"ComDart" vhf/uhf signal geolocation

jonwienke

More Info Coming Soon!
Premium Subscriber
Joined
Jul 18, 2014
Messages
12,974
Location
VA
The "single antenna" bit can't possibly correct. You need at least 4 antennas to calculate separate x and y vectors to pinpoint a location on the ground.
 

dlwtrunked

Member
Premium Subscriber
Joined
Dec 19, 2002
Messages
911
"You need at least 4 antennas to calculate separate x and y vectors to pinpoint a location on the ground" is not quite right (Note I am both a Ph.D mathematician and a consultant to a company doing similar DF product development). Using 4 antennas does not give you location but gives you a direction by simulating rotation--one can do similar by rotating two antennas, but with 4 antennas, you can leave them fixed and electronically rotate and use the Doppler shift. A second remark also, one needs to precisely define what one means by an "antenna". One can call an antenna something the size of a hockey puck that if one takes it apart actually has more than one--say 4--receiving elements. But the article is wrong in that the device only gives a direction. It is meant to be on a moving platform and getting more directions from the motion at different locations as it moves lets one triangulate (cross the direction lines to get a position). Look at the similar RRT (Fredericksburg, VA) THiEF system on the web site below to see why you would call this called "an antenna" even though it has actually separate receiving parts inside: https://radiorecon.com/application/...Public_THiEF_Technical_Data_Sheet_4-18-19.pdf
 
Last edited:

jonwienke

More Info Coming Soon!
Premium Subscriber
Joined
Jul 18, 2014
Messages
12,974
Location
VA
"You need at least 4 antennas to calculate separate x and y vectors to pinpoint a location on the ground" is not quite right (Note I am both a Ph.D mathematician and a consultant to a company doing similar DF product development). Using 4 antennas does not give you location but gives you a direction by simulating rotation--one can do similar by rotating two antennas, but with 4 antennas, you can leave them fixed and electronically rotate and use the Doppler shift.
You don't need to do the simulated rotation/Doppler analysis if you have a separate SDR on each antenna, and do a phase comparison on the signals to calculate your vector. If you have 4 antennas arranged in a tetrahedron, you can calculate both elevation and azimuth angles, and given sufficient altitude, you can locate the RF source on the ground pretty accurately. As long as the SDRs are all driven by common RF and IF oscillators with uniform propagation delays between the oscillator and mixer/ADC, the math to make it work isn't that hard. Take the phase delay between antennas, the physical distance between them, and do a bit of trig to calculate an angle.

The math is simpler if you have 6 antennas in a cubic array instead of 4 in a tetrahedronal array; you calculate separate angles for your X, Y, and Z antenna pairs, and derive your final vector from those.
 

dlwtrunked

Member
Premium Subscriber
Joined
Dec 19, 2002
Messages
911
You don't need to do the simulated rotation/Doppler analysis if you have a separate SDR on each antenna, and do a phase comparison on the signals to calculate your vector. If you have 4 antennas arranged in a tetrahedron, you can calculate both elevation and azimuth angles, and given sufficient altitude, you can locate the RF source on the ground pretty accurately. As long as the SDRs are all driven by common RF and IF oscillators with uniform propagation delays between the oscillator and mixer/ADC, the math to make it work isn't that hard. Take the phase delay between antennas, the physical distance between them, and do a bit of trig to calculate an angle.

The math is simpler if you have 6 antennas in a cubic array instead of 4 in a tetrahedronal array; you calculate separate angles for your X, Y, and Z antenna pairs, and derive your final vector from those.
What you saw simply will not work for any distance from the transmitter unless the antennas are farther apart than the systems in the links where we are confined to a foot or so. I now understand that is why you disputed them from the point of few of a different larger system but even this accuracy becomes an issue at distance. But their system is most certainly similar (seeing the photos) to the one in my link and uses changing position to get a triangulation from bearings rather than a fix from a single position. (I have been deep in those calculations.) It is still difficult, particularly in urban settings, to get accurate positioning due to multipath without doing some tricks.
 

