Forums

multipath over sea for radar

Started by rahul1 January 29, 2006
hi 
i am trying to simulate a multipath environment over sea . am i correct in
assuming that my echo at low grazing angles will comprise of four paths (
combinations of the direct and surface bounced path). A lot of litearture
suggests that only two paths are sufficient where as a few talk about
using all the four. is there any staistical model available for
reference.

is there any method to extract the height of the reflecting surface when
MVDR is not an option?


thanks  



rahul1 wrote:
> hi > i am trying to simulate a multipath environment over sea . am i correct in > assuming that my echo at low grazing angles will comprise of four paths ( > combinations of the direct and surface bounced path). A lot of litearture > suggests that only two paths are sufficient where as a few talk about > using all the four.
Maybe a naive question, but what are the two additional paths? Rune
"rahul1" <skipperrahul@yahoo.com> wrote in message 
news:Qr6dnSpzB5RWXUHenZ2dnUVZ_tWdnZ2d@giganews.com...
> hi > i am trying to simulate a multipath environment over sea . am i correct in > assuming that my echo at low grazing angles will comprise of four paths ( > combinations of the direct and surface bounced path). A lot of litearture > suggests that only two paths are sufficient where as a few talk about > using all the four. is there any staistical model available for > reference. > > is there any method to extract the height of the reflecting surface when > MVDR is not an option? >
There are quite a few variables that might impact on the answers. First, for multipath modeling, there are two paths from the transmitter to the object and there are two paths fromt he object to the receiver. You need an aspect-dependent return model to determine the radar cross section apparent for each of the 4 "paths". Lacking that, you could model the object as if it were omnidirectional but that might be a stretch. Path 1: Direct illumination, direct return Path 2: Direct illumination, bounce path return Path 3: Bounce path illumination, direct return Path 4: Bounce path illumination, bounce path return You might model the direct path as a simple attenuation. You might model the bounce path as a simple attenuation or as a variable-length path due to time-varying surface conditions. Aspect dependency comes in two forms: A) If an object is illuminated at angle "a", then what is the radar cross section from that same angle? B) If an object is illuminated at angle "a", then waht is the apparent radar cross section as observed from angle "b"? The object itself may present a return structure that is highly composite and, thus, Rayleigh distributed (or some distribution) depending on the wavelength and the separation of reflectors in the object. The paths may add similarly - often do. Note that Path 1 is the shortest, Paths 2 and 3 are the same length and Path 4 is the longest. Are these resolvable? It's a simple matter to determine height from the path length difference if you can determine it. For example, Paths 1,2 and 3 are likely the strongest and you may be able to determine the temporal difference between them. This with a pulsed system with adequate temporal resolution. For a CW system and a relatively broadband waveform you might use Lloyd's mirror effect which is more commonly evident in passive sonar where there are only two paths. The idea is that the paths are nearly equal strength so look like a FIR filter with coefficients [1.0 1.0] (scaled) which is a comb filter - it has a sinusoidal frequency response. Measuring the frequency separation between frequency peak responses give you the time delay. Whether that works in an active system with the 4 paths is dependent on the object mostly I would think. The 4-path model probably looks like a 3-tap FIR filter with delay separations of 2D, 1D+1R, 2R (D=direct,R=reflected and D<R). Then the FIR filter relative delays would be zero, R-D,2*(R-D) so the unit delay is R-D, the time difference between the two physical paths - which is what you want. If you have lots of periods of the frequency response then the measurement might be fairly accurate. If you model this path structure as a FIR filter as [1 2 1] then the frequency amplitude response is a raised sinusoid and you may be able to determine the time delay from the distance between its response peaks. The assumption regarding the value of the center coefficient (for the sum of Paths 2 and 3) doesn't change the result that much. There is much dependent on the actual system parameters....... Fred
> >Maybe a naive question, but what are the two additional paths? > >Rune > >
hi rune sorry for not mentioning it in the beginning, the four paths are exactly what fred has mentioned . as fred has mentioned i need high temporal resolution to obtain the exact delay . this is a very difficult thing to have as at low grazing angles my delay will in all circumstances be negligible as compared to the bandwidth of my signal(in my opinion) as i am looking at delays of a couple of nanoseconds and if i am not wrong would require a sampling rate to atleast satisfy Mr Nyquist:-). thus it is implied that i am looking at a fading channel . the problem here is that, in case i use frequency diversity as a soultion to this the diversity will have to be suficiently large to overcome the frequency selctivity of the channel. also as fred mentioned the magnitude of these reflections are dependent on the aspect the target presents to the corresponding path. this is the problem i and a friend discussed the the idea here is to use the characteristics of the medium to estimate the position of the echo source over sea especially where there are no other major reflecting surfaces except for the sea surface.
rahul1 wrote:
> > > >Maybe a naive question, but what are the two additional paths? > > > >Rune > > > > > > hi rune > sorry for not mentioning it in the beginning, the four paths are exactly > what fred has mentioned . as fred has mentioned i need high temporal > resolution to obtain the exact delay . this is a very difficult thing to > have as at low grazing angles my delay will in all circumstances be > negligible as compared to the bandwidth of my signal(in my opinion) as i > am looking at delays of a couple of nanoseconds and if i am not wrong > would require a sampling rate to atleast satisfy Mr Nyquist:-). thus it is > implied that i am looking at a fading channel . > > the problem here is that, in case i use frequency diversity as a > soultion to this the diversity will have to be suficiently large to > overcome the frequency selctivity of the channel. also as fred mentioned > the magnitude of these reflections are dependent on the aspect the target > presents to the corresponding path. > this is the problem i and a friend discussed the the idea here is to use > the characteristics of the medium to estimate the position of the echo > source over sea especially where there are no other major reflecting > surfaces except for the sea surface.
Using the medium as a help, has been tried in sonar. In the sonar application you have changing sound speeds in the water, a generally unknown sea floor and surfeca waves on the sea surface. I am not aware of anybody who have managed to improve on sonar performance in these circumstances. In your case, the scenario seems to be a bit simpler. The speed of light is known and constant, and there is only one reflecting surface, the sea surface. However, the sea surface is highly dynamic, and is a strong reflector. I would not be surprised if you were able to do something smart that works in a dead calm sea. I would believe the problem would be to maintain that performance as the wave heights become larger, and you perhaps illuminate your target from an inadvantageous angle what the surface wave direction is concerned. I don't know how to do what you want, but I am pretty sure you will find the random scattering from the dynamic, rough sea surface to be the main problem you need to overcome. Rune