DSPRelated.com
Forums

Coding for OFDM

Started by kelvin83 April 5, 2011
Hey all,

I am currently working on a set of trial data on OFDM communication.
However, in order for my algorithm to work better, I need to have as low a
BER as I possibly can. 

My colleague suggested coding. However, my data wasnt encoded in the first
place and since the trial data has already being collected, I have to make
do with what I have, i.e post-coding. 

I have very little knowledge on coding. The rough solution I have in mind
is to fix a coding scheme, and decode my transmitted data. This shall be my
'actual' data and the transmitted data is now the 'encoded' data. On the
receiver, I decode the incoming data in hope of getting a better
performance in BER.

I hope that sounds feasible. Comments and suggestions are most welcomed.
Also can u guys suggest any coding scheme that I should implement on my
system?

Thanks,
Kelvin


On Tue, 05 Apr 2011 06:24:59 -0500, "kelvin83"
<piakster@n_o_s_p_a_m.gmail.com> wrote:

>Hey all, > >I am currently working on a set of trial data on OFDM communication. >However, in order for my algorithm to work better, I need to have as low a >BER as I possibly can. > >My colleague suggested coding. However, my data wasnt encoded in the first >place and since the trial data has already being collected, I have to make >do with what I have, i.e post-coding. > >I have very little knowledge on coding. The rough solution I have in mind >is to fix a coding scheme, and decode my transmitted data. This shall be my >'actual' data and the transmitted data is now the 'encoded' data. On the >receiver, I decode the incoming data in hope of getting a better >performance in BER. > >I hope that sounds feasible. Comments and suggestions are most welcomed. >Also can u guys suggest any coding scheme that I should implement on my >system? > >Thanks, >Kelvin
I'm confused. If the trial data has already been collected as you say (i.e., all the information has already been transmitted without a code through the channel), then you can't go back and add a Forward Error Correction (FEC) coding scheme to help improve the attainable BER. I didn't understand what you meant by 'actual' data and 'encoded' data. Unless you can apply a FEC code prior to transmission you'll be stuck with whatever you can do with Maximum Likelihood reception performance, which is typically not very good for OFDM in frequency selective channels. That being said, there are all sorts of codes to pick from. The best performing codes are capacity-approaching codes and are typically either Turbo Codes or LDPCs. Many systems (like WiFi) frequently do well with much less, like a straightforward convolutional code with a Viterbi decoder. Eric Jacobsen http://www.ericjacobsen.org http://www.dsprelated.com/blogs-1//Eric_Jacobsen.php