Penetrating Radar

David Kirkland wrote:
> Eric Jacobsen wrote:
> 
>> On 06 Jan 2005 19:04:32 -0500, Randy Yates
>> <randy.yates@sonyericsson.com> wrote:
>>
>>
>>> I've read somewhere that UWB-based Radar has been used by the military
>>> because of its ability to detect objects through walls. It's not
>>> intuitively obvious to me why a wideband system would have this
>>> capability. Can someone explain it?
>>
>>
>>
>> I think it's kinda practical.
>>
>> As others have mentioned, in order to resolve very close distances you
>> need LOTS of bandwidth.   The usual well-known radar analyses show
>> pretty easily that resolution in the range dimension is a function of
>> bandwidth, so if you want good resolution on small objects you need
>> buttloads of bandwidth (a technical term).
>>
>> There was a lot of noise being made about UWB about five years ago and
>> at the time I saw some interesting images made by commercial UWB
>> wall-imaging devices.   The idea was to find things like pipes and
>> wiring, but another obvious application is to find covert bugs and
>> things like that.
>>
>> Ground penetrating radars are also UWB for the same reason.
>>
>> I would think most decent radar texts would cover the analysis for why
>> resolution is a function of bandwidth.   It's the fundamental reason
>> why most radars chirp instead of pulse.   Most UWBs pulse instead of
>> chirp, though.  I think that's because it's hard to make a system that
>> sweeps that much bandwidth and still stays reasonably linear doing so.
>> The electronics are apparently much easier to build for pulses than
>> for sweeps that wide.  You don't need as much sophisticated signal
>> processing (which is really, really hard and requires hiring expensive
>> geniuses to figure out) for pulsed systems as you do for swept
>> systems.  Since the range is much shorter the power issues aren't as
>> bad.

Sweeps and short pulses are not the only options. Transmitting broadband 
noise is very effect for detailed target analysis and does not require 
highly paid people. At least I thought my pay was lousy when I did that 
stuff. :-)

>> Eric Jacobsen
>> Minister of Algorithms, Intel Corp.
>> My opinions may not be Intel's opinions.
>> http://www.ericjacobsen.org
> 
> 
> Yes, most of the radar texts do cover this, it's usually discussed under 
> "ambiguity function" analysis. In Radar you can often get away with an 
> assumption that the doppler shift is a frequency shifted version of your 
> pulse, unfortunately in Sonar systems you don't have that luxury.

Apart from sidelobe issues, I can't think of anywhere the doppler 
assumption would be measurably wrong.

> BTW - I'm neither highly paid nor a genius. Is there something I'm missing?

Intellect and cash? :-\ (Maybe you are just being modest about the 
intellect)

