[Top][All Lists]

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

[Discuss-gnuradio] GPSDR's - An open standard for time-stamped RF data e

From: L. Van Warren
Subject: [Discuss-gnuradio] GPSDR's - An open standard for time-stamped RF data exchange.
Date: Sun, 2 Aug 2009 19:35:19 -0500

We have a chance to make history, with an array of GPS disciplined SDR's -
GPSDR's for short

This is the ham radio equivalent of HTML - a "lingua franca" for RF data

Enabling radio amateurs to share signals + timing information over the
internet from their GPSDR's opens an amazing vista of experimentation and

It requires defining AND implementing a way of doing business where business
= GPS disciplines Oscillator time-stamps SDR spectral buffers. 

Those signals common to 4 such receivers can be uniquely located in
space-time regardless of their location using multilateration. 

Mapping signal location leads to discovery, mundane as in your light dimmer
or microwave oven, exciting as in tracking black holes, satellites, etc.

If we could figure out how to do it with SoftRocks, that would be awesome
and cost-effective. For technical reasons we need hardware with RF ADC's.

Why? In a nutshell, the time stamp has to be put on the RF sample buffer
before down conversion, or a huge amount of time (and therefore position)
resolution is lost, proportional to the frequency multiple of the carrier
and the "audio" signal. For example 10 MHz / 10 kHz = 1000 fold uncertainty.

If the limit on GPS disciplined oscillators is 50 nanoseconds of jitter and
light travels 1 foot per nanosecond, the time-stamped RF buffer approach can
locate a signal to 50 feet give or take. But if we downconvert first and
correlate, this increases to 50,000 feet, or about 10 miles. Not such a
great foxhunt. So we don't downconvert before correlation.

The problem is that continuous RF sample buffers are too large to ship over
the internet for cross-correlation.

One solution is to place all the GPSDR's in a short period listening mode,
where they listen for a few milliseconds, share information for correlation,
etc. "Listen, share, correlate."

Once signals are located through this shared correlation process, the
time-stamp can be replaced by a position stamp, the samples can be
downconverted, and the continuous (audio) signal can be shared for
traditional SDR consumption via a Panadapter, spectral, or waterfall
display, or for our dear blind friend Ron - a good pair of binaural

"Listen, share and correlate" is inherently parallel and with enough radio
amateurs, becomes a novel form of "cloud computing" with applications to
every slice of the RF spectrum for which sampling receivers can be built.

This even works for noise sources that are stationary, anyway, you get the

An open standard for data exchange would include all the sampling receivers
currently on the market with the flexibility to produce the format.

So what can I do? I can define the format of time stamped spectral sample
buffers and conversion to packets (torrents?) I can define packet exchange
topologies that scale. I can write some software to simulate this.  I will
submit a strawman for your review. You can help out in other ways, providing
your ADC format, and remembering I can't solder...


Van / AE5CC / wdv.com

reply via email to

[Prev in Thread] Current Thread [Next in Thread]