Thread Links Date Links
Thread Prev Thread Next Thread Index Date Prev Date Next Date Index

[802.3ae] WIS defect conditions




Hi Gareth,
I'm putting in my comments on your queries.
> 
> 
> We'd like to check people's interpretation of the ANSI T1.416-1999
> specification referenced by IEEE 802.3ae with respect to WIS.  In
> particular, we have some confusion about a number of the defect
> coniditions.
> 
> In section 7.5 of ANSI T1.416-1999, Path AIS is defined as "an all ones
> signal in H1, H1*, H2, H2*, H3 and the entire SPE".  Does the SPE need
> to be all ones for the whole frame to be classed as a path AIS frame,
or
> can one just assume that if all the header bytes are all ones, then
this
> is path AIS?

SPE also needs to be all ones for the interpretation of PAIS frame.
It is very clear and has been implemented in many designs.

> 
> In section 7.4.1 of ANSI T1.416-1999, Line AIS is defined as a "signal
> containing a valid SOH with bits 6, 7 and 8 of the K2 byte set to 111,
> and a scrambled all-ones pattern for the remainder of the signal". 
Does
> the remainder of the signal mean the whole frame is all ones, apart
from
> SOH (and therefore including K2[6:8] as this is in line overhead), or
is
> it just referring to the SPE?

POH and the complete payload needs to be all ones.

> 
> In section 7.3 of ANSI T1.416-1999, SEF defect is defined... very
> badly!  According to this spec, the occurrence of four or five
> consecutive errored frame alignment words will cause that frame to be
> marked as SEF.  However, two consecutive error-free frame alignment
> words will terminate the error.  If that is the case, then if ten error
> frame alignment words are detected, followed by ten error-free frame
> alignment words, then this would mean that the frame is no longer in
> error, because the error is terminated.  Correct?  This seems very
> strange to me, and I can't believe it is correct.
We've implemented it successfully in our designs. I do not know
why you're finding it strange. Its a simple thing which which can
be taken care of by a simple state machine.
Even in your example (Ten errored and ten error free) u're stating the
things correctly. Pls elaborate more on this query.
> 
> If anybody can clear any of these issues up for us, that would be
> excellent.
> 	
> Cheers
> Gareth
> 


Rgrds.
                                            
       ANWAR SHAHAB               
    Sr. Design Engineer           
    ST Microelectronics           
    Sector 16 A, Noida          
      INDIA -201301            
  Tel:4515262-71 Extn: 6839     
  Email:anwar.shahab@xxxxxx