jonwienke

More Info Coming Soon!
Premium Subscriber
Joined
Jul 18, 2014
Messages
12,974
Location
VA
What you saw simply will not work for any distance from the transmitter unless the antennas are farther apart than the systems in the links where we are confined to a foot or so. I now understand that is why you disputed them from the point of few of a different larger system but even this accuracy becomes an issue at distance. But their system is most certainly similar (seeing the photos) to the one in my link and uses changing position to get a triangulation from bearings rather than a fix from a single position. (I have been deep in those calculations.) It is still difficult, particularly in urban settings, to get accurate positioning due to multipath without doing some tricks.
You're wrong about pretty much everything there. The whole point of this system is that it does NOT require receiving transmissions from multiple locations to calculate the location of the transmitter. It can calculate the location of the transmitter in less than a second from a single location, receiving a single transmission. So it's not doing a simple Doppler rotation to generate an azimuth angle like the system you linked.

The math used here is the same as what's used to calculate a GPS fix, except inverted so you're using multiple receivers whose locations are known to calculate the position of a transmitter whose location is unknown. And it's simplified because you don't have to compensate tor the movement of multiple satellites in 3 dimensions--the locations of the receiving antennas are fixed relative to each other. And distance is NOT an issue. The math only fails when the transmitter is close to the receiver antenna array, or inside it.

Document034.jpg

Pardon my crappy drawing skills, but this is how it works, simplified to one axis. Tx is the location of the transmitter, and R1 and R2 are two phase-locked receivers both running off the same RF/IF oscillators, and separated by distance D. R1 and R2 digitize the RF signal from Tx at the highest sampling rate possible. Pd is calculated by calculating the phase difference between R1 and R2, then converting the phase difference to a distance. Once you have Pd, you can say P2 = P1 + Pd, and a little bit of trig involving P1, P2, and D will give you an accurate angle between line segments P and D, as long as P is "large" compared to D. Using the altitude of the drone as your P value would be pretty reasonable.

The greater the sampling rate of R1 and R2, and the greater the distance D, the more accurately you can calculate the bearing angle betweeen D and P.

Add an additional pair of receivers at right angles to the first, and you can calculate X and Y angles separately, and calculate the transmitter position on the ground with a single transmission and one receiver array, if the array is airborne and the transmitter is inside about a 120-degree cone underneath the drone.
 

dlwtrunked

Member
Premium Subscriber
Joined
Dec 19, 2002
Messages
911
I will not argue with you on this. I am a PhD mathematician who works in this exact area of hardware and the software algorithms to do RF direction finding like this from vehicles, ground, drones, etc. Your drawing is the simple first diagram but ignores a real problem-which is sometimes called "cycle slips"--roughly this means you cannot recover the number of cycle differences (or whole wavelengths(--phase alone cannot do that. If it did, things like extremely high accuracy (sub-inch) GPS would be trivial.
 
Last edited:

jonwienke

More Info Coming Soon!
Premium Subscriber
Joined
Jul 18, 2014
Messages
12,974
Location
VA
I will not argue with you on this. I am a PhD mathematician who works in this exact area of hardware and the software algorithms to do RF direction finding like this from vehicles, ground, drones, etc. Your drawing is the simple first diagram but ignores a real problem-which is sometimes called "cycle slips"--roughly this means you cannot recover the number of cycle differences (or whole wavelengths(--phase alone cannot do that. If it did, things like extremely high accuracy (sub-inch) GPS would be trivial.
ShotSpotter uses similar algorithms, just applied to audio waveforms instead of RF. ShotSpotter solves the multi-cycle problem by looking at the impulse at the beginning of the waveform, and using that to figure out the number of complete cycles between data streams. If the RF waveform has a clearly identifiable starting point, or there is any amplitude or frequency change that can be unambiguously correlated across the data streams from the separate receivers, that does in fact allow a multic-cycle phase difference to be measured. That's easier to do when the RF waveform being targeted is being turned on and off, vice a constant broadcast.

Your point about multipath has some merit, but that is much less of an issue in the context of a ground-based transmitter being surveilled by an airborne receiver array.

