The Precision Time Protocol allows the use of packet networks like Ethernet for mobile backhaul.
Before telecommunications carriers can leverage cost-effective Internet Protocol (IP) networks to handle the increased bandwidth expected from 3G and 4G services, they need to know the issues surrounding time synchronization and the solution provided by IEEE1588v2. Officially named IEEE 1588-2008, this standard addresses network-based timing and synchronization.
Synchronization is a fundamental technology building block of today’s service provider networks. The performance of the traditional wired T1/E1, Sonet, and synchronous digital hierarchy (SDH) networks is based upon constrained clock control theory and phase-locked loop (PLL) design. This type of synchronization signal is a closely bounded frequency. The synchronization signal delivery system used in today’s practice is based on the physical layer (PHY) of the transmission system providing a clock signal.
Electronic circuits discipline the signal to provide the frequency accuracy, stability, phase movement control, wander filtering, and edge jitter (phase noise) levels specified in volumes of standards documents. Each “end customer service” has a specific set of performance requirements detailed in these documents. The various international standards bodies define this performance, which is well understood.
There are two mobile wireless network synchronization schemes. The first uses the same technology as a wired network. This standard is used to provide frequency-division duplex (FDD) radio-based mobile wireless network synchronization signals for ingress/egress of data and accurate radio frequency. For example, WCDMA FDD is the most popular method of radio air interface used in GSM systems.
The second is synchronization of time-division duplex (TDD) radio-based mobile wireless networks. This radio technology requires frequency accuracy, phase alignment, and, in certain cases, time alignment between all basestations in the cell network. CDMA, cdma2000, Mobile WiMAX 802.16e, and Long-Term Evolution (LTE) are all examples of TDD radio systems. The traditional wired “circuit” oriented synchronization signal delivery system cannot be used because there is no phase or time relationship between signal termination points on the clock distribution network.
The four requirements for mobile wireless network synchronization depend on having an inter-basestation aligned timing reference. This is essential to guarantee transport channel alignment for handoff and guard band protection. 3GPP-specified FDD systems require frequency accuracy better than 50 parts per billion (PPB). The 3GPP TDD systems require an inter-basestation time alignment of 2.5 µs to IS-97’s 10 µs in addition to the 50-ppb frequency accuracy. The IEEE 802.16e mobile WiMAX requirements are 20 ppb of frequency accuracy and 1 µs of phase alignment. Ensuring the fulfilment of these requirements reduces the call drop rate and improves the quality of services by decreasing packet loss. This is how it works today and will continue to work when new backhaul techniques are implemented.
So what is the big deal if this technology is so well understood and bounded by standards? T1/E1 circuit-based technologies are dedicated to the individual service they are delivering. Each service end point bears the cost of the entire circuit path 24 hours per day, 365 days a year. Considering the scale of billions of circuits worldwide needed for 3G/4G services, the costs of service delivery present a significant barrier to profitably providing higher-bandwidth service offerings.
For example, these circuits can consume up to 40% of the total operating expense for a mobile wireless operator. Cell-phone customers pay for this in their bill. This market dynamic, plus the competitive nature of the industry, is driving the move to the shared resource of packet-based Ethernet backhaul. The cost per megabit ratio comparison of a T1/E1 circuit versus Ethernet-based packet backhaul averages 6 to 1.
With these metrics, it is easy to understand the motivation of the service provider to migrate to the cheaper Ethernet transport. The question of how to outfit routers and switches with IEEE1588v2-based semiconductor master and slave semiconductors remains. The provision of synchronization for future mobile networks will pose new challenges triggered by the following factors:
- • New mobile networks will offer new content-rich services (e.g., IPTV, VoIP), demanding higher capacity and faster backhaul links.
- • New mobile network technologies such as mobile WiMAX and LTE are hungry for bandwidth. They are also capable of a much higher degree of radio spectrum efficiency than today’s systems. This is key, particularly for licensed spectrum usage.
- • The PHY transport for traditional systems will be packet-based, most likely using Ethernet. Carriage of voice and data traffic in the 3GPP UMTS R5 and above specifies use of IP for data and that compressed voice be carried via VoIP versus being carried over ATM today.
