[EFM-P2MP] Local clock and multipoint MAC control dilemma
Hello all,
In the current Ethernet standard the words clock or counter are not mentioned in clause 31 describing MAC control. Therefore, the original MAC control is not aware of propagation of time although it provides for real-time control and manipulation of MAC sublayer operation. Consequently, it is not ideal that the local clock resides in multipoint MAC control. We could consider that the local clock would reside, let's say, in reconciliation layer. When packets are received, reconciliation layer would pass current local time to multipoint MAC control. If the message happens to include a time stamp, the multipoint MAC control would, in case of ONU, pass information to reconciliation layer how much it should increase or decrease current time in local clock. In case of OLT, multipoint MAC control would calculate RTT.
Multipoint MAC control can make decisions when PDUs are sent and may include time stamps in relevant PDUs. It should pass the starting time of each packet to reconciliation layer. The reconciliation layer would forward the packets at correct time so the time stamps in gate, reg. req., and report messages would hold true.
Antti
Antti Pietilainen
Nokia Research Center
P.O. Box 407
00045 NOKIA GROUP
Finland
tel. +358(0)71-8036660, fax. +358(0)71-8036214