IEEE 1588 Precision Timing Protocol (PTP)

By Jeremy Bennington, Symmetricom

TELECOMMUNICATIONS NETWORKS ARE RAPIDLY MIGRATING from circuit switched to packet technologies to converge services onto a single network, deliver larger amounts of bandwidth, and lower costs. One of the key challenges in this migration is the need for precise frequency synchronization in vital applications such as circuit emulation services (CES) and wireless base station backhaul. Highly accurate timing has always been a cornerstone of the network and is required in the packet networks to enable existing and future applications.

Several technologies exist to provide the precise frequency synchronization needed by these applications. Adaptive clock recovery (ACR) uses the traffic flow itself to reproduce the network clock at the downstream element. GPS-based synchronization receivers can be installed at remote sites to provide a direct synchronization reference. The emerging IEEE 1588 Precision Time Protocol (PTP) standard for a packet-based synchronization promises a higher level of performance and lower cost by leveraging the existing infrastructure.

Pocket Network TIMING ISSUES
In order for service providers to realize the economic benefit of a converged packet infrastructure they must have the ability to deploy existing services without degrading the quality. Two of the most prevalent and profitable services offered today are T1 and E1 TDM leased line services and wireless base station backhaul. While these two services are extremely profitable they are also one of the most challenging services to deploy in a packet network. CES over packet technologies emulate T1, E1, and other services by encapsulating TDM traffic into packets for transport across the packet switched network (PSN) and restoring the TDM traffic at the egress. At first glance it's not difficult to deliver TDM data over IP networks. The challenge arises in the critical nature of timing and synchronization.

Conventional circuit switched networks inherently distribute timing throughout the network based on very accurate primary reference clocks. Building integrated timing supplies (BITS) in each central office distribute and stabilize the digital heartbeat that keeps each switch in sync. However PSNs do not have a timing structure. By its very nature the Ethernet is nondeterministic, which creates problems for time-sensitive applications that require precise synchronization. The PSN imposes delay and delay variation to packets that traverse the network. In order for T1, E1 and wireless backhaul CES to reliably operate, synchronization is needed at the end points of the service to remove the packet delay variation (PDV).

Challenging TIMING REQUIREMENTS
CES and wireless backhaul are the two most prominent applications where timing issues have emerged to date. T1 standards specify a maximum time interval error for the customer interface, measuring the maximum time variation of the clock at the transport level of 8.4 µs over 900 seconds and 18µs over 24 hours. The maximum time interval error at the synchronization level is 1 µs over 2,000 seconds and 2 µs over 100,000 seconds.

Mobile base stations have equally critical timing requirements. A frequency accuracy of ±50 parts per billion (ppb) is needed to achieve successful handoffs and maintain voice and data services. When the handoff between two base stations occurs, the mobile phone must rapidly switch from the frequency and/or time of the current base station to the target of the new base station. If a mobile phone is unable to react quickly enough to synchronization errors between base stations, the result will be a dropped call. Most GSM base stations deployed today recover their synchronization from the TDM network delivered T1 or E1 service. If the T1 or E1 is delivered by CES then the packet network must be enhanced to include synchronization. Similarly if the base station uses native IP backhaul where CES encapsulation is not needed, there is still a need to provide synchronization for the base station frequency.

Timing and SYNCHRONIZATION METHODS
Adaptive Clock Recovery (ACR) is a best effort method utilized today to provide timing and synchronization for CES over PSN applications. This method relies upon the fact that the source is producing bits at a constant rate determined by its clock. When the bits arrive at the destination, they are separated by a random component know as packet delay variation (PDV). Adaptive clock recovery involves averaging these bits in order to even out their gaps and negate the effects of PDV. The weakness of ACR is that it requires an expensive oscillator at the source, and field performance is uncertain under exposure to high levels of PDV present in live networks.

An alternative to ACR is to install a GPS receiver at each base station and use it as a stable clock reference for re-timing the CES packets between the CES modem and the base station T1/E1 input. The timing signal received by the base station is retimed to be precise and stable. The disadvantage of GPS-based retimers is that they involve a substantial cost and implementation burden. First, there is the need to equip each base station with a GPS receiver, involving a significant capital cost. With several million base stations in the world, the required investment is substantial. Another concern is that the existing GPS may not be an acceptable solution for all sites since GPS signals may be weak indoors or in metropolitan areas. Moreover, some wireless operators internationally may not want to use a GPS signal controlled by the United States.

Another alternative is to integrate GPS directly into the base station equipment or deploy stand-alone rubidium clocks. Rubidium based oscillators provide a highly robust solution that has been proven to meet the 50 ppb requirement over the full service life of the equipment. Quartz oscillators, on the other hand, are subject to higher native aging rates and warm-up/restabilization characteristics that make it difficult to assure compliance to the 50 ppb requirement for more than six to 12 months.