- • There will be an increased number of cell sites due to the need for a higher signal-to-noise ratio to support the increase in bandwidth delivered to the users. Increasing the bit density of the radio channel to pack more traffic requires more nodes and different timing distribution techniques.
- • Ethernet is the only candidate for transport technology for future backhaul networks given the lower cost of the facility and switching equipment. Ethernet is inherently asynchronous.
The migration toward a packet-based transport network poses the challenge of providing the level of synchronization requirements defined in the 3GPP, 3GPP2 (LTE), and IEEE 802.16e (WiMAX) specifications.
The transmission of the clock information over a packet network eliminates the need for alternative mechanisms, such as GPS or prohibitively expensive oscillators placed at the receiving nodes. This provides significant cost savings in network equipment as well as in ongoing installation and maintenance. This synchronization solution transmits dedicated timing packets, which flow along the same paths with the data packets, reducing the cost of synchronization and simplifying implementation.
Master and slave semiconductors based on the IEEE1588v2 Precision Time Protocol (PTP) are available now to carry the synchronization data. The IEEE1588v2 protocol is fully compatible with all Ethernet and IP networks. Additionally, the protocol is designed to enable a properly designed network to deliver frequency and phase or time with precision rivalling a GPS receiver.
An IEEE1588v2 protocol implementation can supply FDD and TDD radio systems and CES-based (circuit emulation services) transport systems with the synchronization signals they require. This greatly reduces the costs of clocking all wireless basestation equipment.
- Although IEEE1588v2 systems add a small amount of additional traffic to the network load, they have several advantages. First, they work in the data path, the most redundant and resilient part of the network, resulting in “always on” operation. Next, multiple transmission paths reduce redundant clock system costs. They also use a single synchronization session for all basestation traffic.
- Furthermore, they support legacy systems in mobile networks, where Circuit Emulation Service is employed to carry both TDM traffic (2G) and ATM traffic (3G). They support any generic packet-based transport (such as IP RAN). And, they feature configurable packet rates for network conditions to maintain accuracy (e.g., packet rate less than 1 pps to greater than 100 pps)
The IEEE1588v2 protocol has been extensively tested in these scenarios and has proven viable. Yet it is very important for service providers and equipment vendors to understand IEEE1588v2 is a protocol only. The clock recovery algorithm is the synchronization solution.
- IEEE1588v2 employs a two-way methodology, where packets are sent back and forth from the clock master to the clock slaves. This overcomes high-amplitude, ultra-low-frequency wander that defeats other methods such as adaptive clock recovery techniques. Also, the standard is virtually independent of the physical media and can flow over low-speed twisted-pair, high-speed optical fiber, wireless, or even satellite links without requiring equipment design modifications.
- Additionally, it isn’t limited to TDM circuit emulation like the in-band solutions, but it can support CES better than adaptive clocking by distributing a precise network clock to every inter-working function node in the system. It also can be used for any pure, packet-based network, providing synchronization for future backhaul networks to be deployed by mobile operators.
- The standard can distribute time/phase, frequency, or both. Telecom operators can use it to sell a synchronization service to customers (residential, wireless operators, etc.). It’s resilient because a failed network node can be routed around. It also is resilient because the synchronization can come from one or more grandmaster clock nodes.
- IEEE1588v2 packets fully comply with Ethernet and IP standards and are backward compatible with all existing Ethernet and IP routing and switching equipment. There is no requirement for intermediate switches or routers to be IEEE1588v2 aware. They see these timing packets as normal packet data.
The protocol calls for synchronization packets with time stamps to be sent from master clocks to all slave clocks and for individual slave clocks to send time-stamped packets to the master (Fig. 1). The clock grandmaster maintains a time base locked to a primary reference clock and establishes a separate synchronization session with each of the slaves it serves. The master and slave exchange timing packets according to the syntax of the IEEE1588v2 protocol.
This process provides the timing clock recovery algorithm with the time stamps it needs to precisely recreate the master time base. From this time base, the synchronization signals used by the network equipment are synthesized. The timing clock recovery algorithm filters most noise, packet queuing delay, and propagation delay created by the transport network.
