Professional Documents
Culture Documents
SyncE and PTP Telecom Profiles
SyncE and PTP Telecom Profiles
Paul Norris,
April 2017 www.calnexsol.com
Synchronisation in the Mobile Network
• Frequency Sync
• SyncE
• PTP Frequency Profile
• Time Sync
• PTP Time Profile for Full Timing Support
• PTP Time Profile for Partial Timing Support
Company Confidential 2
Frequency Sync
Company Confidential 3
G.8261: Frequency over Packet Interfaces
4
Synchronous Ethernet (SyncE)
Company Confidential 5
G.8262: Synchronous Ethernet Clock (EEC)
• Based on synchronising the physical layer of the Ethernet
• Identical in performance to SONET/SDH Equipment Clock (SEC)
• Directly replaces SEC in the synchronisation chain
• Standard defines:
• Free run frequency accuracy
• Wander performance: generation, tolerance and transfer
• Jitter performance: generation and tolerance
• Transient behaviour: reference switching and entry into holdover
• Interface types
• New revision (January 2015):
• Includes jitter specification for 40 and 100Gbit/s interfaces
Company Confidential 6
Synchronous Ethernet (SyncE)
Features
• Line rate of the Ethernet Interface used to transfer timing.
• No impact/demand on packet layers.
• Defines the use of a high stability oscillator to generate line frequency.
• Ethernet ‘Classic’: ±100ppm.
• Synchronous Ethernet: ±4.6ppm
• ITU-T Standards in place.
• G.8262: Timing Characteristics for Synchronous Ethernet Equipment
• G.8261: Timing & Synchronisation in Packet Networks
• G.8264: Distribution of Timing Through Packet Networks (ESMC)
7
Synchronous Ethernet (SyncE)
Challenges
• Cost: All interfaces need to be Sync-E compatible.
• Can not be used with existing Ethernet equipment when
transferring synchronisation.
~
PRC
~
PRC
8
Ethernet Synchronisation Message Channel
ESMC
• ESMC PDU transmitted once per
second on the Slow Protocol Channel
• Point to Point Protocol.
• Each network element generates and
terminates ESMC messages
9
Wander and Jitter
• Low speed variations in a signal or clock (10Hz down to uHz)
are termed Wander.
• Wander affects the ability of the internal clocking function to
lock to and track the up-stream reference. Wander therefore
impacts the distribution of clock references across networks
Company Confidential 10
What causes Wander
• Each EEC is locked to the frequency of the
PRC by recovering the clock from the PRC
wander in a network.
option 1 clocks <
20
SEC EEC
Total number of
• Wander is also caused by G.813 clocks in a
sychronisation
SEC SEC
• Slow changes in the relative phase of two clock G.8262 is Compatible with G.813
signals due to climatic temperature changes.
• low-frequency phase noise in a clock oscillator.
Company Confidential 11
What causes Jitter
• Jitter is always present at the output port of any network
element (NE), even if it originates a signal or an entirely jitter-
free digital signal or clock is applied to its input.
