Monday, November 28, 2005

Take a rest

No update today and I'm going to take a vacation for almost 2 weeks.
So, Don;t expect any update for the next two weeks.
Aftre the vacation, I may move to next interesting , which I have not decided yet.

Friday, November 25, 2005

Wi-MAX sc improved

I worked on WiMax-SC to improve its performances.
-DFE is improved to get more accuracy.
-Block Turbo Decoder is improved to get the same architecture as WiMAX OFDM.
- Bug fix of randomizer (inverted bit order)

So, My Wi-Max development has been Almost done.. besides 2 bugs..

Wednesday, November 23, 2005

Wimax-ofdm wrap up

Today, as CTC worked well, I have wraped up WiMax-OFDM.
I have not done AAS/Mesh/subchannelization/MISO &etc..
I'll move to next intersting in the next week.
Also I tried to fix a bug in Wi-Max SC that randomizer's bit order was
The change is easy, but it made some side effects that
BTC and some others got screwed up.
They were not due to the randomizer change,
just different bitstreams makes different results.
And only one out of 200 tests are not passed yet. I'll solve this tomorrow.

Tuesday, November 22, 2005

Problem on CTC encoder.

I could not solve the mismatch I talked yesterday.
So , I decided to change Table 221 Circulation state lookup table.
I know that this is a standard, which I cannot change personally.
But without changeing the table I cannot move forward.
So far , some simple tests with the updated table worked fine in
CTC only environment. And I'm integrating it into OFDM system.

Monday, November 21, 2005


I started CTC encoding and decoding.
The coding is 70% is finished.
I don't really undersatnd How the circulation status is initialized.
I guess the purpose of the initialization makes the last state to zero
in order to add padding bit on the tail of the block for viterbi-decoder.
But the problem is I could not get zero after the encoding process.
I'll solve the problem tomorrow.

Saturday, November 19, 2005

BTC done.

I have done BTC, which is working well now , after I modified a lot today.
The only problem of the BTC is too BIG!! I don't think this can be calculated in ASIC
in reasonable time and cost.
I'll tune the BTC later, but before that I'll do CTC in the next week.

Friday, November 18, 2005

Accidentally passed ??

Today all 200 of tests are passed with BTC.
But I knew the version is not worked as I expected.
So, I believe No error has taken place in modulation(even 64QAM).
The current version does not have capabiltiy to fix errors.
This'll takes more times.

Thursday, November 17, 2005

BTC progress

Today I had to go to Redwood City ( Can you guess where I am ? )
I did not have much time to do BTC.
I tried to improve BTC , but no much progress today.
It seems that BTC is worse than RS-CC code at least in 64 QAM mode with the current schema.
I have another plan to go in order to get better result in the Decoder.
Stay tuned!

Wednesday, November 16, 2005


I started to encode/decode BTC of WiMax OFDM. In this time, only parity modes are defined. I mean that some modes have only take parity , no Hamming code in one or both directions.
As I did not support the case in WiMax-SC , So, I need to implement the case.
It sounds interesting to me. ^x^)v

Tuesday, November 15, 2005

WiFi & clock offset

I had taken sometime to re-release WiFi cmodels, because I got some ideas to improve WiFi during WiMax development.
Now, I come back for Wi-Max.
I added clock offset and found that if clock offset is around 10ppm,
preamble cannot be detected.
So, I'm wondering how clock can be synchronised in Wi-Max OFDM system.

Thursday, November 10, 2005

Preamble detection again

As I had something else to do , no much progress today.
I was strugging to do detect preambles based on DFT64.
Somehow today's version is not working well if I put some delay and noises. (It worked perfectly without pseudo-channel.c)
Let's see it tomorrow.

Wednesday, November 09, 2005

Decoder progress

So far, All RS-CC modes are worked with No-noise Environment.
But No preamble detection were made yet.
Tomorrow, I'll add some noise and clock offset and completes Preamble detection with DFT64.

Tuesday, November 08, 2005


I heard another standard name in wireless League.
It's called "iBurst".
Is it related to Apple computer??
What's the carrier frequency??
I saw several Home pages regarding iBurst.
None of them told me the answers...
My big questions is if iBurst is one of the next thing ??

Monday, November 07, 2005


Preamble detection is not implemented today.
Before that, I tried to write equalizer and Decoders.
Interestingly, Long Preamble did not have odd subcarriers.
So, I needed to generate odd subcarriers from evens.
This was not difficult, but I guess this can be improved.

Friday, November 04, 2005

Preamble Detection

Preamble is different from 11a and WiMax's is more difficult
than the one in 11a.
I don't have a good idea to detect WiMAX OFDM preamble besides
using DFT.
If I take DFT, it may cost for power consumption.
Any better idea??

Thursday, November 03, 2005

RS encoder

Today I compared the encoder result from the Examples in spec pp444.
Still Reed-Solomon Encoder's paritys are not matched.
This problems were pointed out dring development of WiMax-SC.
Still no solution found. Other components are looks OK.
Tomorrow I'll move to decoder implementations.

Wednesday, November 02, 2005

Encoder coding

I'm coding WiMax-OFDM encoder. But this is not as interesting as for Decoder.
It's not a difficult job but so many things to do...
Today I have successfully compiled its Encoder. Tomorrow I should get correct result from Encoders.
BTW, why Wi-Max has 4our different PHYs and any two of them are so different ?? Each one of them looks independently defined from each other.
This makes me crasy for coding their encoders...

Tuesday, November 01, 2005


I forgot to mention AAS & STC in yesterday's article.
It's new. My questions is whether AAS & STC are different ??
It sounds like they are same one in different words.
I 'll treat them as the same one in my cmodel.
Anyway , I'm writing cmodels for OFDM, which does not have
a big difference from 11a. So, I'm writing the cmodel very quickly.