| Thread Links | Date Links | ||||
|---|---|---|---|---|---|
| Thread Prev | Thread Next | Thread Index | Date Prev | Date Next | Date Index |
|
All – Mike Tu and I had some offline conversation and came to the following consensus that 2e-10 RFER is the correct RFER rate to meet our 1e-12 BER objective. The reasoning goes as follows:
For 802.3ch, each RS FEC frame consists of 50 64B/65B blocks, or 3250 input data bits. Now, to align to the BER, we need to the adjust the bit error ratio to by the typical number of errors in an uncorrectable RS-FEC codeword.
The RS-FEC is a 17-error correcting RS code. In gaussian noise, when RS symbol errors are independent the most likely uncorrectable error involves 18 bits received, with a single bit error per (10-bit) RS symbol. These 18 bits in error result in an uncorrectable
RS codeword which represents 3250 64/65B-encoded bits or 3600 bits in total. 2 significant figures should be sufficient, so use 3600.
Therefore 1e-12 BER converts to RFER as 1e-12 * (3600/18) = 2e-10 RFER.
(note, if we use 3250 bits, we get 1.8 e-10 which still rounds to 2e-10). -george George Zimmerman, Ph.D. President & Principal CME Consulting, Inc. Experts in Advanced PHYsical Communications 310-920-3860 To unsubscribe from the STDS-802-3-NGAUTO list, click the following link: https://listserv.ieee.org/cgi-bin/wa?SUBED1=STDS-802-3-NGAUTO&A=1 |