Thread Links |
Date Links |
||||
---|---|---|---|---|---|

Thread Prev | Thread Next | Thread Index | Date Prev | Date Next | Date Index |

*To*: STDS-802-3-100GCU@xxxxxxxxxxxxxxxxx*Subject*: Re: [802.3_100GCU] why transmit all parity bits after 20 data blocks in FEC frame?*From*: Jeff Slavick <jeff.slavick@xxxxxxxxxxxxx>*Date*: Mon, 30 Sep 2013 16:03:29 -0600*Delivered-to*: mhonarc@xxxxxxxxxxxxxxxx*In-reply-to*: <CAMhwfs7NWXW-tK37dm+4QXyY5qfH-jhxyuumj70gs9THZfGnhg@mail.gmail.com>*List-help*: <https://listserv.ieee.org/cgi-bin/wa?LIST=STDS-802-3-100GCU>, <mailto:LISTSERV@LISTSERV.IEEE.ORG?body=INFO%20STDS-802-3-100GCU>*List-owner*: <mailto:STDS-802-3-100GCU-request@LISTSERV.IEEE.ORG>*List-subscribe*: <mailto:STDS-802-3-100GCU-subscribe-request@LISTSERV.IEEE.ORG>*List-unsubscribe*: <mailto:STDS-802-3-100GCU-unsubscribe-request@LISTSERV.IEEE.ORG>*References*: <CAMhwfs74A+h6G1zYpaGrs2kvsnMjhLMLyeVh6zZTZvPMNp+xnQ@mail.gmail.com> <E62249CB378D1840BEB0ACFD537259A41297166A@IRVEXCHMB10.corp.ad.broadcom.com> <CAMhwfs7NWXW-tK37dm+4QXyY5qfH-jhxyuumj70gs9THZfGnhg@mail.gmail.com>*Reply-to*: Jeff Slavick <jeff.slavick@xxxxxxxxxxxxx>

Hi Shen,

The 5140b of data in the codeword are protected by **all** 140b of parity. So you have to parse all 5140b of data first in order to calculate the parity bits, thus we calculate the parity bits as the data goes by and insert the parity at the end of the codeword. This allows for the most efficient implementation of the FEC.

-Jeff Slavick

Avago Technologies

On Sun, Sep 29, 2013 at 6:47 PM, 沈胜宇 <shengyushen@xxxxxxxxx> wrote:

ShenSecond, if the answer to 1 is yes, then why not trasmite the encoded 514 bits and 14 bits parity together, such that the receiver can decode it into 514 bit immediatelly? instead, gustlin transmite the 514 first, followed by another 9 514 bits, and then the 14 parity bits of the first 514 data.First, is the RS(528,514) schema presented in http://www.ieee802.org/3/bj/public/mar12/gustlin_01_0312.pdf page 10 the final fec used in 802.3bj draft?Dear ZhongFengI am sorry that I have not capture your idea.

On 30 September 2013 05:42, Zhongfeng Wang <zfwang@xxxxxxxxxxxx> wrote:

Hi, Shengyu,

RS code is a symbol-based block error correction code.

Major encoding and decoding operations are performed on “symbols”.

In our case, a symbol has 10 binary bits. So a FEC frame has exactly one RS block.

Hope this helps.

Best regards,

------------------------------------------

Zhongfeng Wang

Infrastucture and Networking Group

Broadcom Corp., California, USA

From:沈胜宇 [mailto:shengyushen@xxxxxxxxx]

Sent:Saturday, September 28, 2013 6:54 PM

To:STDS-802-3-100GCU@xxxxxxxxxxxxxxxxx

Subject:[802.3_100GCU] why transmit all parity bits after 20 data blocks in FEC frame?

Dear all:

in http://www.ieee802.org/3/bj/public/mar12/gustlin_01_0312.pdf page 10, the fec frame contains 10 Reed solomon encoded blocks. each blocks contains 514 original data bit and 14 parity bits.

But all these 14*10 parity bits are transmited after all 514*10 data bits. I think this leads to large latency in fec decoding, because decoding the first 514 bits can not happen before its 14 parity bits arrived.

So why not transimit 14 parity bits with their corresponding 514 original data bits?

Shen

**References**:

- Prev by Date:
**November Plenary Registration Reminder** - Next by Date:
**Re: [802.3_100GCU] why transmit all parity bits after 20 data blocks in FEC frame?** - Previous by thread:
**Re: [802.3_100GCU] why transmit all parity bits after 20 data blocks in FEC frame?** - Next by thread:
**November Plenary Registration Reminder** - Index(es):