And the limiting factor in GPS accuracy is atmospheric distortion unevenly refracting the GPS signal (the same effect that causes stars to "twinkle").

Just curious, are you asserting that this claim is false?
“As soon as someone presses the push-to-talk button and utters a few words, we immediately know the location.” ComDart also eavesdrops like other systems, he added, but “the immediate location track is the breakthrough,” Dulberg explained.

If so, on what basis? If not, then how is that capability accomplished, if not by the means I describe?
 

dlwtrunked

Member
Premium Subscriber
Joined
Dec 19, 2002
Messages
911
ShotSpotter uses similar algorithms, just applied to audio waveforms instead of RF. ShotSpotter solves the multi-cycle problem by looking at the impulse at the beginning of the waveform, and using that to figure out the number of complete cycles between data streams. If the RF waveform has a clearly identifiable starting point, or there is any amplitude or frequency change that can be unambiguously correlated across the data streams from the separate receivers, that does in fact allow a multic-cycle phase difference to be measured. That's easier to do when the RF waveform being targeted is being turned on and off, vice a constant broadcast.

Your point about multipath has some merit, but that is much less of an issue in the context of a ground-based transmitter being surveilled by an airborne receiver array.

And the limiting factor in GPS accuracy is atmospheric distortion unevenly refracting the GPS signal (the same effect that causes stars to "twinkle").

Just curious, are you asserting that this claim is false?
“As soon as someone presses the push-to-talk button and utters a few words, we immediately know the location.” ComDart also eavesdrops like other systems, he added, but “the immediate location track is the breakthrough,” Dulberg explained.

If so, on what basis? If not, then how is that capability accomplished, if not by the means I describe?
"Your point about multipath has some merit, but that is much less of an issue in the context of a ground-based transmitter being surveilled by an airborne receiver array."

True of course. Problems there center around weight etc.

"As soon as someone presses the push-to-talk button and utters a few words, we immediately know the location.” ComDart also eavesdrops like other systems, he added, but “the immediate location track is the breakthrough,” Dulberg explained."

Skeptical. It is not clear to me that they are claiming a single unit can do that. Also, at what range. In any case, we both agree a single antenna cannot do such magic. Such advertising often takes a lot of "liberties".

Regarding GPS
"And the limiting factor in GPS accuracy is atmospheric distortion unevenly refracting the GPS signal (the same effect that causes stars to "twinkle")."

It (atmospheric delay and scintillation) is not the only limiting factor, or rather the modeling of atmosphere, is not the only limiting factor. Another is the satellite geometry or "dilution of precision". In any case, processing using correction data from a nearby known location over time takes care of things. Before the DF work for a company, I worked for the U.S. government (an Army Lab) where at times I worked on GPS and gyro position finding. As a hobby, I have survey grade GPS receivers that I experiment with doing things like CORS carrier-phase post-procesing--just recent acquired some new uBlox zed-F9P to experiment with doing that. Back in the 1990's when I first looked at GPS (before working for the government), I had a web page on GPS accuracy which I found out is still sometime referenced even though it has not existed for years. This was when people asked me "Why does anyone need to know position?-just use a map" Sort of like when in 1993 when I was asked "Why does anyone want the INTERNET thing?" by a college computer science department chair.
 

jonwienke

More Info Coming Soon!
Premium Subscriber
Joined
Jul 18, 2014
Messages
12,974
Location
VA
The ComDart’s ability to immediately locate enemy communication transmissions from a single platform represented a breakthrough capability, said Adi Dulberg, vice-president general manager of IAI-Elta’s Intelligence, Communications and Electronic Warfare (EW) Division.

Standard communications intelligence (COMINT) systems usually install an antenna array on a platform, link the array to a receiver and use it to evaluate the direction of the spread of the wavelength, providing an “azimuth axis of where the emitter source is coming from,” Dulberg told Janes.

The main drawback of that approach is that it “only provides direction, not the precise location of the emitter. The other technique is to use a number of platforms and sensors in a single area cell,” according to Dulberg, who suggested that the tactic can be cumbersome and is rarely implemented.