Regards,
Steve

Steve Underwood wrote:
> David Kirkland wrote:
> 
>> Eric Jacobsen wrote:
>>
>>> On 06 Jan 2005 19:04:32 -0500, Randy Yates
>>> <randy.yates@sonyericsson.com> wrote:
>>>
>>>
>>>> I've read somewhere that UWB-based Radar has been used by the military
>>>> because of its ability to detect objects through walls. It's not
>>>> intuitively obvious to me why a wideband system would have this
>>>> capability. Can someone explain it?
>>>
>>>
>>>
>>>
>>> I think it's kinda practical.
>>>
>>> As others have mentioned, in order to resolve very close distances you
>>> need LOTS of bandwidth.   The usual well-known radar analyses show
>>> pretty easily that resolution in the range dimension is a function of
>>> bandwidth, so if you want good resolution on small objects you need
>>> buttloads of bandwidth (a technical term).
>>>
>>> There was a lot of noise being made about UWB about five years ago and
>>> at the time I saw some interesting images made by commercial UWB
>>> wall-imaging devices.   The idea was to find things like pipes and
>>> wiring, but another obvious application is to find covert bugs and
>>> things like that.
>>>
>>> Ground penetrating radars are also UWB for the same reason.
>>>
>>> I would think most decent radar texts would cover the analysis for why
>>> resolution is a function of bandwidth.   It's the fundamental reason
>>> why most radars chirp instead of pulse.   Most UWBs pulse instead of
>>> chirp, though.  I think that's because it's hard to make a system that
>>> sweeps that much bandwidth and still stays reasonably linear doing so.
>>> The electronics are apparently much easier to build for pulses than
>>> for sweeps that wide.  You don't need as much sophisticated signal
>>> processing (which is really, really hard and requires hiring expensive
>>> geniuses to figure out) for pulsed systems as you do for swept
>>> systems.  Since the range is much shorter the power issues aren't as
>>> bad.
> 
> 
> Sweeps and short pulses are not the only options. Transmitting broadband 
> noise is very effect for detailed target analysis and does not require 
> highly paid people. At least I thought my pay was lousy when I did that 
> stuff. :-)
> 
>>> Eric Jacobsen
>>> Minister of Algorithms, Intel Corp.
>>> My opinions may not be Intel's opinions.
>>> http://www.ericjacobsen.org
>>
>>
>>
>> Yes, most of the radar texts do cover this, it's usually discussed 
>> under "ambiguity function" analysis. In Radar you can often get away 
>> with an assumption that the doppler shift is a frequency shifted 
>> version of your pulse, unfortunately in Sonar systems you don't have 
>> that luxury.
> 
> 
> Apart from sidelobe issues, I can't think of anywhere the doppler 
> assumption would be measurably wrong.

You should be able measure the change in pulse duration in an intercept 
receiver.  Not sure if it matters that much.  I've seen good derivations 
of the Doppler effect that use Minkowski diagrams which shows that 
Doppler is dilation of the time axis.  Are you using the term "dopler 
assumption" interchangeably with "narrow band?"
> 
>> BTW - I'm neither highly paid nor a genius. Is there something I'm 
>> missing?
> 
> 
> Intellect and cash? :-\ (Maybe you are just being modest about the 
> intellect)
> 
> Regards,
> Steve

"David Kirkland" <spam@netscape.net> wrote in message
news:b0vEd.9457$TN6.391685@news20.bellglobal.com...
> Peter K. wrote:
> > Eric Jacobsen wrote:
> >
> >
> >>you need buttloads of bandwidth (a technical term).
> >
> > Often abbreviated "bob".
> >
> > Ciao,
> >
> > Peter K.
>
> Great ... just great .... yet another acronym I have to memorize!! :)

Don't you mean yet another TLA?  :-)

-Jon

P.S. TLA = Three Letter Acronym

Stan Pawlukiewicz wrote:

> Steve Underwood wrote:
>
>> David Kirkland wrote:
>>
>>> Yes, most of the radar texts do cover this, it's usually discussed 
>>> under "ambiguity function" analysis. In Radar you can often get away 
>>> with an assumption that the doppler shift is a frequency shifted 
>>> version of your pulse, unfortunately in Sonar systems you don't have 
>>> that luxury.
>>
>>
>>
>> Apart from sidelobe issues, I can't think of anywhere the doppler 
>> assumption would be measurably wrong.
>
>
> You should be able measure the change in pulse duration in an 
> intercept receiver.  Not sure if it matters that much.  I've seen good 
> derivations of the Doppler effect that use Minkowski diagrams which 
> shows that Doppler is dilation of the time axis.  Are you using the 
> term "dopler assumption" interchangeably with "narrow band?"

Surely that is the expected effect of a doppler shift, so there is 
nothing strange there. When I said doppler assumption I was refering to 
the previous paragraph, by David. I believe what David is refering to is 
changes in the apparent doppler caused by the medium. In sonar this can 
be huge, and may flood any real doppler due to target motion. In radar 
it is generally rather hard to detect.

