Re: [Discuss-gnuradio] small polyphase channelization application

[Tom Rondeau]

On Sat, Feb 20, 2016 at 12:43 AM, tom x <address@hidden> wrote:

Hi,

I am working on an application to capture IEEE 802.15.4 traffic on multiple channels simultaneously. (I will mention now that I am new to GNU Radio and DSP.) I can successfully capture on one channel. I read the 2009 paper "Multi-Channel IEEE 802.15.4 Packet Capture Using Software Defined Radio", where L. Choong is able to capture traffic on four channels simultaneously with four frequency translation and FIR filter blocks. I have tried this approach with two channels but did not have success, most likely because the parameters to my filters were incorrect.

This mailing list's archives pointed me to the polyphase channelization block, which I think can simplify this task. Here is a link to a picture of my flow graph:

http://i.imgur.com/sxQnEIw.png

I am tuning the USRP to halfway between channel 25 and 26 (whose corresponding frequencies are shown in the picture). Each channel is 2MHz wide, and the center of each channel is separated by 5 MHz. With this in mind, I set the USRP's bandwidth to 7MHz, which would contain channels 25 and 26 (Is setting this parameter necessary? I would like to scale this design up to four channels.)

The 'target_rate' is the width of a channel, so I am setting the frequency cutoff to half of that, assuming the signal I want would be centered about 0 Hz. I set the transition window to less than half of the sample rate, on the advice of this blog: http://blog.kismetwireless.net/2013/08/hackrf-pt-2-gnuradio-companion-and.html . However, no packets are making it to the sinks, when I transmit on either channel from a separate radio. Does anything about this set up jump out as incorrect? I would appreciate any tips you could offer.

Thanks,

Tom

Tom,

This might help you out:

http://www.trondeau.com/examples/2014/1/23/pfb-channelizers-and-synthesizers.html

One issue is you're offsetting the frequency and shouldn't be. Channel 0 spans -B/2 to +B/2 (B being the bandwidth of the channel) and centered around 0. There should be images in the link (or the paper it references) that should show this.

Tom