The semiconductor suits all wired and wireless service provider network applications. Typically, each synchronization session produces approximately 15 kbits/s of traffic (in each direction, uplink and downlink), which represents a small portion of the total capacity, usually between 100 Mbits/s and 1 Gbit/s, in a backhaul network.
In April 2005, the world’s first IEEE 1588 proof of concept field trial began using IEEE1588v2 slave and master products from Semtech. The company’s ToPSync slave and master devices are fully integrated 1588 devices with built-in protocol support, IP stack, Ethernet frame control, clock recovery, and output generation.
Figure 2 illustrates the test bed of the live trial. Two boards, acting as pre-standard IEEE1588v2 master and slave, were connected to a live metro Ethernet backbone operated by a major U.S. carrier. The master was locked to an atomic clock. The synchronization packets travelled through the metro Ethernet before reaching the slave board, which was connected to an Agilent Omniber 718 measuring the time interval error.
Figure 3 shows the results of the measured maximum time interval error (MTIE). The top picture shows the MTIE against the G.823 synchronization interface mask for an E1 circuit as used in a non-North American GSM basestation. A North American GSM basestation would have a G.824 or T-1 mask target. The measured MTIE is well below the mask, confirming the excellent clock recovery capabilities of the algorithm.
Many carriers have tested the TopSync technology in live field tests, including an extensive European test completed by Vodafone that was presented at the November 2007 International Telecoms Synchronization Forum in London. The testing suite Vodafone used followed the ITU G.8261 specification with some special implementations unique to the company. These results confirmed the performance of the device.
Designing 1588 Into a System
When designing a 1588 solution into a network node as a slave device, the most critical thing to keep in mind is the time stamp located in the media access control (MAC) layer as described in the IEEE 1588-2008 standard. This 64-bit value is the “instant in time value” of the transmitting node’s time base for a 1588 event packet.
As the master and slave devices exchange packets per the IEEE 1588 message syntax, the differences in flight time or packet delay variation period between time stamps will affect the slave clock recovery systems. The system’s architects and designers need to control local system propagation from ingress port to 1588 slave input.
Whatever local system delay does exist needs to be deterministic and as input/output time symmetrical as possible. These design rules minimize the propagation effect from local systems on packet delay variation seen at the ingress of the slave clock. The same architecture rules apply in designing in a master device, even though the master does not perform clock recovery. The outbound delay imposed on the sync messages will be part of the overall packet delay variation seen by the slave. The greater the determinism in this path, the less its impact on the slave’s clock recovery.
When the system is ready for system testing, the performance template is the ITU G.8261 recommendation titled “Timing and Synchronization aspects in Packet Networks.” This document provides a model of a potential service provider carrier Ethernet network with loading parameters as well as target performance masks.
This testing recommendation provides a suite of timed tests with a change in load on both paths, master to slave and slave to master. The test suites and performance masks to date only apply to delivery of frequency from the 1588 slave device. This form of testing is a reasonable means to determine basic system behaviour under closely controlled conditions. Figure 4 shows the overall testing template from the G.8261 document.
It is important to note that G.8261 is a work in progress and as yet does not contain the specific performance targets or the network testing models for IEEE 1588-2008. Figure 4 is the network model being contemplated by the ITU.
Synchronization is an important part of today’s wireless backhaul networks. With the IEEE1588v2 protocol, carriers can achieve synchronization with accuracy matching that of alternative solutions without the cost or need to build overlay networks required by those solutions. This standard provides an essential technology that allows carriers to efficiently deploy IP networks for their wireless backhaul driving down the infrastructure costs of deploying 3G and 4G networks. With field trials showing interoperability and performance, IEEE1588v2 is a proven solution for IP synchronization.
Patrick (Pat) Diamond has more than 30 years of experience in the high-tech field.
He was instrumental in the creation of the IEEE 1588-2008 Precision Time
Protocol working group in 2004. He represents Semtech to the ITU-T,
IEEE-USA and has been with Semtech for more than eight years. He can be reached at email@example.com.