# branch metric of MLSE equalizer for GSM

Started by July 14, 2008
Hi All,

here.

http://www.freescale.com/files/dsp/doc/app_note/AN2943.pdf?fpsp=1&WT_TYPE=Application%20Notes&WT_VENDOR=FREESCALE&WT_FILE_FORMAT=pdf&WT_ASSET=Documentation

1) is the Equation 6 in the article correct?
2) How can I get Equation 6 from Equation 5? Can anybody give some idea or
good detailed reference?

Thank you.

Richard

>Hi All,
>
>here.
>
>http://www.freescale.com/files/dsp/doc/app_note/AN2943.pdf?fpsp=1&WT_TYPE=Application%20Notes&WT_VENDOR=FREESCALE&WT_FILE_FORMAT=pdf&WT_ASSET=Documentation
>
>
>1) is the Equation 6 in the article correct?
>2) How can I get Equation 6 from Equation 5? Can anybody give some idea
or
>good detailed reference?
>
>Thank you.
>
>
>Richard
>

%%%%

Hi

To understand MLSE equaliser properly, try to read Digital Communications
by John G Proakis, 4th edition.

Read section 10.1.2 to 10.1.3, i.e. from page 601 to 607.

he has also given one example.

chintan

On Jul 14, 3:34 am, "richard_zhang" <hardhear...@yahoo.com.cn> wrote:
> Hi All,
>
> here.
>
> http://www.freescale.com/files/dsp/doc/app_note/AN2943.pdf?fpsp=1&WT_...
>
>
> 1) is the Equation 6 in the article correct?
> 2) How can I get Equation 6 from Equation 5? Can anybody give some idea or
> good detailed reference?
>
> Thank you.
>
> Richard

Richard,
I haven't analysed it completely but what you'll want to do to prove
the result is to express the equation in vector form
So, the metric to be maximized can be written as the norm of an L-
sized vector:
M^\tilde[k] = || x - A*h ||^2 where x and h are L*1 vectors and A is
an L*L matrix containing the symbols to be decoded.

Then, use vector algebra to simplify, some of the terms will get
removed because each element of A has magnitude=1 and
you should get the form in equation 6.

Thanks,
Dilip.