# Case Studies

Long Distance White Rabbit High Accuracy time over fiber link

Print

The White Rabbit (WR) link covers an approximate distance of 1,350 km (840 miles) and was deployed in collaboration with Optiver US, a financial company, to connect Chicago and New Jersey trading locations. This link is formed by six long-distance White Rabbit hops using WR-Z16 and WR-ZEN TP devices connected by a combination of DWDM and SyncE-compliant transponders over a public telecommunication fiber network.

Our technology integrates the White Rabbit protocol, the basis for the new high accuracy profile in the IEEE 1588-2019 (PTPv2.1)standard for a precision clock synchronization protocol for networked measurement and control systems.

This technology has become a reference for different sectors thanks to its unprecedented level of accuracy that outperforms current GNSS-based timing solutions, offering a suitable backup solution to deploy time dissemination networks. In the last few years, the White Rabbit technology has been adopted in the financial sector to deploy plug & play local area synchronization and metro area links connecting different data centers in financial hubs.

Although the performance of White Rabbit long distance links has been previously validated, this deployment sets a new distance record while integrating new resiliency and interoperability features using the latest WR-Z16 and the WR-ZEN TP devices.

Figure 1: White Rabbit long-distance link topology

In a series of experiments, the link accuracy and precision were first validated setting a loopback with three long-distance hops, covering an approximate distance of 800 km (500 miles). In this case, the link was measured using an Agilent 53210A time interval counter for a five-day period. This experiment was intended to validate the feasibility of deploying White Rabbit links using commercial SyncE-compliant transponders and commercial telecommunication networks based on DWDM technologies.

Figure 2: Offset measurement for a loopback WR link over the first hop of the long-distance link

This measurement confirmed the sub-nanosecond precision on a loopback and made it possible to perform network effect calibration to minimize the residual offset caused by the link’s asymmetry. In this case, a mean offset of 112 ps, a standard deviation of 139 ps and a peak-to-peak difference of 880 ps were obtained. Additionally, the offsets followed a gaussian distribution with no daily trends impacted by temperature or humidity.

This result proves the ability of high accuracy time over fiber dissemination to fulfill the most demanding telecom requirements (class D telecom boundary clocks) and corroborates one of the conclusions from the Analyzing a More Resilient National Positioning, Navigation, and Timing Capability report released by the RAND Corporation.

White Rabbit can support time transfer with accuracy that substantially exceeds the needs of almost all users; it is better than GPS. Therefore, this method is a strong candidate for backing up GPS time transfer for users that require atomic clock accuracy and for serving as a ‘national backbone’ for time so that secondary users, such as cellular networks, can perform to the limits of its own subdomain without suffering additional inaccuracies of its own master clock. Less accurate methods, like ordinary PTP, could provide timing to the vast number of other users, like mobile and cellular users.

In a second test, the whole link was deployed using GNSS receivers in both ends of the link. The first GNSS receiver was used as the time reference in one end of the link (located in New Jersey) and the second GNSS receiver was used in the second end of the link to compare to the time reference (located in Chicago). Both GNSS receivers are the same model and have a 15 ns RMS jitter specification. In the second location, the local GNSS reference was compared to the remote time reference originating from New Jersey through the long-distance White Rabbit link.

Figure 3: Offset measurement for the long-distance WR link using GNSS as its reference in NY and a second GNSS receiver in Chicago for the comparison.

This resulted in a mean offset of 2.98 ns, a standard deviation of 10.4 ns and a peak-to-peak of 83.3ns. It is noteworthy that due to the time interval counter, some of the samples were filtered. This effect is shown on the histogram but is considered negligible for the purpose of validating the feasibility of the link. Additionally, the White Rabbit protocol automatically corrects daily effects due to temperature or humidity changes in the link, which are not observed in the measurement even when the GNSS receivers are located more than 1,000 km away from each other. This does not represent the real White Rabbit link error but indicates the limitation of using a GNSS receiver to accurately measure the link accuracy.

Figure 4: Offset histogram for the long-distance WR link.

As can be estimated from the measurements, the previous test and the GNSS specifications, the White Rabbit link maintains an approximately ±1 ns precision. In fact, the results show that the GNSS receivers are outperforming their specifications by almost 50%. The final accuracy is influenced by the GNSS receiver calibration. This link can be calibrated using network effect techniques (as shown in the previous experiment) or using the GNSS receivers themselves. This experiment complements the results obtained in the U.S. Department of Transportation GPS backup demonstration showing the White Rabbit technology (which was catalogued as the most accurate alternative technology for time distribution) in a real telecommunications network scenario with very long distances.

Additionally, it justifies the industrial need to deploy high accuracy time distribution when providing an alternative to GNSS-based timing, as it provides a next generation technology that can benefit other PNT systems and several critical infrastructures that rely on time synchronization for daily operations. Beyond this, White Rabbit is demonstrated as a disruptive technology that can be used to measure the performance of GNSS in various locations to develop a warning and backup system, to connect high stability time references (such as atomic clocks) for a suitable long-term ground-based backup to GNSS systems or to connect national metrology institutes around the world to compare different time scales.

This long-distance link represents a new milestone in ultra-accurate time transfer over existing telecom networks that makes it possible to cross-validate different linked references or to accurately synchronize them for coordinated actions at tight relative timing. Furthermore, the deployment through an existing telecom network proves its feasibility at affordable cost and opens the door to new disruptive applications. Atomic clocks have evolved to represent very stable references, but it has been typically assumed that time degrades as it is transferred over long distances making accurate comparisons between different references a challenge. This new generation of long-distance links represent a step further into tightly synchronizing different devices over fiber at long distances, thereby making them resilient to the vulnerabilities of GNSS timing.

This level of accuracy has become essential for certain applications and sectors. Matt Nassr, Data Engineering Lead at Optiver, remarked “Partnering with Seven Solutions has allowed us to better optimize for the highly distributed nature of the financial markets. Being able to establish a nanosecond-level precision link across the Chicago-NJ path further improves our ability to provide liquidity across the major US exchanges.”

Francisco Girela, Americas Tech Responsible at Orolia, stated, “This is not a first step on the deployment of long distance time synchronization networks, this is just one step further. We have been working for years on improving our devices, easing the monitoring and management, enhancing the resiliency, integrating failover features, adding interoperability with IEEE 1588 (PTP), NTP or PPS and allowing the integration of high accuracy timing in third party devices thanks to the HATI core. During this journey, we have learned how to integrate the technology to work at its best and we have proved its effectiveness in telecom, datacenters, finance, defense, or power grids among others”.

When asked about next steps, Girela added “We want to demystify this technology. People think that its deployment is complex or overkill, but that is not true. We see White Rabbit as the core technology to build alternative PNT systems, setting a sub-nanosecond timing foundation that will push the development of a more diverse and more accurate technology ecosystem. We have worked to evolve White Rabbit devices to be a market commodity when deploying time synchronization in local areas but providing a level of accuracy that will fulfill the requirements for the current applications and for the next decade. For long distances, complex deployments, or projects we are always happy to assist our customers and partners to get the best from our devices. We are sub-nanosecond natives; we care for every single picosecond.”

Related Resources Related Resources