Hi All, I would like to realize an algorithm to build the best possible dvb-t sdr experience. The idea is, that no matter how many sdrs and what kind of antenna you have, the software gets the best out of your hardware. This is what I have thought: The user community of a system depends on how useful it is, how easy it is to use, build and how much money it costs. Rtl-sdrs are quite cheap, but for now a user has no benefit of having multiple sdrs in its system. That is why I'm searching a way to correlate the signals of the sdrs without hardware modification. I think everyone of you has seen the noise from the power source. Has anyone tried to build a filter to use that noise to calculate the delay between multiple dongles? I mean, they get the same noise. I don't know if it is sufficient, but it is also a good spot, to put some artificial noise in, as it can be easily accessed and does not interfere with rf circuitry when it is switched off. Is that an useful approach? Do you think this could work? Did I miss something? regards, Steve
Stefan Groißmeier 2017-01-10 17:37:
That is why I'm searching a way to correlate the signals of the sdrs without hardware modification.
Just my 2 cents: Intriguing idea, BUT (sorry...): For corrleation you need a common time base, that is a common clock source. That is not the case for multiple RTL-SDR dongles. Moreover you need a stable and known phase relation (runtimes of the clock signal) for your dongles. The USB bus does not provide the clock for the dongles, so this is no way to synchronize them The key word here would be MIMO - Multiple In Multiple Out. That is what WLAN 801.11n does, in one device. I do not know if any of the currently availabe dongles have a clock input (for a common source) or a clock output (to act as clock master) 73 de Patrick OE6PSE -- Engineers motto: cheap, good, fast - choose any two Amateur Radio Operator / Ham / Funkamateur / QTH: JN77rb http://oevsv.at/ One of the lucky 10.000: http://xkcd.com/1053 Use Mail Encryption Today! PGP Key ID: 0xDF8A127E5A120903 Patrick Strasser-Mikhail OE6PSE <oe6pse at wirklich dot priv at>
I've only heard of the possibility to solder out the reference oscillator of all but one dongle and distribute the oscillator of that to the others. That's a modification, and also, only eliminates frequency offset - the random absolute phase (which you can remove by correlation, assuming your antennas /mainly/ see the same signal) stays, and what's more, the phase noise of the synthesizer inside the tuner chip. Point is that these devices are *exactly for reasons of integration* this cheap: there's one chip that does all the LO synthesizing and mixing. You can't get the LO out or into that. But, since Phase noise really is usually uncorellated to the signal we're interested in, Patrick's right – with that modification, you can build Multiple-input receivers. You'll get diversity gain. How much gain (measured in reception quality improvement equivalent to an SNR improvement, hence the name) that is depends on the method you're using – but the fact is: The first SDR you use gives you the most gain (namely, being able to observe the signal at all); the second still a bit, regardless of diversity method, the third always a bit less and so on. So whilst your complexity tends to grow worse than linearly, the increase with every additional receiver dwindles. What's making this worse: All these methods depend on your ability to know the phase of every receiver path. With RTL-SDRs you can't know that (there's no phase synchronization at all, and even worse, no time synchronity to even start with); you have to estimate it from your received signal, every single tune, and continously, since the LOs *will* drift against each other in any normal PLL'ed device. Now, estimating the phase of a reception is a pretty hard problem and usually requires intricate knowledge of what signal you're looking at – that's why all wireless standards have things like preambles, pilot tones, sync words etc. With a pure cross-correlation, your phase estimate's variance is pretty much proportional to your sum(SNR^-1); not a good thing :( So, in general, this is hard, and a receiver that allows you to circumvent at least parts of these problems will be a lot more modular and complex than an RTL-SDR dongle :( Greetings, Marcus On 26.02.2017 21:47, Patrick Strasser-Mikhail wrote:
Stefan Groißmeier 2017-01-10 17:37:
That is why I'm searching a way to correlate the signals of the sdrs without hardware modification. Just my 2 cents:
Intriguing idea, BUT (sorry...):
For corrleation you need a common time base, that is a common clock source. That is not the case for multiple RTL-SDR dongles. Moreover you need a stable and known phase relation (runtimes of the clock signal) for your dongles.
The USB bus does not provide the clock for the dongles, so this is no way to synchronize them
The key word here would be MIMO - Multiple In Multiple Out. That is what WLAN 801.11n does, in one device.
I do not know if any of the currently availabe dongles have a clock input (for a common source) or a clock output (to act as clock master)
73 de Patrick OE6PSE
Hi Stefan,
Rtl-sdrs are quite cheap, but for now a user has no benefit of having multiple sdrs in its system. Hmmm, with several rtlsdr units one can monitor several different frequency bands. I am sure there are users who find that beneficial:-)
That is why I'm searching a way to correlate the signals of the sdrs without hardware modification. I think everyone of you has seen the noise from the power source. Has anyone tried to build a filter to use that noise to calculate the delay between multiple dongles? Surely it is possible to synchronize two or more dongles in software. They just have to receive the same signals to a sufficient amount.
One would have to apply a fractional resampler to all but one rtlsdr to compensate for different sampling clocks. Those resamplers have to be adaptive to accomodate different frequency drifts. This means one has to find the time shift for best correlation, and establish how that time drifts to update the resampling ratio for the time shift to stay constant. There will be many interesting complications if one just uses the normal antenna signals. It would however be fairly easy to inject pulses at perhaps 30 Hz and use them for the synchronization. Those pulses would be known accurately so they can be subtracted from the data stream to not have any adverse effect on the desired signals.
I mean, they get the same noise. I don't know if it is sufficient, but it is also a good spot, to put some artificial noise in, as it can be easily accessed and does not interfere with rf circuitry when it is switched off. Synchronization can not be switched off, stability is not good enough for that - but the artificial noise is known so we can subtract it:-)
Is that an useful approach? Do you think this could work? Did I miss something? Surely it can work. I do not think it will be trivial to get it running however, but you should be able to make a multi-channel receiver that could combine phase-coherent signals from several antennas for drastically improved S/N.
73 Leif
regards,
Steve
participants (4)
-
Leif Asbrink -
Marcus Müller -
Patrick Strasser-Mikhail -
Stefan Groißmeier