Dear all,

I am urgently cramming for a course report... I need to write a report on
Space Time Coding in 2-3 days... Can anybody recommend some good tutorials
and notes on this topic? Consider my poor math, I hope the tutorial/notes
will be understandable in a few days...

Thanks a lot,

-Walala

On Mon, 1 Dec 2003, walala wrote:

> Dear all,
>
> I am urgently cramming for a course report... I need to write a report on
> Space Time Coding in 2-3 days... Can anybody recommend some good tutorials
> and notes on this topic? Consider my poor math, I hope the tutorial/notes
> will be understandable in a few days...
>
> Thanks a lot,
>
> -Walala
>

i know about books, but,

space-time trellis codes are based on the idea of taking cardinal products
of contellations of multiple antennas and applying trellis codes on this
new set of constellations.  the tarokh paper is quite good on how one
should optimize the performance.

space-time layered codes are based on interference cancellation.  examples
are the variants of bell labs' codes.

space-time block codes are based on design theory, specifically orthogonal
arrays.  again, the tarokh paper is quite good, although i may be biased.

there is newer work by jafarkhani, and also by fitz, on combining block
codes and trellis codes.  the main idea is that one can consider the block
codes as a map to the space of multiple antennas, so at the end one can
apply trellis codes again.

so i would say that you should start with trellis codes first, because the
other ones are easier to understand once you are familiar with trellis
codes.

hope that helps,
julius

-- 
The most rigorous proofs will be shown by vigorous handwaving.
http://www.mit.edu/~kusuma

opinion of author is not necessarily of the institute

On Tue, 2 Dec 2003 09:19:29 -0500, Julius Kusuma <k...@mit.edu>
wrote:

>so i would say that you should start with trellis codes first, because the
>other ones are easier to understand once you are familiar with trellis
>codes.

Julius, Really good comments I felt. I havn't known also where I need
to go for first starting to learn this topic. Now, I'll go back to
trellis papers, e.g. your space-time coding paper I saw before.

By the way, I heard that he is worry about too complex mathmatical
descriptions since he don't have much time to do his work. If this
fact is really true, rather than to start somehow complex algorithms
like trellis codes, I suggest for you to start first STTD (space-time
block code) being for just two antenna also called Alamouti code, and
next, its extentions for more than two antennas, which are easily
founded in 3GPP website as TR25.869. 

BR,
------
James K. (t...@hotmail.com)
- Any remarks, proposal and/or indicator would be greatly respected.
- Private opinions: These are not the opinions from my affiliation.
[Home] http://home.naver.com/txdiversity

On Wed, 3 Dec 2003, James K. wrote:

[snip]
> By the way, I heard that he is worry about too complex mathmatical
> descriptions since he don't have much time to do his work. If this
> fact is really true, rather than to start somehow complex algorithms
> like trellis codes, I suggest for you to start first STTD (space-time
> block code) being for just two antenna also called Alamouti code, and
> next, its extentions for more than two antennas, which are easily
> founded in 3GPP website as TR25.869.
>
> BR,
> ------
> James K. (t...@hotmail.com)

james,

i am not an expert on space-time coding, but as far as i know, extensions
of space-time block codes to the case of more than 2 antennas is based on
design theory, which is quite mathematical.

on the other hand, space-time trellis codes are based on ungerboeck's (or
some would say forney's) framework on using cosets.  thus it is somewhat
easier to extend to the case of more than two antennas.  it just depends
on your ability to draw cardinal products of constellations of multiple
antennas.

with respect to trellis codes, the best reference is the book by divsalar
et.al., unfortunately a bit hard to find, which is non-mathematical and
very intuitive.

to be honest i find it very hard to read standards, plus standards tend to
give methods, but don't say why.  maybe you've had much more luck with
standards than i have!

julius

-- 
The most rigorous proofs will be shown by vigorous handwaving.
http://www.mit.edu/~kusuma

opinion of author is not necessarily of the institute

On Wed, 3 Dec 2003 10:11:30 -0500, Julius Kusuma <k...@mit.edu>
wrote:
> [snip]
After Alamouti codes, people are interested in that area to extend
such a space-time code for more than two antennas. To do this, we have
some points to be considered before to start the development of this
idea. I now want to descried these points in two subpart: one of them
is by putting my idea on your last comment, the other is regarding to
the fundamental characteristics of the space-time code.

>to be honest i find it very hard to read standards, plus standards tend to
>give methods, but don't say why.  maybe you've had much more luck with
>standards than i have!

