We seek to understand and document all radio transmissions, legal and otherwise, as part of the radio listening hobby. We do not encourage any radio operations contrary to regulations. Always consult with the appropriate authorities if you have questions concerning what is permissable in your locale.

Show Posts

This section allows you to view all posts made by this member. Note that you can only see posts made in areas you currently have access to.

Messages - benji

Pages: [1]
Software / Re: directKiwi project
« on: June 04, 2018, 0555 UTC »
Hi Linkz,

I noted the addition of microkiwi_waterfall.py to Christoph's kiwiclient.git which indeed takes a snapshot of a specified host kiwi. With default arguments it shows the calculation result over the whole 30 Mhz range but can be narrowed to a high zoomlevel. Locally I use it for manually selecting a suitable antenna for a chunk of the spectrum. This "global" SNR gives a pretty good indication for SNR at a certain kiwi like Marco shows at http://sibamanna.duckdns.org/sdr_map.html.

Another indication of SNR could be the dB value of signal mean over standard deviation just for the passband. The idea of suggesting a measure of SNR is an attempt to avoid kiwis with high RSSI readings due high local QRN or QRM end up on top of the list.

As you mentioned for voice signals within the passband of the selected mode it remains to be seen if there is any benefit of having such an SNR measure. Probably best way to find out is to plot this together with RSSI in a real time graph.

To go a step further I guess for digital signals measured BER would be a better qualifier. For speech SNR is not well defined since the power spectral density of a voice signal is highly variable with time and almost as random as the noise itself. There are routines trying to separate speech from noise first such as using VAD (voice activity detection) or syllabic rate detection and then estimate SNR. However this is probably overkill for the present project.

Anyway thank you for coming back to my suggestions and a filter for to exclude the worst remotes will already be a good addition.

Best regards, Ben.

Software / Re: directKiwi project
« on: May 24, 2018, 1908 UTC »
Hi linkz,

I really do like your python program and it is one of the very few using the kiwiclient interface. After working with it a couple of times a few comments / ideas.

a. S-meter.
Perhaps it may be  better to replace the RSSI S-meter reading by a measure of signal to noise. There are different definitions for SNR. It could be expressed as the decibel value of the ratio of the mean over the standard deviation of the measured signal.

Another way of calculating SNR is found at http://rx.linkfanel.net/snr.html .
Quote from the KiwiSDR  forum :
"Auto Scale" button sets WF max/min by taking one waterfall result, sorting the resulting FFT bins, calling the 50% bin the noise level and the 98% bin the signal level. Then WF min is set to the noise level -10 dB and WF max the signal level +30 dB. Works best when zoomed out so there is a variety of signal levels. The sort/bin idea is from Pierre, who uses a similar idea for SNR scoring at http://rx.linkfanel.net and http://rx.linkfanel.net/snr.html
End Quote

I believe this method is also used to give the reported dB value for WSJT modes. Of course the -10 and + 30 dB are not needed for this purpose. Would be interesting to see that as a time plot together with S-meter readings and look under various QRN and QRN conditions at channels with voice and data transmissions to see if this is a useful to quantify what we hear and perceive as "better audio quality". I guess the kiwiclient is the place to add a SNR computation.

b. Scanning over a limited number of kiwi sites and frequencies
I find myself entering a number of frequencies in the GUI of your directKiwi project and then looking at a number of different kiwis to see how they compare.

Actually if this was automated that would be really nice. Something I was thinking of is have the user enter a list of say 10 frequencies, then select from the list of 200+ kiwis his/her favourite ones (perhaps also 10) to monitor. Scanning thru this table will give a dynamically updated view of the "best" kiwi site and frequency combination. Scanning could be sequential per kiwi ip address or perhaps in parallel with multiple web sockets open at the same time.

A 10 by 10 matrix having for instance frequencies as columns and host kiwi as row could display the latest SNR value. When enhanced with some colour coding akin to a heat map this would look quite very cool. The "best" selection of the moment could further be used to provide audio.
Using kiwclient to interrogate remote kiwis via the internet involves of course some time delay so a realistic refresh rate for such a matrix is probably 30-60 seconds.

Anyway just throwing around some ideas for this already nice directKiwi program.

Best regards, Ben

Pages: [1]