discuss-gnuradio
[Top][All Lists]
Advanced

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

[Discuss-gnuradio] USRP2 high data throughout and slow PC - possible sol


From: Fabrizio Tappero
Subject: [Discuss-gnuradio] USRP2 high data throughout and slow PC - possible solution
Date: Fri, 22 Oct 2010 09:48:30 +0200

Hello,
I recently successfully collected with usrp2_rx_cfile.py some WiFi
data using a USRP2 with decimation=5 (which gives you about 20MHz of
bandwidth). This means about 1GB of data in few seconds or so. Because
of the write speed limitation of the had disk, it is basically
impossible to do it using a normal HD.
One straightforward solution is modify usrp2_rx_cfile so that it
stores all your data in one big array and saves it in the HD at the
end.
These few lines describe an alternative solution using a "ramdisk"
partitions of type "tmpfs" which allows you to use use any GNU Radio
function without modification. You basically dump your file in your
ram.

sudo mkdir /var/ramdisk
sudo mount -t tmpfs none /var/ramdisk -o size=1100m
cd /var/ramdisk
df -H
sudo usrp2_rx_cfile.py -v -s -e eth1 -f 2462.0M -g 25 -d 5 -N 250M
wifi_d5_g25.bin
cp ./wifi_d5.bin /home/ctae/Desktop/
sudo rm ./*; df -H; cd /
sudo umount /var/ramdisk
cd ~/Desktop

This method seems to me to allow you to overcome the problem of the HD
writing speed. Make sure however to have enough ram, in my case 2GB
was enough (linux without GUI). Please let me know if a better method
exists.

QUESTIONS
1) For some reasons I could not use a decimation=4 (with a XCVR2450
front end) with the greatest and latest USRP2 FPGA firmware, any hint
why?
2)  I understand to read the collected file I can use the
read_cshort_binary.m function (included in GNU Radio ), is this
function the appropriate one considering that I used the option "-s"
with usrp2_rx_cfile.py?
3) with the function read_cshort_binary.m, what is the best way to
understand I collected meaningful wifi data? Measure the signal power
in the collected signal? How?

Best regards,
Fabrizio



reply via email to

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