Re: FEC (Forward Error Correction)
Ravi Ranjan Kumar <raviranjan_kumar-HTy/[email protected]> Mon, 3 May 2004 05:30:17 -0400
| Newsgroups | gmane.org.handasarabia.nour |
|---|---|
| Message-ID | <[email protected]> |
Hi Mohamed, Please find my answers embedded in the mail marked as Ravi>>. -- Ravi ---- On 28 Apr 2004 10:41:21, Mohamed Salem ([email protected]) wrote: > Dear NOUR team, > > Regarding completion of bitstream HW specs, FEC 1/3 and FEC 2/3 details > have to be figured out. > > *FEC 1/3 for Header: > Input bit should be repeated three times. 0 => 000 ; 1 => 111 and so on. > Data input in serial fashion and also output. > Suggest an implementation and clarify the input data rate and output > data rate. Ravi>> It is very clear from the application of this block it self that each bit will be repeatede thrice. So the the input data rate will be 1/3M Bits/s clock at the clock of 1 Mhz. All previous components on 1 MHZ system clock , so The > output from FEC 1/3 will be then at 3 times faster rate or not ? Ravi>> By making the input bitstream 3 times slower we achieve the rate of 1Mega Bits / second at the output fo the FEC1/3 block. > Refer to the signal description in bitstream document > http://www.handasarabia.org/document.php?id=110 > for previous components to know global signals for all components > including FEC. > > *FEC 2/3 For payload: > > 2 bits at input correspond to 3 bits at output. In our scheme every set > of 10 bits is "injected" into the architecture shown in spec document > and then we Append 5 parity bits to them. > In the implementation of the block how can we inform the component that > after each 10 bits append 5 bits in 5 clock cycles ; and also how can > incoming bits not being lost while appending the parity 5 bits;in the > time of 5 cycles needed to append parity ; input data doesn't stop. Ravi >> The rate of incoming bits can be controlled as it is not a stream of data but it will be output of a tx_buffer. For each bit that has to be read from the buffer there will be a read strobe issued. Now in case of 2/3 FEC, there will be read_strobe asserted for 10 clocks continuously and then it will be deasserted for another 5 clocks after that, this will continue till the end of the packet. > > Suggest an Implementation from a controlling point of view as the HW > architecture is default for FEC 2/3 according to BT standard. > > It will be worth discussing also how the Parity(redundant added bits in > each FEC coding type) bits (check bits) are used to detect errors and > correct it in both types of FEC used for Header and Payload. > > Note: Once, this is clarified the bit-stream HW specs will be ready and > complete then HDL coding phase should be taking place in advance. > > LET US LEARN and Progress NOUR. > > Thanks > > Best Regards, > > Mohamed A.Salem > > > > -------------------------- > NOUR Mailing List > Handasa Arabia Organization > http://www.handasarabia.org > [email protected] > > Message number 136 > > > ________________________________________________ Get your own "800" number Voicemail, fax, email, and a lot more http://www.ureach.com/reg/tag -------------------------- NOUR Mailing List Handasa Arabia Organization http://www.handasarabia.org [email protected] Message number 137