IEEE 1588 Precise Time Protocol:
The new standard in time synchronization
WHITE PAPER
Overview – Precise Time and the Trouble with Ethernet
As network computing becomes more complex and the world more intercon-nected, the need for more precise time synchronization has greatly increased. Network time protocol (NTP) and IRIG time code have been the most common protocols governing time transfer, but the recently developed IEEE 1588 Precise Time Protocol (PTP) now prom-ises to revolutionize time synchroniza-tion by improving accuracy and reducing cost. While certain other precise sync protocols require significant investment in hardware and cabling, PTP makes highly precise timekeeping possible using the most widely deployed medium for network connectivity – Ethernet. While Ethernet has proven to be a ubiq-uitous and inexpensive medium for con-nectivity, it has not been well-suited for applications requiring precise synchro-nization. By nature it is nondeterminis-tic, which creates difficulty for real-time or time sensitive applications that require synchronization. On an NTP-based LAN, network devices and com-puter operating systems add latency and jitter that reduces synchronization accu-racy to 1 to 2 milliseconds. Applications requiring greater accuracy have often had to deploy separate cabling systems and dedicated clocks – the IRIG B proto-col, for instance, requires a dedicated system of coaxial cables to carry timing signals directly between IRIG B clocks separate from any data network.
PTP overcomes the Ethernet latency and jitter issues through hardware time stamping at the physical layer of the network. The result can be an unprece-dented accuracy in the 10 to 100 nanosecond range that is achieved using an Ethernet network to carry the timing packets, allowing for remarkable cost savings. This paper examines how PTP works, compares its effectiveness to other protocols, and explains why it is so cost effective.Choosing a Standard: NTP vs. IRIG
vs. PTP
NTP has been the most common and
arguably the most popular synchroniza-
tion solution because it performs well
over LANs and WANs and is relatively
inexpensive to implement, requiring little
in the way of hardware. And while NTP
should be able to deliver accuracy of 1 –
2 milliseconds on a LAN or 1 – 20 mil-
liseconds on a WAN, this is far from
guaranteed network-wide. This is largely
due to the use of switches and routers
on LANs and WANs and the fact that
many NTP clients run on non-real-time
operating systems such as Windows or
Linux, that were not designed for time
keeping accuracy and therefore rarely
deliver it. On Windows, for instance, one
can often witness clock corrections of 10
– 50 milliseconds because the system
was busy performing tasks it deemed
more important than timekeeping.
IRIG time code provides increased accu-
racy – up to 1–10 microseconds – and is
often used where precision timing is
mission critical: in military, aerospace
and power utility instrumentation. But
improved accuracy comes at a cost. IRIG
systems eschew Ethernet in favor of
dedicated coaxial timing cabling
between dedicated hardware clocks,
which are not only an added expense but
which create an extra burden on the
physical infrastructure of the facility.
PTP, on the other hand, offers the cost-
effectiveness of NTP by using existing
Ethernet LANs, and it exceeds the accu-
racy of the IRIG clocks. PTP can coexist
with normal network traffic on a stan-
dard Ethernet LAN using regular hubs
and switches, and yet provides synchro-
nization accuracy to the sub microsec-
ond level. With the addition of IEEE 1588
boundary clocks or transparent switch-
es, 20-100 nanosecond synchronization
accuracy is achievable. The key to this
caliber of performance is hardware-
assisted time stamping.
Figure 1Comparison of the synchronization requirements of NTP, IRIG time code and IEEE 1588
GLOSSARY
Grandmaster Clock:
Within an IEEE 1588 sub domain, a
Grandmaster clock is the ultimate
source of time for clock synchroniza-
tion using the IEEE 1588 protocol.
Boundary Clock:
Generally a switch with more than a
single IEEE 1588 equipped port, which
is a slave on one port and a master on
all others.
IEEE 1588 Ordinary Clock:
A IEEE 1588 clock with a single port.
Precision Time Protocol (PTP):
The protocol defined by the IEEE 1588
standard.
