FPGA Central

Hi,

I have a question about channel coding. As I know, coding is used for
improving system performance. At the same time, coding will result in
lower spectral efficiency, except TCM. For higher modulation, the
coding gain is useful only when its performance is better than that of
lower order modultion without coding. Except TCM, other coding schemes
are general of bandwidth expansion type. Even for pragmatic TCM(I, Q
with separate Viterbi coding. For example, 2/3 rate in IEEE 802.16), I
find the high order modulation with 2/3 pragmatic Viterbi code is worse
than that of low order modulation without coding in a high dispersive
channel, even with deep interleaving. This kind of coding is used only
in relatively mild channel?

In a paper(IEEE, trans. on communication, Vol.53, No.11, 2005,
Optimized decision-feedback equalization for convolutional coding with
reduced delay, Page 1859-1866), it gave an example on page 1865, Fig.8.
It displayed an improved DFE with 8-PAM convolutional coding, whose
coding rate is 1/3. The improvement is 1 or 2 dB. Although I did not do
further simulation myself for that example, I suspect that the lower
order modulation (for example, 4-PAM) will do much better than 8-PAM
with 1/3 convolutional coding. Am I wrong? Please help me if you could.


Thank you very much.

On 3 Jul 2006 19:07:50 -0700, "fl" <[email protected]> wrote:

>Hi,
>
>I have a question about channel coding. As I know, coding is used for
>improving system performance. At the same time, coding will result in
>lower spectral efficiency, except TCM.

The presence of TCM doesn't have much to do with it, IMHO. Only the
overall code rate and error performance is needed for this sort of
analysis.

> For higher modulation, the
>coding gain is useful only when its performance is better than that of
>lower order modultion without coding.

Higher order modulation and higher rate coding provides improved
spectral efficiency. This allows one to trade off SNR for throughput
and reliability as the channel conditions change.

> Except TCM, other coding schemes
>are general of bandwidth expansion type. Even for pragmatic TCM(I, Q
>with separate Viterbi coding. For example, 2/3 rate in IEEE 802.16), I
>find the high order modulation with 2/3 pragmatic Viterbi code is worse
>than that of low order modulation without coding in a high dispersive
>channel, even with deep interleaving. This kind of coding is used only
>in relatively mild channel?

I'm not sure what you're saying, but if the spectral efficiencies are
different than there is still likely a tradeoff to be made between the
two schemes.

>In a paper(IEEE, trans. on communication, Vol.53, No.11, 2005,
>Optimized decision-feedback equalization for convolutional coding with
>reduced delay, Page 1859-1866), it gave an example on page 1865, Fig.8.
>It displayed an improved DFE with 8-PAM convolutional coding, whose
>coding rate is 1/3. The improvement is 1 or 2 dB. Although I did not do
>further simulation myself for that example, I suspect that the lower
>order modulation (for example, 4-PAM) will do much better than 8-PAM
>with 1/3 convolutional coding. Am I wrong? Please help me if you could.

An easy way to see the tradeoffs is to plot throughput (or net
spectral efficiency) vs SNR for each scheme together on the same plot.
The outside trace of the overlapping curves is often called the "hull
curve" of the ensemble of modulations and coding scheme (MCS) sets.
Throughput increases with increasing SNR as one selects overlapping
higher order modulations and code rates to continue up the curve. Any
scheme that doesn't touch or contribute to the hull curve (i.e., lies
completely inside it) isn't useful as the sets that do touch the curve
provide better performance.
Eric Jacobsen
Minister of Algorithms, Intel Corp.
My opinions may not be Intel's opinions.
http://www.ericjacobsen.org