Prev: Overlap-add induces ring-modulation
Next: Resampling/interpolation (uniform and non-uniform case)
From: awasthi on 7 May 2010 18:16
Hi,
I have a fundamental doubt regarding bit-loading. I was trying to
simulate VDSL2 system in matlab. (Issue is also applicable to ADSL2+
standard..... might be applicable to others).

Fact#1) Bit-loading algorithms assume that the "energy of subcarrier"
increases monotonically with information (i.e. as number of bits increase)
....... if the distance-between-constellation-points is fixed (4,16,64-QAM
have same "d"), each incremental bit requires double energy than the
previous increment (and every 2 bits will need 4 time the previous
increment).

Conclusion from Fact#1:
Thus, as QAM size increases (due to bit-addition) .......average energy
must increase.


FYI: for example Cioffi book, chapter#4 (
http://www.stanford.edu/group/cioffi/book/chap4.pdf ) .....I'm using
Levin-Campello algorithm.



Fact#2) VDSL and ADSL standards ......are however a bit confusing.
They seem to indicate that average energy of all QAM constellations is same
(so incremental bits would be added with no additional energy).

Reference: VDSL2 standard ......section 10.3.4 ( Link:
http://www.itu.int/rec/dologin_pub.asp?lang=e&id=T-REC-G.993.2-200602-I!!PDF-E&type=items
)

NOTE: Standards do provide an option of gain, gi (for each sub-carrier)
...... but it's only for minor margin adjustment.

From: awasthi on 9 May 2010 14:34
Anyone ?

Do all QAM-constellations in DSL have same average energy ? If, yes
.....then how do they perform bit-loading ?


>Hi,
> I have a fundamental doubt regarding bit-loading. I was trying to
>simulate VDSL2 system in matlab. (Issue is also applicable to ADSL2+
>standard..... might be applicable to others).
>
>Fact#1) Bit-loading algorithms assume that the "energy of subcarrier"
>increases monotonically with information (i.e. as number of bits
increase)
>....... if the distance-between-constellation-points is fixed
(4,16,64-QAM
>have same "d"), each incremental bit requires double energy than the
>previous increment (and every 2 bits will need 4 time the previous
>increment).
>
>Conclusion from Fact#1:
> Thus, as QAM size increases (due to bit-addition) .......average
energy
>must increase.
>
>
>FYI: for example Cioffi book, chapter#4 (
>http://www.stanford.edu/group/cioffi/book/chap4.pdf ) .....I'm using
>Levin-Campello algorithm.
>
>
>
>Fact#2) VDSL and ADSL standards ......are however a bit confusing.
>They seem to indicate that average energy of all QAM constellations is
same
>(so incremental bits would be added with no additional energy).
>
>Reference: VDSL2 standard ......section 10.3.4 ( Link:
>http://www.itu.int/rec/dologin_pub.asp?lang=e&id=T-REC-G.993.2-200602-I!!PDF-E&type=items
>)
>
>NOTE: Standards do provide an option of gain, gi (for each sub-carrier)
>...... but it's only for minor margin adjustment.
>
>
 | 
Pages: 1
Prev: Overlap-add induces ring-modulation
Next: Resampling/interpolation (uniform and non-uniform case)