Transparent Clock:
In IEEE 1588 terminology, it is a switch
that compensates for its own queuing
delays. Neither master or slave.
How PTP Works: Hardware-Assisted Time Stamping
The two primary problems that must be overcome in network timekeeping are oscillator drift and time transfer latency.Regardless of the protocol used, oscilla-tor drift can be mitigated by using higher quality oscillators and by deriving time from a more accurate source, such as GPS. The time transfer latency problem is more difficult and is two-fold in
nature: there is latency associated with processing of time packets by the oper-ating system, and network latency creat-ed by the hubs, switches, cables and other hardware that exist between
clocks. It is in the area of reducing oper-ating system latency that PTP is most successful.
PTP combines time stamping units (TSU) with an innovative method for exchanging time stamp detail between master and slave clocks. A TSU placed between the Ethernet Media Access Control (MAC) and the Ethernet PHY transceiver sniffs both inbound and out-bound traffic and issues a time stamp when the leading bits of a 1588 packet are identified, precisely marking the arrival or departure of 1588 time pack-ets.
In order to estimate and mitigate operat-ing system latency, the master clock periodically sends a Sync message
based on its local clock to a slave clock on the network. The TSU marks the exact time the Sync message is sent,and a Follow_Up message with the exact time information is immediately sent to the slave clock. The slave clock time stamps the arrival of the Sync message,compares the arrival time to the depar-ture time provided in the Follow_Up and is then able to identify the amount of latency in the operating system and adjust its clock accordingly.
Network related latency is reduced by measuring the roundtrip delay between master and slave clock. The slave peri-odically sends a delay request message (Delay_Req ) to the master clock which issues a delay response message
(Delay_Resp ). Since both messages are precisely time-stamped, the slave clock can combine this information with the detail from the Sync and Follow_Up mes-sages to gauge and adjust for network-induced latency. The protocol for exchanging precise time stamps is detailed in the graphic below.
Determining Target Accuracy
Choosing broadcast intervals and oscillator types
In PTP, the desired accuracy of timing determines how often sync messages are broadcast and what kind of oscillator is used. More frequent broadcasts result in more accurate sync, but also in more network traffic, although the bandwidth used is extremely small. Higher quality oscillators also result in more accurate sync. It may be tempting to try to achieve target accuracy more economically by using a lower quality oscillator while increasing broadcast frequency, but this is unadvisable. Low quality oscillators lack the stability needed to achieve high precision with PTP, so shortening the broadcast interval offers diminishing returns.
Accuracy is also a function of the IEEE 1588 master clock, called the grand-master, that is the ultimate source of time on the network. Grandmasters are typically referenced to GPS so that they are both very stable and very accurate.Accuracy to UTC, TAI or GPS is typically 30 nanoseconds RMS or better . By start-ing with such an accurate clock with an absolute time reference, time on a PTP-enabled network can be very well syn-chronized. A quality grandmaster also provides other measurement features to characterize the latency and jitter char-acteristics of network elements and to measure slave accuracy relative to the grandmaster . Ironically, a quality grand-master clock also supports NTP and IRIG timing requirements. This makes sense as there are legacy systems in place that will need synchronization as well as new PTP deployments. Selecting other hardware
PTP has been readily implemented to work on Ethernet LANs where router buffer delay and switch latencies under-mine the accuracy of time transfer .Consequently, the other variables in achieving target time transfer accuracy are the use of hubs versus switches,boundary clocks or transparent clocks. When evaluating cost versus packet
delay, hubs are generally considered the best value. Switches may move packets more efficiently, but reduce accuracy because of packet jitter .
Figure 2Hardware set up for a system deploying IEEE 1588. The time stamp unit monitors packets at the
MII layer, between the Ethernet MAC and PHY, and issues a precise time stamp when a 1588 timing packet
passes through it.
Figure 3The sequence of packets used to transfer time from the 1588 master to a 1588 slave. Sync packets are stamped as they leave the master (T1)and arrive at the slave (T2). Follow-up packets communicate the sync packet departure time to the slave. Delay response packets are stamped as they leave the slave (T3) and when they arrive at the master (T4). Sync and follow-up packet pairs are sent by the master on a periodic basis, as are the delay request and delay response packets. The formula for slave clock correction= 0.5 (T1-T2-T3+T4).
