84 seconds to locate transmitter with the KrakenSDR

Ubbe

Member
Joined
Sep 8, 2006
Messages
8,944
Location
Stockholm, Sweden
Those elements allow not only finding the direction, but also the DISTANCE to the emitter via DTOA (Differential Time Of Arrival) of the signal. That allows a single Kraken unit to quite accurately locate an emitter.
To PA8W's defense I where also confused by your incorrect statement. A stationary Kraken system can only detect the direction to the target, and it does that by using DTOA instead of the more common doppler technique. You will either need another Kraken user you can connect to thru internet or you will have to move your own system to the side from the target direction to get another angle to it. Then you can triangulate, like with any other direction finding system, and calculate how far away the target are. The Kraken application will make that calculation for you and plot on a map.

/Ubbe
 

popnokick

Member
Premium Subscriber
Joined
Mar 21, 2004
Messages
2,837
Location
Northeast PA
It's not my intention to mislead anyone, so please help me to understand my earlier "incorrect statement".

Per the info on the KrakenSDR website KrakenSDR CrowdSupply describes the system as having RTL-SDR receivers and using five separate receiving channels simultaneously. Are the following correct statements, or is CrowdSupply misrepresenting the features and abilities of the KrakenSDR system? -
- Is the (now "automatic") calibration and coherence synchronization determining the distance between the receiver antennas (among other things)?
- To locate emitters via TDOA wouldn't KrakenSDR use one or more of the following algorithms and calculations?
Object Tracking Using Time Difference of Arrival (TDOA) - MATLAB & Simulink
- Are the examples of tracking shown on the CrowdSupply website omitting the fact that multiple KrakenSDR system units were used to perform location? Or did they use a single KrakenSDR unit?
Thank you for any clarifications that might be necessary.
 

Token

Member
Joined
Jun 18, 2010
Messages
2,379
Location
Mojave Desert, California, USA
You're missing a key piece of the understanding of the KrakenSDR information. Unlike legacy RDF systems that use a single antenna and radio to find only the direction of the emitter, the Kraken has multiple antennas and radios working together with a microprocessor. Those elements allow not only finding the direction, but also the DISTANCE to the emitter via DTOA (Differential Time Of Arrival) of the signal. That allows a single Kraken unit to quite accurately locate an emitter. If you have more than one Kraken unit (as in legacy RDF) then the accuracy is even greater. That is how a single Kraken located an emitter in 84 seconds.

Errr...no.

The Kraken does not calculate distance by DTOA (or TDOA, or multilateration).

The Kraken measures the phase of arrival of a signal on each antenna element. Knowing the phase on each element, and having the element configuration and size as an input value (and so element spacing), it can then calculate angle or direction of arrival (DOA) from that phase. Of course, it can be argued that phase is time, and so the Kraken does determine (if not measure) DTOA, but the important part is not what it is measuring, but rather what it is calculating via that measurement. It calculates direction of arrival, not distance.

The base, the distance between elements, is too small to measure distance via TDOA at any meaningful distance.

If the Kraken never moves that is all you get, angle or direction of arrival. But if the Kraken moves you now create a synthetic aperture. At location A the Kraken has a DOA. The software plots that DOA on a map (or in 2D space). A few seconds later, at location B, it takes another DOA, since the platform is in motion this results in different DOA, and the cuts intersect someplace. A few seconds later, at location C, it takes another, and another, etc.

Each of these cuts is from a different location (the platform is in motion, yes?) It is like having multiple RDF sources reporting direction cuts.

The software then continually calculates the intersection point of each of the cuts, refining the location of the emitter. To the best of my knowledge, it only does this when using the App, and having access to the GPS location of the device the App is on. If it can do this from another GPS source I have not figured out how yet.

The first thing I did with my Krakken was to find out who it was in my neighborhood that had the temperature sensors pinging on 434 MHz. The second thing was to track down a jammer hitting a local GMRS freq. Slick, easy, and relatively accurate with minimal effort.

T!
 

popnokick

Member
Premium Subscriber
Joined
Mar 21, 2004
Messages
2,837
Location
Northeast PA
So by moving the Kraken you've created multiple locations ("... a few seconds later, at location B") then a few seconds later at location C, etc. you've created not only a directional plot but a calculated distance to the emitter.... from a single Kraken unit. How far do you have to move the Kraken unit in those "few seconds" to get enough data for direction (and distance)?
 

Token

Member
Joined
Jun 18, 2010
Messages
2,379
Location
Mojave Desert, California, USA
So by moving the Kraken you've created multiple locations ("... a few seconds later, at location B") then a few seconds later at location C, etc. you've created not only a directional plot but a calculated distance to the emitter.... from a single Kraken unit. How far do you have to move the Kraken unit in those "few seconds" to get enough data for direction (and distance)?