Utilizing PSN for TIMING AND SYNC
All of these existing timing methods involve considerable capital investment for hardware at a large number of customer sites or base stations around the world. For these reasons, telecommunications providers have been seeking an alternative that would eliminate these expenses by making it possible to deliver timing and synchronization over the packet-based network. Many have looked at Network Time Protocol (NTP), the most popular protocol for time synchronization over LANs and WANs. NTP, however, currently does not meet the accuracy requirements for CES and base station timing and synchronization. The problem is that NTP packets go through the Ethernet PHY and Media Access Control (MAC) layers in the switch like any other packets so timing is not addressed until the packets reach the software stack. The timing signals are thus delayed by an indefinite amount depending on the operating system latency.

The Precision Timing Protocol (PTP) or IEEE 1588 is an emerging standard that addresses the weaknesses of current NTP implementations and provides the ability to deliver timing and synchronization over PSNs. The basic difference between PTP and NTP is that PTP time stamping is implemented in hardware as shown in Figure 1. A time stamping unit (TSU) is placed between the MAC and PHY to sniff both inbound and outbound traffic and issues a precision time stamp when the leading bits of an IEEE 1588 PTP packet are identified.

How IEEE 1588 PTP WORKS
PTP (IEEE 1588) utilizes clocks configured in a tree hierarchy. Master clocks can be installed in an existing building integrated timing supply (BITS) located in master switching centers, and simple slave devices can be installed in remote base stations. The master clocks send messages to their slaves to initiate synchronization. Each slave then responds to synchronize itself with the master. Incoming and outgoing PTP packets are time stamped at the start of frame of the corresponding Ethernet packet. The protocol then exchanges information between the master and slave using IEEE 1588 message protocol.

These messages are used to calculate the offset and network delay between time stamps, apply filtering and smoothing, and adjust the slave clock phase and frequency. This sequence is repeated throughout the network to pass accurate time and frequency synchronization. IEEE 1588 networks automatically configure and segment themselves using the best master clock (BMC) algorithm. The BMC enables hotswapping of nodes and automatically reconfigures the network in response to outages or network rearrangements.

In order to estimate and mitigate operating 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 containing 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 departure time provided in the follow up message, and then is able to identify the amount of latency in the operating system and adjust its clock accordingly.

Network related latency is compensated for by measuring the roundtrip delay between the master and slave clocks. The slave periodically sends a delay request message to the master clock and the master clock issues a delay response message. Since both messages are precisely time-stamped, the slave clock can combine this information with the detail from the sync and follow up messages to gauge and adjust for network induced latency. The protocol for exchanging precise time stamps is detailed in Figure 2. While network asymmetries and PDV will exist in the PSN, PTP (IEEE 1588) can be deployed to minimize these effects well enough to provide synchronization for CES and wireless backhaul.

Deployment Issues
PTP (IEEE 1588) grand master servers should be located as close as possible to network end points. This will improve timing accuracy by reducing the number of hops between the grand master server and the slaves. This approach is very efficient since it removes the need to provide tight timing over the entire network by providing the synchronization locally. As shown in Figure 3, different types of network elements can introduce widely varying amounts of PDV into PSNs.

Most service providers will be able to leverage their existing investment in BITS that are used to provide synchronization for both wireline and wireless networks. IEEE 1588 blades can be integrated with the existing BITS architecture, sharing a GPS reference and multiple input references to provide a high level of holdover stability as shown in Figure 4. Internal oscillators hold synchronization and timing in case the GPS signal is unavailable. When the GPS signal is lost, the clock switches to holdover mode and relies on the free-run performance of the internal oscillators. Integrating IEEE 1588 servers with the existing BITS can also reduce management requirements.

IEEE 1588 VERSION 2
The movement towards IEEE 1588 in the telecommunications industry can be expected to increase with the release of version 2 of the 1588 standard, which is due in the first half of 2007. Version 2 increases the maximum number of timing packets from 1 per second now to 30 to 40 per second with the new version. Increasing the number of packets per second reduces the impact when a packet is discarded or has a long delay. This has the effect of improving accuracy and making it possible to reduce the cost of slaves by using less expensive oscillators. Version 2 also features a unicast transmission mode in addition to the multicast transmission mode used in version 1. The addition of unicast transmission makes it possible to tune each client independently, which improves timing accuracy.

A parallel effort to IEEE 1588 version 2 is an improvement to NTP. This investigation has just begun, but may leverage the hardware implementation of PTP and use the existing NTP protocol. With improvements to NTP and the new IEEE 1588 protocol, the packet network is well positioned to support existing and future applications.

Wireline and mobile backhaul networks are migrating to packet networks to converge their services onto a single network, deliver larger amounts of bandwidth, and reduce operating costs. Wireline operators need to meet timing and synchronization requirements for CESs and other services over PSNs that isolate remote elements from their former source of synchronization as shown in Figure 5. Mobile operators need to ensure they can support the synchronization accuracy needed to avoid dropped calls and maintain Quality of Service (QoS). The PTP protocol based on the IEEE 1588 standard uses the existing Ethernet distribution networks and the existing BITS infrastructure to deliver timing and synchronization that meets the requirements of these and other applications.

Jeremy Bennington is a Senior Business Development Manager at Symmetricom, San Jose, CA. He can be reached by e-mail at