The boundary clock is a multi-port switch containing one port that is a PTP slave to a master clock, while the other ports are masters to downstream slave clocks. Boundary clocks provide a decent method for regulating time to a number of subnets. But using cascading boundary clocks accumulates non-linear time offsets in their servo loops, result-ing in unacceptable degradation of accu-racy.
The transparent clock is another poten-tial hardware option for the PTP-based network. This is a PTP enhanced switch which modifies the precise time stamps in the Delay_Resp and Follow-Up mes-sages to account for receive and trans-mit delays within the switch itself. The result is improved sync between slave and master clocks. But the transparent clock can also create security issues when the original packet crypto check-sum doesn’t match the final packet arriving at the slave.
The graphic below shows overlaid his-tograms of the clock-offset errors rela-tive to each device type. Inconsistent device latency and jitter results in increased clock offset errors represent-ed by the increased dispersion in the data.Enabling Cost-Effective Real-time
Applications
In addition to extraordinary accuracy,
PTP also facilitates the operation of
real-time or near real-time data acquisi-
tion systems over Ethernet. Traditionally,
these systems have depended upon trig-
gering techniques and delay compensa-
tion independent of the data network in
order to coordinate synchronized data
collection. But PTP eliminates the
necessity – and the cost – of such meas-
ures.
Consider the example of a real-time
control system that is set up over
Ethernet with a variety of independent
sensors. Each sensor uses an onboard
1588 clock slaved to a grandmaster
clock and is synchronized with the other
sensors on the network to an accuracy
of 1 microsecond. Every 5 milliseconds,
each sensor takes a measurement and
sends the information back to the con-
troller, using the same LAN that is coor-
dinating the synchronization of the sen-
sors. The result: data is acquired syn-
chronously from the entire system at
precise intervals, without using trigger-
ing techniques or delay compensation.
There is no trigger variation caused by
trigger propagation delay between the
near and far sensors. Because the 1588
modules are relatively inexpensive and
can easily be added to the Ethernet-
based network, the system allows for
cost-effective real-time or near-real
data acquisition.
PTP and Military and Aerospace
Applications
Because of its precision, cost-effective-
ness and ease-of-use, PTP can be
expected to be widely used in military
and aerospace applications. Systems
that are deployed in military theater sit-
uations to identify enemy threats can be
constructed from ad hoc Ethernet net-
works that connect a variety of sensors.
Sonar systems in submarines can
improve acoustic intercept and ranging
accuracy by deploying PTP time stamp
technology close to the sensors. And
PTP is replacing IRIG systems in aircraft
flight test because it provides tighter
sync and eliminates the need for addi-
tional cabling.
The Future of PTP
PTP has justifiably received considerable
attention since its introduction in 2002,
and its influence is growing. A variety of
vendors are producing hardware that
supports PTP, including Intel, which has
recently embedded an IEEE 1588 TSU
into one of its networking microproces-
sors. The next version of the 1588 proto-
col is currently being defined and is
expected to increase accuracy even
more. And improved fault tolerance and
compatibility with SNMP are expected to
enhance PTP interoperability with exist-
ing network infrastructure. With its
nanosecond accuracy, ease-of-deploy-
ment, and cost-effectiveness, PTP is
poised to transform the landscape of
time-synchronized applications in any
number of fields.
To learn more about IEEE 1588, visit
https://www.wendangku.net/doc/2718179391.html,/
Figure 4Histograms showing the precision of time synchronization between IEEE 1588 master and slave using three different kinds of Ethernet connection devices.
?2005 Symmetricom. Symmetricom and the Symmetricom logo are registered trademarks of Symmetricom, Inc. All specifications subject to SYMMETRICOM, INC.
2300 Orchard Parkway San Jose, California 95131-1017
tel: 408.433.0910fax: 408.428.7896
info @https://www.wendangku.net/doc/2718179391.html,