Practical Time Transport

By Ray Savich   January 15, 2014

We’re not talking about time travel here, at least not in the way used by Marty McFly and Doc Brown in the Back to the Future films. Instead, we’re talking about accurate synchronization in mobile networks. Synchronization isn’t likely to be the basis for a blockbuster movie franchise but it’s really important in the latest mobile networks using the LTE-TDD or LTE Advanced standards. These standards typically require all of the base stations in a network to be synchronized to a common reference with a worst-case error of 1.5 µs. That’s 0.0000015 of a second, or the time it takes light to travel 1,476 ft in a vacuum.

Accurate synchronization is an essential part of satellite navigation systems like GPS, and GPS receivers offer one method to distribute a synchronizing signal with the required accuracy. However, this solution is not without some drawbacks in a mobile network. GPS is vulnerable to deliberate or accidental interference and it’s not always easy to locate a GPS receiver with a clear view of the sky – think about installations indoors, in subway tunnels, or at street level in an area of high-rise buildings.

It’s often more convenient to synchronize base stations to a centralized reference using timing messages carried in the backhaul network. The technology of choice for this task is the Precision Time Protocol (PTP) as defined in IEEE1588-2008. PTP allows us to synchronize a slave clock to a master clock, in roughly the same way as the existing Network Time Protocol (NTP), but with a much higher level of accuracy. Importantly, PTP provides support for hardware platforms that measure the time of transmission and reception of messages in hardware. This technique is often referred to as hardware time stamping. To achieve the 1.5 µs end-to-end accuracy needed for mobile networks, every router, switch or wireless link must support hardware time stamping and must compensate for the delay that individual timing messages incur at each point in the transport network. PTP provides us two ways of doing this: Boundary Clocks (BC) and Transparent Clocks (TC). There will normally be several BCs and or TCs between a master and a slave clock, and network planners are typically looking for less than 50 ns of constant time error in each router, switch or link in the network to be sure of achieving the overall accuracy of 1.5 µs.

The Cambium Networks development team has been working on an IEEE1588 Transparent Clock feature for sub-6 GHz wireless microwave links, and it’s fair to say that we’re quite excited by the performance we’ve achieved. The test result below shows the time error introduced by a single TDD wireless link with the Transparent Clock feature enabled. The result is smoothed by a 10 s moving average filter to approximate the function of a client clock. The fixed time error of 2.8 ns is 0.0000000028 of a second, or the time it takes light to travel just 33 inches.

This was on a good day; we can’t guarantee achieving 2.8 ns under all circumstances, but we’re confident that we’ll be able to achieve the required 50 ns under a wide range of configurations.

Now, if we could just figure out how to get 1.21 gigawatts into the flux capacitor, we could probably crack time travel as well.