates > >>Digital Signal Labs > >>http://www.digitalsignallabs.com > > > > Dunno, but I wonder if the field of medical ultrasound would > > be a good place to look for info. Seems like these issues > > must come up there all the time, and though I haven't looked > > I would guess there might be more publications available: > > Hot field, and no security restrictions. > > Thanks, Bob. As it turns out, this is an ultrasound application. I don't > know much more - just looking at an abbreviated job description. > -- > Randy Yates > Digital Signal Labs > http://www.digitalsignallabs.com
Ultrasound usually gets generated via piezoelectric transducers - those are high Q with narrow bandwidth around resonance frequency. You simply don't get much of w change. So this is not it. There are tons of other effects: target shape or texture, multi-path, clutter etc You need to provide more details to get some answers