Company Confidential 12
Sync-E Jitter
Tx Rx Tx
• In both Ethernet and Synchronous Ethernet, pulse edges carry timing information
• The timing information is recovered in the Rx and used to sample the data
• In Sync-E, the timing information is also passed on to the Tx side
Clock Packet
recovery
Sampler Data processing
Clock
13
Inside the sampler
•If jitter exceeds device tolerance, sampling moves away from eye
centre towards edges
•The input sampler mistakes 1s for 0s and vice-versa – bit errors
•The packet processing in the device then drops any packet with one
or more bit errors
14
How Sync-E passes the clock on
Rx
Clock Packet
recovery
Rx
Sampler Data processing
Tx
Frequency lock
Narrowband
clock Tx clock PHY
filtering
15
G.8261: Frequency over Packet Interfaces
Packet
M Network
S
PRC Up to 20 SECs SSU Up to 20 SECs SSU SSU Packet Master and Slave End
(G.811) or EECs* (G.812) or EECs* (G.812) (G.812) (slave: G.8263) Equipment
(G.813/G.8262) (G.813/G.8262)
Company Confidential 16
PTP Telecom Profile for
Frequency (G.8265.1)
Company Confidential 17
Prime Objectives
• To permit the distribution of frequency using PTP over existing
managed, wide-area, packet-based telecoms networks
• To allow interoperability with existing synchronization networks
(such as SyncE and SDH)
• To define message rates and parameter values consistent with
frequency distribution to the required performance for telecom
applications
• To allow the synchronization network to be designed and
configured in a fixed arrangement
• To enable protection schemes to be constructed in accordance
with standard telecom network practices
Company Confidential 18
Key design decisions
• No on-path support (e.g. boundary and transparent clocks)
• Must operate over existing networks
• IP adopted as the network layer
• Ubiquitous and end-to-end
• Unicast transmission adopted over multicast
• Guaranteed to work over wide-area telecoms networks
• Upstream multicast often not enabled in carrier networks
• Continuity of SSM with SONET/SDH and SyncE
• PTP Announce message adapted to carry the Quality Level (QL) indications
• BMCA (Best Master Clock Algorithm) replaced by static provisioning
• Planned synchronization and protection flows, rather than dynamic
adjustment
Company Confidential 19
General Packet Timing Architecture – G.8265
~ PRC
PTP
Master
Packet Timing Signals
PTP
Slave
PTP
Slave
PTP
Slave
Company Confidential 20
Unicast negotiation - G.8265.1
• Slave requests service from master
• Unicast only no multicast
• Slave requests message rate
• Grants are for limited duration
• Default 300 seconds
• Adjustable 60 – 1000 secs
• Slave can request
• Announce messages
• Sync Messages
• Delay-Resp messages
Company Confidential 21
Rate of Timing Messages
• The rate of timing messages required is dependent on several factors
• Amount of noise in the network
• Local oscillator stability
• Efficiency of clock servo algorithm
• The Telecom Profile defines the range of message rates that Masters
and Slaves should support:
Message rates Minimum Maximum Default
Announce 1 msg. every 16s 8 messages/s 1 msg. every 2s
Sync 1 msg. every 16s 128 messages/s Not defined
Delay_Request 1 msg. every 16s 128 messages/s Not defined
Company Confidential 23
Source Traceability
• Encodes QL values in the clockClass field of the Announce
message
• Provides end-to-end traceability of the reference source along the
synchronization chain
• Informs the slave clock (and subsequent devices) of the quality of the
timing source
• Allows the timing chain to be managed in a similar way to existing
synchronization networks
End-to-End Source Traceability
PTP PTP
GM Slave
PTP
Master
(secondary) PTP
Slave
Company Confidential 25
Master Selection and Protection
• Telecom slave clock consists of several logical protocol instances,
each communicating with a different grandmaster
• Selection process follows G.781 selection rules:
• Availability, Traceability, Priority
Slave Telecom
PTP GM
1
Protocol
Instance 1
Slave Clock
G.781-
Slave based
PTP GM Packet
Protocol Master
2 Network Instance 2 Selection
Process
Slave
PTP GM
Protocol List of N
N Instance N
Grandmasters
Company Confidential 26
Separate PTP domains
Example of how PDV affects OCs
PDV Accumulation
Timing packets have to fight there way through network devices against traffic
28
Packet Delay Variation
• “Green Light” protocol
• Some (small) number of cars will drive through a town with every traffic
light at green, and arrive at their destination with the least delay
• This “minimum delay” is consistent, and cannot be improved (at least,
not without breaking the speed limit)
• The number of cars achieving this “minimum delay” reduces with
congestion
• Same applies to packets in a network
• Queuing in a network element is like waiting at a traffic light
• The more congestion on the network, the less chance there is to avoid
queuing
Company Confidential 29
Packet Delay Distribution
Frequency of Cluster
occurrence range
Low
congestion
Medium
congestion
High
congestion
Ideally Capture for >24 hours to see the changes in PDV throughout a day and
capture any intermittent issues 31
PDV Metrics
Floor Packet Population (G.8261.1)
• Packet Delay Variation network limit at
point C of figure 3/G.8261.1 (for HRM-1)
• Floor Packet Percent (FPP)
• Window interval W = 200s
• Fixed cluster range δ = 150µs (from floor delay)
• At least 1% of packets must fall in this cluster.