A look down radar suffers so much from ground clutter that you have 
little hope of detecting moving targets unless you DC filter in one form 
or another, and detect them by their doppler. Look up radars have 
relatively trivial problems. If the assumption that apparent doppler is 
solely due to target motion is ever wrong, look down radars become a 
nightmare (maybe even impractical) to design. Look down radars are 
usually on a fast moving platform, so you need to worry about doppler 
due to your own motion not being the same for the sidelobes as it is 
down the main boresight. For beams of any significant width you need to 
worry about this at the edges of the main beam too.

Regards,
Steve

:-\

-- 

r b-j                  rbj@audioimagination.com

"Imagination is more important than knowledge."

Steve Underwood wrote:
> Stan Pawlukiewicz wrote:
> 
>> Steve Underwood wrote:
>>
>>> David Kirkland wrote:
>>>
>>>> Yes, most of the radar texts do cover this, it's usually discussed 
>>>> under "ambiguity function" analysis. In Radar you can often get away 
>>>> with an assumption that the doppler shift is a frequency shifted 
>>>> version of your pulse, unfortunately in Sonar systems you don't have 
>>>> that luxury.
>>>
>>>
>>>
>>>
>>> Apart from sidelobe issues, I can't think of anywhere the doppler 
>>> assumption would be measurably wrong.
>>
>>
>>
>> You should be able measure the change in pulse duration in an 
>> intercept receiver.  Not sure if it matters that much.  I've seen good 
>> derivations of the Doppler effect that use Minkowski diagrams which 
>> shows that Doppler is dilation of the time axis.  Are you using the 
>> term "dopler assumption" interchangeably with "narrow band?"
> 
> 
> Surely that is the expected effect of a doppler shift, so there is 
> nothing strange there. When I said doppler assumption I was refering to 
> the previous paragraph, by David. I believe what David is refering to is 
> changes in the apparent doppler caused by the medium. In sonar this can 
> be huge, and may flood any real doppler due to target motion. In radar 
> it is generally rather hard to detect.
> 
> A look down radar suffers so much from ground clutter that you have 
> little hope of detecting moving targets unless you DC filter in one form 
> or another, and detect them by their doppler. Look up radars have 
> relatively trivial problems. If the assumption that apparent doppler is 
> solely due to target motion is ever wrong, look down radars become a 
> nightmare (maybe even impractical) to design. Look down radars are 
> usually on a fast moving platform, so you need to worry about doppler 
> due to your own motion not being the same for the sidelobes as it is 
> down the main boresight. For beams of any significant width you need to 
> worry about this at the edges of the main beam too.
> 
> Regards,
> Steve

So you are using the term "doppler" and "narrow band" interchangeably. I 
mentioned an intercept receiver because in Sonar the pulse return is 
usually stretched by the target not being a point radiator.  A intercept 
receiver will typically get a clean direct path signal, ar at least a 
leading edge, before any forward scattered reverberation and multi paths.