“ComDart is a highly compact system using just a single antenna,” he said. “As soon as someone presses the push-to-talk button and utters a few words, we immediately know the location.” ComDart also eavesdrops like other systems, he added, but “the immediate location track is the breakthrough,” Dulberg explained.


It's pretty clear that they are in fact claiming to be able to use a single device to locate a transmitter from a single transmission. The only way to do that is to use an airborne platform, and use some version of the process I described to calculate at least x and y bearing angles between the drone and the transmitter, and plot the transmitter location as where those angles intersect the ground. And as I noted before, increasing the range makes the angle calculations more accurate. So your objection re range is invalid. The limiting factor is signal strength.
 

cg

Member
Premium Subscriber
Joined
Dec 13, 2000
Messages
3,804
Location
Connecticut
The products own website:
"Geo-location is achieved through a single, small form factor, Vector Sensor Antenna (VSA). The VSA design is proprietary technology developed solely by ELTA - the antenna’s construction enables concurrent measurement of elevation and azimuth from received RF waves. While other technologies require multiple antennas on one or more platforms for geolocation, ELTA's solution delivers instantaneous geo-location from a single antenna, on a single platform."
 

dlwtrunked

Member
Premium Subscriber
Joined
Dec 19, 2002
Messages
911
The products own website:
"Geo-location is achieved through a single, small form factor, Vector Sensor Antenna (VSA). The VSA design is proprietary technology developed solely by ELTA - the antenna’s construction enables concurrent measurement of elevation and azimuth from received RF waves. While other technologies require multiple antennas on one or more platforms for geolocation, ELTA's solution delivers instantaneous geo-location from a single antenna, on a single platform."
Good link which I had missed. I am aware of Vector Sensor Antenna's but the saying "one antenna" in the literature is misleading. VSA's when closely looked at in reality are actually more than one electronic antenna (typically being 3 electric dipoles and 3 magnet dipoles). Although a VSA is referred to as one VSA antenna, measuring the multiple field components of the EM wave essentially use what are separate antennas in the usual sense. I still think it actually has to do it with their system in motion. Although they say "one transmission", that does not exclude doing more than one measurement as the platform with their device moves. A former student (35 years ago) of mine works at iai, perhaps I can get information from her. A good IEEE reference is a paper by Lincoln Labs at Direction-Finding Results for a Vector Sensor Antenna on a Small UAV
 

jonwienke

More Info Coming Soon!
Premium Subscriber
Joined
Jul 18, 2014
Messages
12,974
Location
VA
Combining the antennas into an integrated small-form package eliminates the multi-cycle conundrum, at least for UHF and below. It demands a higher sampling rate for digitizing the RF, though, to get sufficiently accurate phase comparisons to get accurate vector angles.
 

Token

Member
Premium Subscriber
Joined
Jun 18, 2010
Messages
2,161
Location
Mojave Desert, California, USA
To sum up:

The only reason this works is because it is an airborne platform, presumably looking for a ground, or near ground, based transmitter.

The "single antenna" is actually a single array, multiple electronic antennas in a single, one assumes compact, package. The VSA.

The system achieves an accurate AOA in 2 axis, call them elevation and azimuth. The point in space of the sensing platform (the UAV) is known. The system geolocates by drawing a line from the position of the platform along the indicated elevation and azimuth and plotting this line on a high detail, 3 D, map. Where the line intersects the surface of the ground is the transmitter location.

Of course, this will not work for any transmitter not at or near ground level, such as another UAV or aircraft.

T!
 

jonwienke

More Info Coming Soon!
Premium Subscriber
Joined
Jul 18, 2014
Messages
12,974
Location
VA
Correct. And the closer the X/Y angles are to horizontal, the less accurate the location plot will be with regard to distance from the vehicle. As the transmitter nears the horizon, the system gradually devolves to calculating a compass heading only.

If the antenna array can calculate X, Y, and Z angles, the system can calculate both a compass azimuth and an elevation angle to distinguish between ground and airborne transmitters.
 
Top