PDV Metrics
Vendor Specific
Requirement A
Node B requirements:
• The Node B will lock within a 16 minute period if either
a) 1% of packets within 20µsec of Noise Floor, or
b) 99% of packets with <0.3msec spread.
Requirement B
• If case b) is greater than 0.3msec, then
• If 99% with <3msec spread, will lock
within 60 minutes.
• If 99% with <10msec spread, will lock
within 180 minutes.
PEC-M PEC-S-F
Company Confidential 34
PTP Telecom Profile for Time
using Full Timing Support
(G.8275.1)
Company Confidential 35
Prime Objectives
• To facilitate interoperability between clocks designed for the
ITU-T time/phase architecture (G.8275)
• To permit the distribution of time and phase using PTP over
managed, wide-area, packet-based telecoms networks
• To allow the synchronization network to be designed and
configured in a fixed arrangement
• To enable protection schemes to be constructed in accordance
with standard telecom network practices; specifically the ability
to use multiple geographically separate T-GM clocks
• To define message rates and parameter values consistent with
time/phase distribution to the required performance for
telecom applications
Company Confidential 36
Key design decisions
• Uses a boundary clock at every node in the chain between PTP
Grandmaster and PTP Slave
• Reduces time error accumulation through the network
• Boundary clocks defined with a filter bandwidth of 0.1Hz
• Recommends the use of Synchronous Ethernet to syntonise
each boundary clock to a stable frequency
• Defines Sync and Delay_Request message rate of 16 messages/s
• Operates over a Layer 2 Ethernet network
• Uses Ethernet multicast addresses identified in IEEE1588-2008 Annex F
• Support of unicast IP not agreed due to potential for insertion of non
PTP-aware equipment
• Supports multiple active grandmasters for redundancy
Company Confidential 37
Hypothetical Reference Model
Up to 20 BCs in a chain
SyncE
frequency
distribution
networks
PRC
frequency
references
[PRCs may be separate or common]
Frequency Plane
Company Confidential 38
Source Traceability
Uses clockClass attribute to determine phase/time traceability:
defaultDS. Input Frequency Time
Phase/time traceability description
clockClass SyncE QL Traceable flag Traceable flag
T-GM connected to a PRTC in locked mode
6 n/a TRUE TRUE
(e.g. PRTC traceable to GNSS)
PRC / PRS TRUE TRUE
T-GM in holdover, within holdover specification 7
Non-PRC FALSE TRUE
T-GM T-TSC
T-BC T-BC T-BC T-BC
Features
• Every element in the path must be “PTP aware”
• T-BC case covered in standards, T-TC case under development
• Uses a combination of SyncE and PTP, where SyncE provides the
frequency and PTP the phase/time
Company Confidential 41
PTP with Full Timing Support
Benefits
• Controlled, deterministic environment suitable for both
frequency and time/phase transfer
• “Building block” approach to network construction,
with example time error budgets in G.8271.1
• Profile, architecture and clock performance defined by ITU-T,
published May 2014
Challenges
• All equipment in path needs to be PTP aware
• No control of asymmetry in the network
!
