T-VIPS CP525 cMux, Руководство пользователя

Страница: 275 / 292

Forward Error Correction in IP Networks 275
ID: um_tsmux CP525 cMux User’s Manual Rev. 5.20.12 (3944)
Figure B.11 FEC latency and buffer size
Also shown is the resulting overhead and the number of packets that can be corrected. In
column-only FEC there is one FEC packet per column, resulting in a 1/D increase in transmission
overhead, D being the matrix column depth. I.e. in a 10 row matrix (D=10) the added overhead
is 10%. The minimum allowable column depth of 4 will produce 25% overhead.
In two-dimensional FEC there will be D+L FEC packets in a DxL matrix (L being the row
length). Thus the added overhead is D+L/DxL, which for a 10 by 10 matrix amounts to 20%.
Adding row-FEC will increase the error correcting capability without significantly increasing the
latency or buffer size requirement. Applying row- and column-FEC also enables use of iterative
FEC calculations to recover more missing packets. The equipment manufacturer is at liberty to
determine the algorithm used in error recovery as long as the requirements and limitations of
the specification are respected.