QUESTION:
What is the number of the Networks Elements that one Stratum 1 clock can handle? What ITU recommendations talk about this?
ANSWER:
There are two ways of interpreting your question. I will try to answer both.
Case 1: In the case where the PRS is colocated with Stand Alone Syncronization Equipment (SASE), also called a Synchronization Supply Unit (SSU), the PRS output can be fed as the reference to the SSU which can in turn distribute this clock to all the other Network Elements in the Office. The number of elements in the Office that can be timed is limited by the size of the SSU (number of output ports).
Case 2: You may be wondering how long of a chain I can have without getting into trouble. For example, NE #1 can be timed by a PRS; the output of NE#1 is the input to NE#2 which uses the recovered clock to time its outputs; NE#3 receives an output from NE#2 and uses that particular recovered clock to time the outputs of NE#3; and so on. It would appear "logical" that such a chain could be made indefinitely long. This is not true. Every span introduces some additive jitter and wander that the NE PLL cannot remove. Thus every span introduces a degradation that is cumulative, albeit in anon-linear fashion. Thus after some time the notion of "PRS-traceable" is a joke. Recognizing this, the ITU-T (see Recommendation G.803) has provided some guidelines:
ITU-T says the following regarding maximum number of NEs in a chain:
- From a traceable PRC source, the maximum number of consecutive NE should be no more than 20 between a PRC source and a SSU, or between a hierarchical SSU. In a single chain though, there is a maximum limit of 60 NEs from a single PRC source regardless of how many SSUs you insert in this chain.
- The maximum number of SSUs in a chain should not exceed 10.
Keep in mind that this is only a recommendation. It is my recommendation to insert an SSU between every 2 or 3 NEs to reduce the clock-noise buildup.
|