The Kraken, and Kraken interface running on the Pi, measures only direction or arrival, nothing more. The external software running on the App takes those instantaneous DOAs and plots them in space. The more angle cuts you get, the more refined the emitter location plot becomes, averaging out multipath issues.

The further you move the Kraken between cuts, or the more varied cuts you have, the more accurate the solution tends to be. And if you don't move, you will not get a location plot, as the cuts will not converge.

You may end up with a rough possible solution in just a few 10s of seconds of motion. But that solution is continually improved as you are in motion, refining its accuracy. Although you may have a rough location plot in seconds, getting a good, accurate, plots take minutes, not seconds.

T!
 

Token

Member
Joined
Jun 18, 2010
Messages
2,379
Location
Mojave Desert, California, USA
Here is a video of the Kraken doing basic DOA of a pulsed signal that is active about every 30 seconds. You see it reporting a direction of arrival, in this case between 123 and 132 degrees true, every time the signal transmits. These reported bearings would be the information used by the App (running on a separate smart phone or tablet) to plot and find a geolocation, if I were doing so. In this case I was only interested in determining the DOA, so I did not run the App and only used the web based GUI.


T!
 

gldavis

KE7MQF
Premium Subscriber
Joined
Oct 17, 2002
Messages
1,405
Location
Bountiful, UT
I have now taken this system out for a test drive 4 times, with mixed results. If you can drive to the xmitter location, then it does a good, quick job. (travel time)
If your transmitter is not local, or you can't "drive to it", then the results are mixed. I have driven around, attempting to "DF" a few sites that are/were at least 35~50+ klicks(Km), from where I was located/driving. Due to desert and government property, I may not be able to get closer.
20 degree resolution, at that distance, gives a large area of error.
Another issue was 3 different sites using the same control freq, witch added confusion to the equation. But I haven't given up yet. There are several "adjustments" that can be made to better the results.
 

Token

Member
Joined
Jun 18, 2010
Messages
2,379
Location
Mojave Desert, California, USA
I have now taken this system out for a test drive 4 times, with mixed results. If you can drive to the xmitter location, then it does a good, quick job. (travel time)
If your transmitter is not local, or you can't "drive to it", then the results are mixed. I have driven around, attempting to "DF" a few sites that are/were at least 35~50+ klicks(Km), from where I was located/driving. Due to desert and government property, I may not be able to get closer.
20 degree resolution, at that distance, gives a large area of error.
Another issue was 3 different sites using the same control freq, witch added confusion to the equation. But I haven't given up yet. There are several "adjustments" that can be made to better the results.

You can improve the resolution by tailoring the antenna elements and spacing to the frequency in question. Use the antenna array calculator to determine the largest array spacing you can use for that frequency (not to exceed 0.5 wavelength spacing on the target freq, I normally shoot for just under that) and the resolution gets as good as it is going to be. I often get well under 10 degrees, and sometimes down around 6 or 7 deg repeatable resolution.

When shooting for a signal you may not be able to get close to, don't drive directly towards the signal, but off to the side of the signal. And get free of anything around you to reduce multipath issues.

When working with a crowded frequency try setting the VFO B filter bandwidth very tight, and set the squelch level higher as you get near a target. If you can get one signal noticeably stronger than the others, for example getting closer to that source or putting terrain / obstructions between you and the other sources, than you can more easily break it out from the others.

DF is what it is, and there are variables and issues with it, no matter what system you use. The Kraken is not perfect, but does very well for its price point. But if you really want accuracy you need more complex, and expensive, systems. For example, maybe a distributed system that uses TDOA or something like that.

T!
 

gldavis

KE7MQF
Premium Subscriber
Joined
Oct 17, 2002
Messages
1,405
Location
Bountiful, UT
6~7 degrees ambiguity is still a large area, when listening 35+ clicks away. I have been moving parallel to the xmitter, when possible. The freq spectrum isn't to crowded, but I will try the filter tip. I have been trying different squelch/gain settings, to focus on the signal of interest.
RDF is a peculiar animal. It requires some knowledge and experimenting. In most cases, the KrakenSDR is a good method/system (and affordable), but as with another system, there are strong points and weak points. I just need to use it to learn what and how.
-side not, I am waiting on the cables for the Arrow Antenna and will see how that works.
 

gldavis

KE7MQF
Premium Subscriber
Joined
Oct 17, 2002
Messages
1,405
Location
Bountiful, UT
I've been using the KRAKENSDR system for a few months now. It is impressive at what it does. You do need to have an understanding of what RDF can and cannot do. And definitely need to become familiar with how this works. From a fixed location, it can only give a bearing. After moving around, it can/will give an approximate location. This location is refined as it gets more "hits".
 

PA8W

Newbie
Joined
Nov 14, 2017
Messages
17
Thanks for your user insight. I think the overall conclusion should be that the title of this thread is misleading, that's why I started asking questions in the first place...
 
Top