discuss-gnuradio
[Top][All Lists]
Advanced

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

Re: [Discuss-gnuradio] Constellation Object


From: Tom Rondeau
Subject: Re: [Discuss-gnuradio] Constellation Object
Date: Mon, 10 Nov 2014 14:12:12 -0500

On Mon, Nov 10, 2014 at 1:31 PM, Richard Bell <address@hidden> wrote:
Hi Tom,

I read that page before making my post, and that's what lead me to conclude the symbol map wasn't working as intended. I ran a test by changing the symbol map array [0, 1, 3, 2] vs [1, 0, 3, 2] and noticed the output didn't change. Only changing the constellation point order made a change in the output.  [1+1j, -1+1j, -1-1j, -1+1j] vs [-1+1j, 1+1j, -1-1j, -1+1j]

This lead me to conclude the symbol table wasn't controlling anything, it was the index of the constellation point array that was mapping symbols. Correct me if you experience different. 

Rich

Yes, I've used the symbol map a lot to change behavior.

A simple way to explore this is to create a test application with a vector_source -> constellation modulator -> throttle -> qtgui constellation sink.

The vector_source can be set to just repeat the same symbol over and over again 0 -> 0x00, 1 -> 0x55, 2 -> 0xAA, and 3 -> 0xFF. That way, you'll only output one constellation point (assuming QPSK here). Then change your symbol map table to see how it affects where the constellation point goes.

Tom

 
On Nov 10, 2014, at 10:04 AM, Tom Rondeau <address@hidden> wrote:

On Tue, Nov 4, 2014 at 9:11 PM, Richard Bell <address@hidden> wrote:
The symbol map feature of the constellation object doesn't seem to do anything. In fact, you can leave it blank. The index of the constellation points is used to map the complex points to bits. The documentation leads one to believe otherwise. Working as intended?

v/r,
Rich


The symbol map is a list that maps to the constellation points. So constellation_points[i] holds the symbol defined in symbol_map[i]. An easy test to see this working is to take the default Constellation Object in GRC, which is defined for QPSK, and change the map to [0,1,3,3] and notice that the symbol 2 (10b) is missing in the output constellation. Without a symbol map defined, it defaults to it being [0, ..., i-1]. The alternative is to crash or catch that the vector is empty and exit on an error.

There is the Digital Modulation page in the manual that describes these things. If you have suggestions for how to improve them, please send them along:


Tom



reply via email to

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