Friday, September 30, 2005

Still Turbo product decoder

I'm still on makinf the turbo product decoder.
Today finally I got error correction capability on
a single line. But with 2-d deocde, it does not work..
This situation does not makse sense to me..
Anyway, I need more tune on the deocder.
It seems thi may take whole of next week...

Thursday, September 29, 2005

Finally I got an idea??

for the turbo product decoder.
I analysed how the parity calculation works
and force every cycle measure the ditance between the current LFSR
and its zero state.

I'm on the way to calculate the idea in Cmodel.

Wednesday, September 28, 2005

"Delta Viterbi" ??

I found myself in wrong direction for the turbo product decoder.
I tried to tune the decoder over yesterday's implementation.
BUt in this morning I found its "viterbi_turbo.c" did not have a capability to fix only a single bit errror. So, I'm trying to study more about SIOS, Delta Viterbi , SOVA and etc..
No update today. Stay tuned.

Tuesday, September 27, 2005

Turbo product decoder

Turbo product Decoder is ready.
It seems it's working. But its performance seems that it's not
good enough.
I'll tune it tomorrow.

Monday, September 26, 2005

Decoding type2

Today I wrote FEC decoder with type2 which takes a kind of viterbi decoder.
This works well . I also started development of Decoder of Type4 with turbo product.
I'm not sure now about this. But I 've already got an idea to decode it with a kind of
viterbi Decoder on Linear Feedback Shift Register.

Friday, September 23, 2005

FEC encoders

Today I read spec regarding FEC of SC downlink.
RS and convolution encoder is not new for me.
TurboProduct is new for me , but it's not difficult to understand.
The problem is RS + 1bit parity. What's this ??
It's inner Coder, so that it should be done before RS decoder??
It does not make sense.. I need to investigate more.

Thursday, September 22, 2005

RS Enc/dec

Today I still tried to match my cmodel with an exmaple of the spec.
But I could not. So I leave this problem , and I did not take any further development. I just added some programmabilities , R(number of parity) and paddinf bytes.
Tomorrow I'll develop all of FEC/Encoder.

Wednesday, September 21, 2005

Confusing day...

To day I tried to reverse the order of calcuration from the original code.
But it's harder than I thought...
Another cmodel from http://www.ka9q.net/code/fec/ was doing
in the correct order, so I was going to switch to use the cmodel.
But I found an example of RS encode in 802.16-2004 spec,
which is different from both of cmodels.
So, I lost a direction to go...
No zip update today...

Tuesday, September 20, 2005

Reed Solomon

I have chosen a cmodel from www.eccpgae.com
It was written in 1989!!
And a configulation (255,223,32) for Wi-Max is working correctly.
As the cmodel is not good for ASIC development,
I need to rewite it using Galoir's multipliers and etc..
By the way, though I would be able to complete the RS code,
its theory is too difficult for me. Anybody can suggest me how it's working?

Monday, September 19, 2005

Hello

Hello,
I have moved to this "blogger" from other blog site.
The previous blog site is still alive for Wifi's PHY.
Please use the blog if you have any concserns on WiFi (or IEEE802.11).

Anyway , today I started to consider Wi-max's PHY.
First of all I'll start from Reed-solomon FEC.
Today I downloaded some of Reed-solomon 's cmodel,
but none of them fitted to WI-MAX's FEC.
So I need to select a cmodel and modify it for Wi-MAX.