Company Confidential 42
G.8271.1: Time Error Budget Example
G.8271.1 Network Reference Points
A, B C D
±100ns
(PRTC/T-GM) cTE uses up 70% of the
±200ns dTE
(random network network equipment budget
variation)
±550ns cTE ±250ns cTE
(node asymmetry, (link asymmetry
±50ns per node) compensation)
Class A T-BCs:
Class B T-BCs:
±250ns
±420ns cTE ±380ns cTE (short term holdover)
(21 nodes, ±20ns per node) (link asymmetry
±150ns
compensation)
(end application)
Company Confidential 44
Relevant Standards for FTS
Published recommendations related to full timing support:
• G.8271 – General considerations for time synchronization
• G.8275 – General network architecture for time synchronization
• G.8275.1 – PTP Profile for full timing support
• G.8271.1 – Network limits for full timing support
• G.8272 – Primary Reference Time Clock Specification
• G.8273.2 – Telecom Boundary Clock Specification
Company Confidential 45
Partial Timing Support
Company Confidential 46
G.8275.2 “Partial Timing Support” Profile
• Why a second time/phase profile?
• Some service providers need to operate time/phase synchronization over
existing networks
• Reduces barriers to entry into LTE-A systems; don’t need to build an
entirely new network
• Allows operation over 3rd party network providers (given appropriate
quality guarantees)
• Requested by four large service providers (Sprint, AT&T, Verizon, T-Mobile)
• Not every network element needs to include a BC or TC
• Much smaller, localised network than for full timing support
• Intended to support both pure PTP case and PTP backup to GNSS
Company Confidential 47
Partial Timing Support Use Cases - 1
PTP backup to GNSS (“assisted partial timing support”, or APTS)
Reference Reference
Point Point
C D
Not all switch/routers on the path between
GNSS T-GM and AP-TSCP contain a T-BC
GNSS
T-GM T-BC
AP-TSC
Features
• Objective is backup to GNSS, i.e. “assisted holdover”
• GNSS monitors PTP service quality and network asymmetry
• PTP can maintain timebase when GNSS is out of service
(e.g. due to jamming or antenna failure)
Company Confidential 48
PTP with Assisted Partial Timing Support
Benefits
• Mutual co-operation between GNSS and PTP
• PTP provides an initial time fix to assist the GNSS during signal acquisition
• GNSS calibrates the PTP asymmetry, and monitors its suitability for service
• PTP can monitor GNSS timing quality, e.g. antenna failure, spoofing, jamming
• Operates over existing networks, including third party access networks
that may not have built-in PTP support
• Profile now agreed, architecture and clock performance still under
definition in ITU-T
Challenges
• Less deterministic path from T-GM to APTSC, because not every
network element assists in the timing flow
• May need constraints on traffic load and span of the packet network
• Little agreement in ITU-T on the scope of the network
49
Partial Timing Support Use Cases - 2
PTP distribution over a local LAN (“partial” or “no timing support”)
Time distribution from rooftop
antenna to multiple small cells,
using PTP over the building LAN
Features
• Objective is to distribute time over a small PTP-unaware (or
partially unaware) network
• Small network, potentially only a single in-building network
• Places GNSS source as close to the end clock as possible
Company Confidential 50
Partial Timing Support Use Cases - 3
Network Bridging
T-GM T-TSC
T-BC T-BC T-BC T-BC
Features
• Objective: bridge between two full timing support networks
• Example: a mobile operator may not own the access network,
and need to bridge across a third party network
Company Confidential 51
PTP with Partial Timing Support
Benefits
• Simple deployment over existing networks
• Operators do not need to own or manage the network
• Can be leased from a third party, e.g. building owner
• Profile now agreed, architecture and clock performance still under
definition in ITU-T
Challenges
• Less deterministic path from T-GM to T-TSC, because not every
network element assists in the timing flow
• Switches/routers not designed with PTP asymmetry in mind, so
device asymmetry is uncontrolled
• May need constraints on traffic load and span of the packet network
• Little agreement in ITU-T on the scope of the network
52
Relevant Standards for APTS and PTS
