| Thread Links | Date Links | ||||
|---|---|---|---|---|---|
| Thread Prev | Thread Next | Thread Index | Date Prev | Date Next | Date Index |
|
Kevin: The presence of SyncE makes the 1588 time-of-day recovery easier, but not necessarily more accurate. The big benefit of having SyncE with 1588 is that it provides timing redundancy. If the 1588 master dies,
the SyncE-provided clock will allow a BC to maintain its time-of-day accurately until another 1588 master takes over. If SyncE is not available when the 1588 master dies, a BC’s time-of-day will drift based on the accuracy and stability of its local oscillator.
The definition for the use of TCs in the ITU is still in an early phase. However, they have already been used successfully in some non-telecom applications.
Rich From: Bross, Kevin [mailto:kevin.bross@xxxxxxxxx] I was encouraged to review ITU-T G.8265 through G.8275.1 relative to these timing requirements. In particular, G.8271 talks about the synchronization requirements for various technologies, and G.8271 Amendment 1 has an updated table of
synchronization requirements. The values we’ve talked about encompass that range, though obviously the 5G requirements aren’t finalized yet. Unfortunately, many of these ITU-T docs have a preference for SyncE and the use of Boundary Clocks (BCs), with Transparent Clocks (TCs) largely being a subject of future study. I’m hopeful that future ITU-T documents will show that standard
10GbE (without SyncE) can also meet the necessary timing requirements and that TCs can actually have less error accumulation (at lower cost) than BCs. --kb ------------------------------------------------------------------ From: Richard Tse [mailto:Richard.Tse@xxxxxxxx]
Kevin: I agree that the synchronization error will never be greater than +/-1.5us. This should be very easy to achieve in a RoE network, which is built with specialized support for precision timing. However, because massive MIMO and Tx diversity will be commonplace in future networks, I believe the requirement for RoE in the near future will be much closer to, and maybe even more stringent than, +/-65ns.
If the # of RoE hops is limited, +/-65ns should already be achievable with present-day IEEE 1588 equipment. Rich From: stds-1904-3-tf@xxxxxxxx [mailto:stds-1904-3-tf@xxxxxxxx]
On Behalf Of Bross, Kevin All, We’ve had an assumption in 1904.3 that the endpoints in a link will be synchronized by some method; there are multiple methods to get this synchronization (GNSS/GPS, 1588, SyncE, CSAC, BITS, etc.), so we did not want to prescribe the method
of synchronization in 1904.3. Can we bound what the expectations are for this synchronization, at least up through LTE? (5G is still in definition.) ·
For example, proper handoff between cells requires that the cells be synchronized to within 3 µs. This means that if all cells are within ±1.5 µs of “true time” like TAI, then the cells can do proper handover. ·
Under LTE Advanced, this is reduced to ± 1 µs. ·
The tightest synchronization requirement I see in the current 3GPP requirements is in 3GPP TS 36.104 (there are similar requirements in 3GPP TS 25.104 and 3GPP TS 36.133 as well): 6.5.3.1 Minimum Requirement
For MIMO or TX diversity transmissions, at each carrier frequency, TAE shall not exceed 65 ns. For intra-band contiguous carrier aggregation, with or without MIMO or TX diversity, TAE shall not exceed 130 ns. For intra-band non-contiguous carrier aggregation, with or without MIMO or TX diversity, TAE shall not exceed 260 ns. For inter-band carrier aggregation, with or without MIMO or TX diversity, TAE shall not exceed 1.3 μs. For discussion purposes, can we agree that synchronization requirements for RoE nodes is somewhere in the range of ±65 ns to ±1.5 µs? --kb ------------------------------------------------------------------ |