Stan Pawlukiewicz wrote:
> Steve Underwood wrote:
> 
>> Stan Pawlukiewicz wrote:
>>
>>> Steve Underwood wrote:
>>>
>>>> David Kirkland wrote:
>>>>
>>>>> Yes, most of the radar texts do cover this, it's usually discussed 
>>>>> under "ambiguity function" analysis. In Radar you can often get 
>>>>> away with an assumption that the doppler shift is a frequency 
>>>>> shifted version of your pulse, unfortunately in Sonar systems you 
>>>>> don't have that luxury.
>>>>
>>>>
>>>>
>>>>
>>>>
>>>> Apart from sidelobe issues, I can't think of anywhere the doppler 
>>>> assumption would be measurably wrong.
>>>
>>>
>>>
>>>
>>> You should be able measure the change in pulse duration in an 
>>> intercept receiver.  Not sure if it matters that much.  I've seen 
>>> good derivations of the Doppler effect that use Minkowski diagrams 
>>> which shows that Doppler is dilation of the time axis.  Are you using 
>>> the term "dopler assumption" interchangeably with "narrow band?"
>>
>>
>>
>> Surely that is the expected effect of a doppler shift, so there is 
>> nothing strange there. When I said doppler assumption I was refering 
>> to the previous paragraph, by David. I believe what David is refering 
>> to is changes in the apparent doppler caused by the medium. In sonar 
>> this can be huge, and may flood any real doppler due to target motion. 
>> In radar it is generally rather hard to detect.
>>
>> A look down radar suffers so much from ground clutter that you have 
>> little hope of detecting moving targets unless you DC filter in one 
>> form or another, and detect them by their doppler. Look up radars have 
>> relatively trivial problems. If the assumption that apparent doppler 
>> is solely due to target motion is ever wrong, look down radars become 
>> a nightmare (maybe even impractical) to design. Look down radars are 
>> usually on a fast moving platform, so you need to worry about doppler 
>> due to your own motion not being the same for the sidelobes as it is 
>> down the main boresight. For beams of any significant width you need 
>> to worry about this at the edges of the main beam too.
>>
>> Regards,
>> Steve
> 
> 
> So you are using the term "doppler" and "narrow band" interchangeably. I 
> mentioned an intercept receiver because in Sonar the pulse return is 
> usually stretched by the target not being a point radiator.  A intercept 
> receiver will typically get a clean direct path signal, ar at least a 
> leading edge, before any forward scattered reverberation and multi paths.

I'm sorry. I don't follow why you say I am using doppler and narrow band 
interchangeably. If I transmit broad band noise, what I get back from 
the target is subject to the same doppler shift as when I transmit a 
narrow band signal. Doppler is doppler. Bandwidth doesn't come into it. 
As with sonar, radar target returns splurge in time, due to the size of 
the target. You don't usually get reverberation, though, and multipath 
issues are insignificant compared to sonar. I think some spreading in 
time will always occur. No arbitrarily shaped object is going to be a 
clean reflector of any kind of energy.

Are we talking from different angles because in radar the doppler has 
only a tiny effect on overall bandwidth, and in sonar it is large?

Regards,
Steve

Steve Underwood wrote:
> Stan Pawlukiewicz wrote:
> 
>> Steve Underwood wrote:
>>
>>> Stan Pawlukiewicz wrote:
>>>
>>>> Steve Underwood wrote:
>>>>
>>>>> David Kirkland wrote:
>>>>>
>>>>>> Yes, most of the radar texts do cover this, it's usually discussed 
>>>>>> under "ambiguity function" analysis. In Radar you can often get 
>>>>>> away with an assumption that the doppler shift is a frequency 
>>>>>> shifted version of your pulse, unfortunately in Sonar systems you 
>>>>>> don't have that luxury.
>>>>>
>>>>>
>>>>>
>>>>>
>>>>>
>>>>>
>>>>> Apart from sidelobe issues, I can't think of anywhere the doppler 
>>>>> assumption would be measurably wrong.
>>>>
>>>>
>>>>
>>>>
>>>>
>>>> You should be able measure the change in pulse duration in an 
>>>> intercept receiver.  Not sure if it matters that much.  I've seen 
>>>> good derivations of the Doppler effect that use Minkowski diagrams 
>>>> which shows that Doppler is dilation of the time axis.  Are you 
>>>> using the term "dopler assumption" interchangeably with "narrow band?"
>>>
>>>
>>>
>>>
>>> Surely that is the expected effect of a doppler shift, so there is 
>>> nothing strange there. When I said doppler assumption I was refering 
>>> to the previous paragraph, by David. I believe what David is refering 
>>> to is changes in the apparent doppler caused by the medium. In sonar 
>>> this can be huge, and may flood any real doppler due to target 
>>> motion. In radar it is generally rather hard to detect.
>>>
>>> A look down radar suffers so much from ground clutter that you have 
>>> little hope of detecting moving targets unless you DC filter in one 
>>> form or another, and detect them by their doppler. Look up radars 
>>> have relatively trivial problems. If the assumption that apparent 
>>> doppler is solely due to target motion is ever wrong, look down 
>>> radars become a nightmare (maybe even impractical) to design. Look 
>>> down radars are usually on a fast moving platform, so you need to 
>>> worry about doppler due to your own motion not being the same for the 
>>> sidelobes as it is down the main boresight. For beams of any 
>>> significant width you need to worry about this at the edges of the 
>>> main beam too.
>>>
>>> Regards,
>>> Steve
>>
>>
>>
>> So you are using the term "doppler" and "narrow band" interchangeably. 
>> I mentioned an intercept receiver because in Sonar the pulse return is 
>> usually stretched by the target not being a point radiator.  A 
>> intercept receiver will typically get a clean direct path signal, ar 
>> at least a leading edge, before any forward scattered reverberation 
>> and multi paths.
> 
> 
> I'm sorry. I don't follow why you say I am using doppler and narrow band 
> interchangeably. If I transmit broad band noise, what I get back from 
> the target is subject to the same doppler shift as when I transmit a 
> narrow band signal. Doppler is doppler. Bandwidth doesn't come into it. 
> As with sonar, radar target returns splurge in time, due to the size of 
> the target. You don't usually get reverberation, though, and multipath 
> issues are insignificant compared to sonar. I think some spreading in 
> time will always occur. No arbitrarily shaped object is going to be a 
> clean reflector of any kind of energy.
> 
> Are we talking from different angles because in radar the doppler has 
> only a tiny effect on overall bandwidth, and in sonar it is large?