Published recommendations related to partial support:
• G.8271 – General considerations for time synchronization
• G.8275 – General network architecture for time synchronization
• G.8272 – Primary Reference Time Clock Specification
• G.8275.2 – PTP Profile for partial timing support
54
Prime Objectives
• To allow interoperability between the various phase and time clocks
used in the architecture defined in G.8275
• To permit the distribution of phase and time using PTP over existing
managed, wide-area, packet-based telecoms networks
• To define message rates and parameter values consistent with
phase/time distribution to the required performance for telecom
applications
• To allow the synchronization network to be designed and configured
in a fixed arrangement
• To enable protection schemes to be constructed in accordance with
standard telecom network practices, including the use of multiple
geographically separated reference clocks
• To allow reference source selection based on traceability information
supplemented by local priority, as well as automatic establishment of
topology
55
Key design decisions
• Operates over existing switches and routers, using unicast IP
• Allows use of boundary and transparent clocks at some nodes in the
chain between PTP Grandmaster and PTP Slave
• Cleans up time signal as it passes through the network
• Supports multiple active grandmasters for redundancy
• Mixture of both G.8265.1 and G.8275.1 profiles, with necessary
modifications for time/phase operation
• IPv4/IPv6 support as in G.8265.1
• Unicast messaging and negotiation as in G.8265.1
• Message rates similar to G.8265.1 (Sync/Delay_req 1 to 128 msgs/s)
• Uses end-to-end messaging (Sync/Delay_req) for delay estimation
• ClockClass scheme and ABMCA from G.8275.1
Currently in Last Call: expected publication May 2016 if no objections
56
Rate of Timing Messages
• The rate of timing messages required is dependent on several factors
• Amount of “noise” in the network (e.g. caused by other traffic)
• Local oscillator stability
• Efficiency of clock servo algorithm
• The profile defines the range of message rates permitted:
Message rates Minimum Maximum
Announce 1 message/s 8 messages/s
Sync (and follow-up) 1 message/s 128 messages/s
Delay_Request 1 message/s 128 messages/s
58
Source Traceability
Uses clockClass attribute to determine phase/time traceability:
defaultDS Frequency Time
Phase/time traceability description
clockClass Traceable flag Traceable flag
T-GM connected to a PRTC in locked mode
6 TRUE TRUE
(e.g. PRTC traceable to GNSS)
T-GM in holdover, within holdover specification 7 TRUE/FALSE* TRUE
T-GM or T-BC without time reference since start-up 248 TRUE/FALSE* FALSE
Slave only OC (does not send Announce messages) 255 n/a n/a
Company Confidential 60
PDV - Unpredictable delay is the problem
Timing packets have to fight there way through network devices against traffic
61
Packet Delay Variation
• “Green Light” protocol
• Some (small) number of cars will drive through a town with every traffic
light at green, and arrive at their destination with the least delay
• This “minimum delay” is consistent, and cannot be improved (at least,
not without breaking the speed limit)
• The number of cars achieving this “minimum delay” reduces with
congestion
• Same applies to packets in a network
• Queuing in a network element is like waiting at a traffic light
• The more congestion on the network, the less chance there is to avoid
queuing
Company Confidential 62
Packet Delay Distribution
Frequency of Cluster
occurrence range
Low
congestion
Medium
congestion
High
congestion
• For networks that don’t have BCs or TCs at every node (known
as “Partial Timing Support” (PTS), the network limit will use the
metric pktSelected2WayTE:
Company Confidential 64
Provisional Time Error Budget for APTS
• Provisionally agreed values for pktSelected2WayTE:
• Percentile selection: 0.5%
• Window size: 100s
• Peak-to-peak Limit: 1150ns
±100ns
• Implicit asymmetry budget: 200ns after calibration by GNSS
(PRTC/T-GM)
±1050ns dTE
(random network variation)
±200ns cTE
(GNSS calibration uncertainty)
±150ns
(end application)
±1000ns dTE
(random network variation)
±250ns cTE
(reference switching)
±150ns
(end application)
Paul Norris
Paul.norris@calnexsol.com
+44 (0) 1506-671-416