Re: [Discuss-gnuradio] detect if certain channels are being used

[Martin Braun]

On 05.02.2014 13:13, address@hidden wrote:
> Let me be more clear of what is my goal here:
> We receive the 13cm ham band (2300Mhz - 2400 Mhz)
> This is converted down (-900Mhz) to 1400 - 1500 Mhz) (L band)
> This way normal satelite receivers can rx amateur television stations in 13cm 
> band.
>
> We have 3 input frequencies in use: 2330 Mhz, 2360 Mhz and 2390Mhz
> For each input there are 2 receivers: 1 analoge end 1 digital. (this makes 6 
> receivers)
> these receivers are powered on 24/7. this consumes a lot of power +-120Watt
>
> Are station has te become "green"
>
> So thats why i was thinking of making a "detection" system. so that when a 
> signal comes in (being it analoge tv or digital tv)
> the pair of receivers can be switched on for a while... and the video signal 
> can be recieved and relayed...
>
> The detection may run at slow speeds... 2 or 3 times per second. to reduce 
> processing power.
>
> I'm a newbee with gnuradio... but eager to learn!

As Marcus said, this can become difficult fast, and a lot of people are 
getting PhDs on topics like this. Since your subject seems pretty 
clearly defined, you might be able to implement it without much effort.

Some thoughts:
- Say you don't want an X300, but a USRP with a lower bandwidth (e.g. 
B210). Since your update rate is small, you can simply tune to several 
segments and process them one at a time.
- If your signals are generally narrow-band, it might suffice to 
channelize the signals (e.g. using FFTs) and define a threshold. Since 
USRPs are generally not calibrated, you need to do that -- e.g. by 
estimating a mean noise floor.

In any case, you'll accidentally miss signals and have false alarms 
(e.g. you detect a signal that's not there), each with a certain 
probability, that will never be zero. It depends on many things, such as 
averaging time etc.

If you can live with this, you can definitely implement this with GNU Radio.

MB