On 03/10/2013 03:53 PM, jeff kephart wrote:
That was me. I've attached the source code. The problem is, this was just a weekend project and I never took the time to polish it up. If you are willing to fiddle around, and especially if you know some python, then I can show you how it works (after I remind myself how it works!) Other than not having an interface, it works well (but probably only in Linux). It is hard-coded to the specs for my device (ezcap) and knows which part of the received band doesn't work (the center frequency and the edges are just noise). It probably can be adapted if your device is different.Someone said they had a script to scan and log active freq via a usb tv tuner? (realtek rtl2832u?) I should have kept the post, but didn't, friend and I are interested in trying to make a scanner app.
The principle of operation is that it computes statistics and measures how must the signal deviates from being normally distributed. With this heuristic it is smart enough ignore the bands that appear to have continuous but never changing signal (noise from computer?) It continually walks through the whole spectrum and collects signals, when anything other than noise ever shows up, it gets logged. The end result is data that can be loaded into gnuplot showing how likely each band is to have something. Takes a few hours to gain good stats over the whole multi GHz range the device can receive. Let me know if you are interested and want more information on how to use it.
Attachment:
rtlsdr_scan.tar.gz
Description: application/gzip
