Re: [Discuss-gnuradio] usrp_spectrum_sense.py input parameters in USRP B200

[Marc Pàmies Massip]

Hi Soumaya,

Which is your channel bandwidth? According to your message I think you are not setting this parameter, so I don't know which value the program might be using. I say that because in line 195 (self.freq_step = self.nearest_freq((0.75 * self.usrp_rate), self.channel_bandwidth)) the freq_step is set to discard the bins on both ends of the spectrum, and the formula used depends on both the sample rate and the channel bandwidth. Maybe it is related with your problem (or not?).

I wanted to join this conversation because I am also using osmocom_spectrum_sense and having some trouble with the frequencies as well... I don't really understand the way center_freq is set. In my case I want to find peaks in a band that goes from 925MHz to 960MHz in steps of 200KHz ((960M-925M)/200K=175steps), but when I execute the program printing the variables "center_freq" and "freq" at each iteration of the main loop they don't change as I was expecting.

I thought that in the first iteration center_freq would be 925.1MHz (to compute the power in the 200KHz band that goes from 925.0MHz to 925.2MHz), the second iteration should be cenetered in 925.3MHz and so on until the 175th iteration where the band from 959.8MHz to 960.0MHz should be analysed (so freq_center would be 959.9MHz in this last case). However, looking at the outputs given by the program I see that the center_freq remains the same for a while while freq goes increasing in intervals of 200KHz (input parameter of channel_bandwidth), but the value of freq starts far away from 925MHz so this makes me think that it is not looking at the first band of the spectrum that I am asking from the command line.

It doesn't make much sense to me, con anyone explain how it works?

Thank you in advance,

Marc.

PD: In case it helps, here it is the command I am using:

osmocom_spectrum_sense -a hackrf -s 10000000 -g 20 -b 200000 -F 1024 925000000 960000000

On 17/01/2017 11:42:06, Soumaya el barrak <address@hidden wrote:

_______________________________________________ Discuss-gnuradio mailing list address@hidden https://lists.gnu.org/mailman/listinfo/discuss-gnuradio

Hi Marcus,

I understand the usb bandwidth limitations, but I have to experience this spectrum scanning in the Raspberry pi 3. I used 8 MS/s, fft-size equal to 512 and tune-delay equal to 0.25s. I followed your recommendation, and I scanned only 48 Mhz per time. The energy levels obtained are logic, but while switching from one center frequency to another, almost 3.6 Mhz are dropped, I don’t receive anything along this bandwidth, it scans almost 2 Mhz and drop 3.6 Mhz. I don’t understand what is the reason behind those results and how can I improve them.

Please if you have any ideas, give me a hand!

Best Regards

Soumaya      

2017-01-16 14:24 GMT+01:00 Soumaya el barrak <address@hidden>:

---------- Forwarded message ----------
From: Soumaya el barrak <address@hidden>
Date: 2017-01-16 12:48 GMT+01:00
Subject: Re: [Discuss-gnuradio] usrp_spectrum_sense.py input parameters in USRP B200
To: Marcus Müller <address@hidden>

Dear Marcus,

Thank you again for your explanations, but I installed Gnuradio on a Raspberry pi 3, model B board, so the usb ports are USB2 not 3. That’s why I cannot use a high sample rate. I got an error because of the USB bandwidth limitations. If you can guide me to the right parameters to choose with this kind of host machine?

Also, Can you please explain for me what does this parameter refer to : $((2**16))

The script which I use is this one:

https://github.com/gnuradio/gnuradio/blob/master/gruhd/examples/python/usrp_spectrum_sense.py.  

Thank you in advance

Kind Regards

Soumaya

2017-01-16 12:28 GMT+01:00 Marcus Müller <address@hidden>:

Hi Soumaya,

while tuning, there will simply be transients, mostly of oscillator energy leaking into the receiver. You cannot use the data that you get while tuning.

You might want to define "strange power values"; 2.4 GHz is an ISM band, so a lot of traffic is to be expected there, and that traffic is also probably very bursty, so that your observed values might temporarily vary a lot, since your observation time is much longer than the bursts in e.g. WiFi, your observed power is basically the product of number of burst, burst energy of the individual transmitter, and path loss between your receiver and the transmitter.

