direction finding of short pulses

Discuss KerberosSDR - 4x Coherent RTL-SDR
Post Reply
mousedude
Posts: 2
Joined: Fri Sep 27, 2019 7:06 pm

direction finding of short pulses

Post by mousedude » Fri Sep 27, 2019 7:20 pm

What is the minimum duration of a radio signal required to obtain a direction bearing?

I've seen demonstrations of direction finding for continuous radio signals, but can this be used for short pulses? The reason I'm asking is because this system might be useful for tracking wildlife radio collars. These collars usually transmit short (20ms) pulses, about once per second. Is it possible to locate a signal of such short duration?

and on a related note, Could multiple stationary Kerberossdr antenna arrays be used to triangulate the position of a moving transmitter in real time?

ckoval7
Posts: 15
Joined: Fri Sep 27, 2019 8:54 pm

Re: direction finding of short pulses

Post by ckoval7 » Fri Sep 27, 2019 8:57 pm

On most computers the system samples about every half second. 20mS pulses are too short for the system to pick up. With some optimizations to the code and a really fast computer, it can sample faster. Apparently a problem with sampling too fast is that it will lose sync, then the bearings become garbage.

myke
Posts: 37
Joined: Wed Aug 28, 2019 7:03 pm

Re: direction finding of short pulses

Post by myke » Sat Sep 28, 2019 3:33 am

I read a recommended sampling rate as :

Under receiver configuration enter the center frequency of your signal of interest, and a sampling rate. For sampling rates we recommend 2.56 MSPS for passive radar, and 1.0 MSPS for direction finding.

So at 20ms transmission burst the system would collect 200,000 comparative samples every 500ms The Android app with its "minimum required power" and "minimum required confidence" setting could be enough to make this work... Would need to be proven ..... As for stationary sites and triangulation - I have seen that being discussed here in the forum... do a little searching.

rtlsdrblog
Site Admin
Posts: 2428
Joined: Mon Nov 19, 2012 11:54 pm

Re: direction finding of short pulses

Post by rtlsdrblog » Mon Sep 30, 2019 10:00 am

Honestly I can't say for sure if such a short pulse will work. The problem is that the code collects a chunk of samples and processes that chunk. If the chunk contains only a short pulse and mostly noise it might not get a good result.

We're working on new code now which will allow you to set the sample chunk size to anything you like, so a smaller chunk might work better for these short pulses.

I'm checking with Tamas if there is any other solution to this, but we might need to wait for some code updates.

And yes, if you did network multiple distributed kerberos units that could all receive the signal you could indeed track a moving TX source.

myke
Posts: 37
Joined: Wed Aug 28, 2019 7:03 pm

Re: direction finding of short pulses

Post by myke » Mon Sep 30, 2019 6:11 pm

The problem is that the code collects a chunk of samples and processes that chunk. If the chunk contains only a short pulse and mostly noise it might not get a good result.

We're working on new code now which will allow you to set the sample chunk size to anything you like, so a smaller chunk might work better for these short pulses.

I'm checking with Tamas if there is any other solution to this, but we might need to wait for some code updates.

Bear with me! just trying to learn some python along side of Kerberos.... The recommended sample rate setting is 1.0 MSPS for direction finding. Does that define the Sampling Size (1MB sample less for FIR & Decimation influences) you are referring to as data chunk? Identified also as "iq_samples" and these are pulled every quarter sec??

EDIT:
Well - Never mind - I see BUFF_SIZE and where that is passed in hydra_main_window.py ....geez some time I need to drop back to 'HELLO WORLD'!

rtlsdrblog
Site Admin
Posts: 2428
Joined: Mon Nov 19, 2012 11:54 pm

Re: direction finding of short pulses

Post by rtlsdrblog » Wed Oct 02, 2019 1:38 am

1 MSPS is just arbitrary, it's just chosen lower than 2.4MSPS just in case the computer can't keep up with the work.

The BUFF_SIZE is the number of samples that is collected in a chunk for processing. The program collects a chunk of samples, works on that, then processes another chunk and so on. Different buffer sizes may be better for different types of signals/applications, but for now either 128 or 256 seems to work the best.

bob_dvb
Posts: 1
Joined: Thu Oct 03, 2019 6:50 am

Re: direction finding of short pulses

Post by bob_dvb » Thu Oct 03, 2019 6:57 am

I am very interested in this, someone I know has asked me to look into tracking a cat which keeps going missing and then turning up where it shouldn't. Doesn't sound unusual for a cat, except when they put a tracking collar on the feline it was removed by someone.

Next step is to try a low power beacon that can be hidden in the collar but that will need to be tiny and low power.

I am torn between trying to to RSSI with LoRa and getting a Kerberos SDR. I've wanted a Kerberos SDR since before it existed, but I need to justify the purchase and this tracking project might just be it! Sure, I could just use a Yagi with a scanner, but where's the fun in that?!

I'd be grateful if you did figure out the optimisation for pulsed signal direction finding and I am sure the cats owner will be to!

Edit: I missed this other thread where this is also being discussed:
viewtopic.php?f=9&t=4602

Post Reply