First, about your comments. The fact that you (Julius) pointed above,
seems correct with very 100% on my consideration, which, frankly
speaking, I forgot at that moment. As you said, up to my experience,
people in the standard don't have much interested in mentioning the
technical reason for the proposed method. Only a few of them appended
the details, but usually briefer than in a paper, to their Tdocs.  

Second, about the fundamental characteristics of the space-time code,
designing new space-time codes, we should consider below:
1. Pure block code for more than 2 antennas provides additional
diversity gain, but, effective gains from more than two-order
diversity is a very little.
(In addition, it was proved that full rank transmission is not
possible to send information through more than two antennas).
2. Because of the fact told in the number 1, we can add the coding
gain by considering addition of trellis in the conventional two-order
diversity, which scheme was noted by Julis in his previous article.
3. One another way to prevent such a fact, to combine some
beam-forming feature to the open-loop system. Using both features in a
same time, we will take the pros from both systems with little
overhead.

In brief, I also seconded on Jullis opinion that in standard, not many
times to provide in detail reason for reader's technical
understanding. Only a part of technical information are founded in the
TR (Technical report), like TR25.869 (Tx Diversity) in 3GPP, are
nearly not founded in TS (Specification). Additionally, some points to
be considered the designing of extended the space time block code were
discussed very shortly.

On Sun, 07 Dec 2003 00:33:57 +0900, James K. <t...@hotmail.com>
wrote:
(I just revised to this for better understanding, and easy to read.)

On Wed, 3 Dec 2003 10:11:30 -0500, Julius Kusuma <k...@mit.edu>
wrote:

Since the Alamouti code was developed, it has been interesting to many
researchers for both its good performance and simplicity, and some
codes have been considered to extend this code for more than two Tx
antennas because the original code was proposed only for two Tx
antennas. Regarding to the extended codes, it has been widely
announced that we have some facts to be taken into account when
developing them. In next two parts, I now describe two issues: one is
my answer to the previous comment, and the other is about the
fundamental characteristics of the space time code being used as an
extended version of Alamouti code.

> [snip]
>to be honest i find it very hard to read standards, plus standards tend to
>give methods, but don't say why.  maybe you've had much more luck with
>standards than i have!

The fact dictated previously by Julius is really true, even on my
experience; frankly speaking, at that moment I forgot it for a while.
As you said, it seems like that people in the standard are not much
interested in mentioning the technical reason for their proposed
method. Only a few of them appended the details, but usually briefer
than explained in the conference paper, to their Tdocs.

Second, I summarize some issues being accounted as fundamental
characteristics when designing new extended space time codes:

1. A pure block code for more than 2 antennas provides additional
diversity gain, but this gain is not effective as much as gain from
the two-order diversity. In addition, it was proved that full rank
transmission is not possible to send information through more than two
antennas, especially when information are modulated in complex
constellation.

2. To resolve the problem in (1), we can combine trellis code into
block code and, hence add the coding gain, instead of adding possible
full diversity. In the previous article, Julius also said that it has
been one of the interesting areas.

3. One another way is to combine some beam-forming feature to it. By
combining both features in a same time, we can take both advantages of
two schemes, and just with little overhead.

In brief, I seconded on Julius opinion that in the documentations of
the Standard, not many times it is provided of the detail reason to
help reader's technical understanding. Indeed, a little technical
information are founded only in the TR (Technical report), e.g.
TR25.869 (Tx Diversity), but are not found nearly in TS
(Specification). Additionally, some facts being considered when
designing the extended space time block code were discussed in very
short.

When consider space time code, I would like talk about hassibi's
linear dispersion code. I think it provide a broader scheme to design.
Also it is easy to understand. I don't know if I understand it well.
However, I think it is a good ideas though we have much things to do
to improve it. How do you think about it?

On Sun, 07 Dec 2003 00:33:57 +0900, James K. <t...@hotmail.com>
wrote:

>On Wed, 3 Dec 2003 10:11:30 -0500, Julius Kusuma <k...@mit.edu>
>wrote:

>>to be honest i find it very hard to read standards, plus standards tend to
>>give methods, but don't say why.  maybe you've had much more luck with
>>standards than i have!

>First, about your comments. The fact that you (Julius) pointed above,
>seems correct with very 100% on my consideration, which, frankly
>speaking, I forgot at that moment. As you said, up to my experience,
>people in the standard don't have much interested in mentioning the
>technical reason for the proposed method. Only a few of them appended
>the details, but usually briefer than in a paper, to their Tdocs.  