Not large, but measurable in the sense that one can observe it in some 
restricted circumstances if one is motivated to look for it. I don't 
think it has any impact on how people design systems.
> 
> Regards,
> Steve

Stan Pawlukiewicz wrote:
(snip)

>> I'm sorry. I don't follow why you say I am using doppler and narrow 
>> band interchangeably. If I transmit broad band noise, what I get back 
>> from the target is subject to the same doppler shift as when I 
>> transmit a narrow band signal. Doppler is doppler. Bandwidth doesn't 
>> come into it. As with sonar, radar target returns splurge in time, due 
>> to the size of the target. You don't usually get reverberation, 
>> though, and multipath issues are insignificant compared to sonar. I 
>> think some spreading in time will always occur. No arbitrarily shaped 
>> object is going to be a clean reflector of any kind of energy.

>> Are we talking from different angles because in radar the doppler has 
>> only a tiny effect on overall bandwidth, and in sonar it is large?

> Not large, but measurable in the sense that one can observe it in some 
> restricted circumstances if one is motivated to look for it. I don't 
> think it has any impact on how people design systems.

In http://www.uwbgroup.ru/eng/ourworks/uwbradar_2/chapt6.htm
they detect a person breathing through a wall.  You can almost stop
moving, but you can't stop breathing.

In http://www.uwbgroup.ru/eng/ourworks/uwbradar_2/chapt2.htm
they detect a heartbeat, though maybe not through a wall.

-- glen

Stan Pawlukiewicz wrote:
> Steve Underwood wrote:
> 
>> Stan Pawlukiewicz wrote:
>>
>>> Steve Underwood wrote:
>>>
>>>> Stan Pawlukiewicz wrote:
>>>>
>>>>> Steve Underwood wrote:
>>>>>
>>>>>> David Kirkland wrote:
>>>>>>
>>>>>>> Yes, most of the radar texts do cover this, it's usually 
>>>>>>> discussed under "ambiguity function" analysis. In Radar you can 
>>>>>>> often get away with an assumption that the doppler shift is a 
>>>>>>> frequency shifted version of your pulse, unfortunately in Sonar 
>>>>>>> systems you don't have that luxury.
>>>>>>
>>>>>>
>>>>>>
>>>>>>
>>>>>>
>>>>>>
>>>>>>
>>>>>> Apart from sidelobe issues, I can't think of anywhere the doppler 
>>>>>> assumption would be measurably wrong.
>>>>>
>>>>>
>>>>>
>>>>>
>>>>>
>>>>>
>>>>> You should be able measure the change in pulse duration in an 
>>>>> intercept receiver.  Not sure if it matters that much.  I've seen 
>>>>> good derivations of the Doppler effect that use Minkowski diagrams 
>>>>> which shows that Doppler is dilation of the time axis.  Are you 
>>>>> using the term "dopler assumption" interchangeably with "narrow band?"
>>>>
>>>>
>>>>
>>>>
>>>>
>>>> Surely that is the expected effect of a doppler shift, so there is 
>>>> nothing strange there. When I said doppler assumption I was refering 
>>>> to the previous paragraph, by David. I believe what David is 
>>>> refering to is changes in the apparent doppler caused by the medium. 
>>>> In sonar this can be huge, and may flood any real doppler due to 
>>>> target motion. In radar it is generally rather hard to detect.
>>>>
>>>> A look down radar suffers so much from ground clutter that you have 
>>>> little hope of detecting moving targets unless you DC filter in one 
>>>> form or another, and detect them by their doppler. Look up radars 
>>>> have relatively trivial problems. If the assumption that apparent 
>>>> doppler is solely due to target motion is ever wrong, look down 
>>>> radars become a nightmare (maybe even impractical) to design. Look 
>>>> down radars are usually on a fast moving platform, so you need to 
>>>> worry about doppler due to your own motion not being the same for 
>>>> the sidelobes as it is down the main boresight. For beams of any 
>>>> significant width you need to worry about this at the edges of the 
>>>> main beam too.
>>>>
>>>> Regards,
>>>> Steve
>>>
>>>
>>>
>>>
>>> So you are using the term "doppler" and "narrow band" 
>>> interchangeably. I mentioned an intercept receiver because in Sonar 
>>> the pulse return is usually stretched by the target not being a point 
>>> radiator.  A intercept receiver will typically get a clean direct 
>>> path signal, ar at least a leading edge, before any forward scattered 
>>> reverberation and multi paths.
>>
>>
>>
>> I'm sorry. I don't follow why you say I am using doppler and narrow 
>> band interchangeably. If I transmit broad band noise, what I get back 
>> from the target is subject to the same doppler shift as when I 
>> transmit a narrow band signal. Doppler is doppler. Bandwidth doesn't 
>> come into it. As with sonar, radar target returns splurge in time, due 
>> to the size of the target. You don't usually get reverberation, 
>> though, and multipath issues are insignificant compared to sonar. I 
>> think some spreading in time will always occur. No arbitrarily shaped 
>> object is going to be a clean reflector of any kind of energy.
>>
>> Are we talking from different angles because in radar the doppler has 
>> only a tiny effect on overall bandwidth, and in sonar it is large?
> 
> 
> Not large, but measurable in the sense that one can observe it in some 
> restricted circumstances if one is motivated to look for it. I don't 
> think it has any impact on how people design systems.
> 
>>
>> Regards,
>> Steve

In the Radar texts the assumption about Doppler being a frequency shift 
is often called the NB assumption.

In Sonar with CW pulse signals the received frequency bandwidth can be 
extremely large - so much so that you need several doppler shift 
replicas for your matched filter (alternatively you can sometime use an 
FFT - it depends on the Time Bandwidth product).

The NB assumption isn't based on the bandwidth of the signal you are 
transmitting but rather the receive bandwidth (doppler spread) wrt to 
your centre frequency. Just to remind those who forget that a common 
speed of sound in water is 1500 m/s, thus relatively small velocities
can produce a large frequency spread.

Hope that helps.

Cheers,
David