|
From: | Marcus Müller |
Subject: | Re: [Discuss-gnuradio] Using two USRPs - how to synchronise a receiver (scan a range of frequencies) looking for a transmitter which is transmitting on a specific frequency |
Date: | Fri, 21 Jul 2017 13:06:55 +0200 |
User-agent: | Mozilla/5.0 (X11; Linux x86_64; rv:45.0) Gecko/20100101 Thunderbird/45.8.0 |
Hi Neil,
1) in this case, the method of choice would be to simply generate a 1MHz wide baseband signal and only fill the channel you're interested in with payload signal – the "Rotator", or the "Frequency Xlating FIR" or OFDM or something like a PFB synthesizer would be viable. 2) well, that very much depends on your channel, noise and signal models!!! Whatever you pick, you will probably need to write down a mathematical justification, a false alarm rate and a detection rate, soooo it might be worth reading into the theory of estimators and detectors. In any case, this sounds pretty simple to me: Use a PFB channelizer (the Doxygen explains how they work, and Tom Rondeau's blog has some info, too) to divide your 1MHz reception into 5 channels, and do independent detection concurrently on each of those. Detection can be done in a lot of ways. For your own signal, you'd probably look for something like a preamble after a matched filter (hint: you can use a matched filter as prototype filter for your PFB). For the question whether a channel is occupied by something else or not: uh, you'll need a signal model for that. That signal model would lead to something like an average signal power in each band of the same PFB as above, to something like an estimate for positions of indepent sinusoids in the whole 1 MHz passband, to cyclostationary estimation of unknown patterns in the signal to... whatever your signal model indicates might be a good idea! 3) Well, that sounds like a back channel! In other words, you're
designing a Media Access Control method. A common thing to do is
dividing the channel into time slots. Sender A selects channel,
says "Hello, I use this channel now" on that channel, then
switches to reception mode on the same channel. B hears the
"Hello", detects it, and answers within a limited time window with
a "Hey, yeah, sure, let's use this channel", which A listens for.
As soon as it got that, it switches back to transmission and sends
the data it wants to send. But that's just one out of millions of imaginable schemes for
this. If you think your channel will be seldom used at all, you
might want to omit the handshake completely (assuming your
detection probability is high enough, of course), as it just
wastes time (if you just send whenever you please, that media
access scheme would be called ALOHA. Understanding ALOHA should
definitely be in the syllabus for every comms theory student!).
Best regards, Marcus On 07/20/2017 09:04 PM, neil shelley
wrote:
|
[Prev in Thread] | Current Thread | [Next in Thread] |