Still, why are you using such small sampling rates as 8 MS/s? With a good USB3 controller and a reasonably fast computer, you should easily be able to do 40 MS/s with the usrp_spectrum_sense program. In fact, on my throttled thinkpad X240, I'm currently doing

./usrp_spectrum_sense.py -s 40e6 -g 76 -F $((2**16)) 2.4e9 2.5e9

and it works just fine.

Best regards,

Marcus

On 01/16/2017 10:32 AM, Soumaya el barrak wrote:

Hi Marcus,

Thank you so much for your reply, I followed your recommendations, I used a sample-rate of 8MS/s and FFT-size 1024 and I kept the tune-delay and dwell-delay at 0.25 s. But I still receive some strange power values while scanning the 2.4 GHz band.

I also obtained high power picks while tuning from one center frequency to another. Could you please explain for me where I am mistaken? And what is the explanation of those power picks?

Best..

Soumaya

2017-01-13 14:46 GMT+01:00 Marcus Müller <address@hidden>:

Hi Soumaya,

the B200 tunes slower and might need up to half a second when tuning further.

You should hence minimize the amount of tuning, and instead of 4MS/s use a higher rate, a larger FFT size, and tune less often.

The power normalization is truly, 100%, arbitrary. Since these dB are all "relative to digital 1", they only have linear relation with real-world powers.

Best regards,

Marcus

On 01/13/2017 11:40 AM, Soumaya el barrak wrote:

Dear all,

I am working with USRP B200 for spectrum sensing, I want to scan the band from 2.4 to 2.5 Ghz, I use fft size 256, and sample_rate 4M, and tune _delay 0,001s. I am not sure if the parameters are correctely chosen because when I try to scan another band without wireless activities, I still receive hight power levels. Can anyone has any idea about where I am mistaken ? 

Also, in the formula used to calculate the PSD : 

 Power_dB = 10*math.log10(m.data[i_bin]/tb.usrp_rate); 

Normally it is the magnitude squared averaged , I want to know why it is averaged by the usrp_rate ?  

Thank you in advance !

Soumaya  

--

EL BARRAK Soumaya

Telecommunication & Networking Engineer / Phd Student

Innovative Technologies Laboratory

National School of Applied Sciences of Tangier - (ENSA Tanger)

Abdelmalek Essaâdi University -  Morocco

Tel : (+212) 6 65 9 625 40/ (+ 39) 327 198 5064

_______________________________________________
Discuss-gnuradio mailing list
address@hidden
https://lists.gnu.org/mailman/listinfo/discuss-gnuradio

_______________________________________________ Discuss-gnuradio mailing list address@hidden https://lists.gnu.org/mailman/listinfo/discuss-gnuradio

--

EL BARRAK Soumaya

Telecommunication & Networking Engineer / Phd Student

Innovative Technologies Laboratory

National School of Applied Sciences of Tangier - (ENSA Tanger)

Abdelmalek Essaâdi University -  Morocco

Tel : (+212) 6 65 9 625 40/ (+ 39) 327 198 5064

--

EL BARRAK Soumaya

Telecommunication & Networking Engineer / Phd Student

Innovative Technologies Laboratory

National School of Applied Sciences of Tangier - (ENSA Tanger)

Abdelmalek Essaâdi University -  Morocco

Tel : (+212) 6 65 9 625 40/ (+ 39) 327 198 5064

--

EL BARRAK Soumaya

Telecommunication & Networking Engineer / Phd Student

Innovative Technologies Laboratory

National School of Applied Sciences of Tangier - (ENSA Tanger)

Abdelmalek Essaâdi University -  Morocco

Tel : (+212) 6 65 9 625 40/ (+ 39) 327 198 5064

--

EL BARRAK Soumaya

Telecommunication & Networking Engineer / Phd Student

Innovative Technologies Laboratory

National School of Applied Sciences of Tangier - (ENSA Tanger)

Abdelmalek Essaâdi University -  Morocco

Tel : (+212) 6 65 9 625 40/ (+ 39) 327 198 5064