[Top][All Lists]

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

Re: [Discuss-gnuradio] Wrong ideal points on USRP1(gnuradio 3.2 rev.1122

From: Stefan Brüns
Subject: Re: [Discuss-gnuradio] Wrong ideal points on USRP1(gnuradio 3.2 rev.11223)? results form IQ const. plotting.
Date: Wed, 28 Oct 2009 17:20:49 +0100
User-agent: KMail/1.9.9

On Tuesday 27 October 2009 22:56:04 tom_unaff wrote:
> Hello,
> I do IQ plotting for DBPSK, DQPSK, 8PSK. (Please have a look below):
> For DBPSK:
> http://omploader.org/vMm16Zg/dbpsk_IQ_plotting.png
> For QPSK:
> http://omploader.org/vMm16Zw/dqpsk_IQ_plotting.png
> Can someone explain why for DQPSK positions of ideal points are wrong?  It
> should be [1,1], [-1,1],[-1,-1],[1,-1]

Welcome to the real world, you just swallowed the blue pill ;-)

Radio signals are subject to 
- attenuation (signal gets weaker when moving tx/rx apart)
- gains from antennaes, PAs (power amplifier) on tx, low noise amplifiers on 
rx side
- multipathing
- frequency deviation (doppler shift, oscillator mismatch between rx/tx)
- deviation in symbol timing (rx samples e.g. 2.001 symbols while tx is only 
sending 2.0 - after 1000 symbols, you are off one symbol)
- ambient noise and interferers

There are several signal processing blocks which deal with this problems, e.g. 
AGC (automatic gain control) for attenuation, carrier recovery for frequency 
deviation. These blocks mitigate the effects, but only to some extent.

You seem to have some of these blocks in you flowgraph already, otherwise you 
would get something looking quite diferrent.

For dqpsk, you have your ideal positions with a seemingly constant phase 
offset, and some additional white noise.

For white noise: Try to have a stronger signal at the receiver. The ambient 
noise level is given for your current environment, but you can: Use antennaes 
with higher directionality, increase the tx power level.

For phase offset: Have a look at the combined carrier/symbol timing recovery 
block gr_mpsk_receiver. Google for Costas and Mueller&Müller for the 
algorithms implemented there.

> Also for DBPSK, Why symbols on right site (of Y-axis ) are far away from
> point, [1,0]? How to fix it?

Difficult to say. This is a DC offset, and it should be dealt with in the 
USRP. Unfortunately, you did not give any information about your flowgraph, 
so you are on your own here ...

In general: get a good book about analog and digital transmission systems. It 
is impossible to build a rx/tx system without a good understanding of the 
existing problems and how to deal with them.


Stefan Brüns
Department of Wireless Networks, RWTH Aachen University
+49 2407 575-7045

reply via email to

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