It is not the job of a  standard document to explain why something was
standardized, only what the standard is.  Sometimes they don't even do
that well, but they're often confusing enough just trying to that that
sometimes I think it's a good thing that they don't do any more than
that.

Many standards do contain informative (in addition to the normative)
text if is deemed warranted or necessary for someone to properly
interpret the intent of what is being standardized.  Some standards
bodies (like the IEEE) maintain the documentation trail of the "why"
part of it, but having gone through that a few times it is extremely
tedious.  Capturing "why" something got in the standard can often
involve years of debate and sorting through several candidate
technologies across many companies, etc., etc.  Trust me on this one,
we really don't want standards to capture why something appears in the
standard, only that it does.

>Second, about the fundamental characteristics of the space-time code,
>designing new space-time codes, we should consider below:
>1. Pure block code for more than 2 antennas provides additional
>diversity gain, but, effective gains from more than two-order
>diversity is a very little.

That's true.  Adding diversity, any kind of diversity, provides
diminishing returns after a point.


Eric Jacobsen
Minister of Algorithms, Intel Corp.
My opinions may not be Intel's opinions.
http://www.ericjacobsen.org

e...@delete.ieee.org (Eric Jacobsen) wrote in message
news:<3...@news.west.earthlink.net>...
> Capturing "why" something got in the standard can often
> involve years of debate and sorting through several candidate
> technologies across many companies, etc., etc.  Trust me on this one,
> we really don't want standards to capture why something appears in the
> standard, only that it does.

During a session on one of the first acoustics conferences I went to, an 
anecdote was told about why some acoustics standard ended up like it did. 

Apparently, two "hotshots" in acoustics, (sorry, I don't remember the 
names, except one of the debattants was either Bruhl or Kjær, founder
of the Danish acoustics company. I suspect the other part was Sabine, 
who was an early authority on room acoustics), or even the subject under 
debate. The deabattants had quarreled over this question to be standardized
for ages, they held incompatible views, and everybody else were utterly 
fed up about it since no one could do any measurements or build any 
measurment equipment until a standard was established. 

The matter was setteled during some conference when the two guys were 
locked into a room and told that they were not let out of there until 
they had agreed on a standard. And thus the standard was established.

Perhaps somebody out there know what this was all about, and can fill 
in the details?

Rune

Rune Allnor wrote:

> e...@delete.ieee.org (Eric Jacobsen) wrote in message
news:<3...@news.west.earthlink.net>...
> 
>>Capturing "why" something got in the standard can often
>>involve years of debate and sorting through several candidate
>>technologies across many companies, etc., etc.  Trust me on this one,
>>we really don't want standards to capture why something appears in the
>>standard, only that it does.
> 
> 
> During a session on one of the first acoustics conferences I went to, an 
> anecdote was told about why some acoustics standard ended up like it did. 
> 
> Apparently, two "hotshots" in acoustics, (sorry, I don't remember the 
> names, except one of the debattants was either Bruhl or Kjær, founder
> of the Danish acoustics company. I suspect the other part was Sabine, 
> who was an early authority on room acoustics), or even the subject under 
> debate. The deabattants had quarreled over this question to be standardized
> for ages, they held incompatible views, and everybody else were utterly 
> fed up about it since no one could do any measurements or build any 
> measurment equipment until a standard was established. 
> 
> The matter was setteled during some conference when the two guys were 
> locked into a room and told that they were not let out of there until 
> they had agreed on a standard. And thus the standard was established.
> 
> Perhaps somebody out there know what this was all about, and can fill 
> in the details?
> 
> Rune

It sounds like the choosing of a pope. The story goes that long ago, the
College of Cardinals would meet in a convenient town and harangue there
until the local food was exhausted, then move on to another. After the
entire countryside was denuded -- sometimes after more than a year --
they would somehow manage to agree on one candidate. When they reached
one cathedral city -- Sienna? -- early in one selection process,
irreverent townspeople locked them into the cathedral, allowing no one
in or out until a pope was chosen. The cardinals were provided with two
kinds of fuel; one made black smoke, the other, white. White smoke
signaled that a pope had been elected, and the doors were unbarred. The
practice has become a tradition, and electing a new pope has seldom
taken long since then.

Jerry
-- 
Engineering is the art of making what you want from things you can get.
¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯¯