English
NOTE: This blog is a summary of a paper presented with Calnex at ITSF 2020.
Timing is a challenging element of modern telecommunications networks. To allow the smooth transfer of calls and data across the network, the individual base stations (BS) of a cellular network must be highly synchronised. That means there must be a common, dependable reference time source that all cells can be synchronised to.
In most cellular networks, that source is the time signal broadcast by GPS and other global satellite navigation systems (GNSS). It’s acquired by a GNSS receiver in the network’s Primary Reference Time Clock (PRTC), which uses it to generate a 1PPS time signal. That acts as the reference signal for time, phase and frequency synchronisation across all of the other clocks in the network.
To prevent the subsidiary clocks from drifting out of sync, as they will tend to do within a matter of hours, the PRTC must continuously re-obtain GNSS time and redistribute it over the network.
GNSS time may not always be reliable
While architectures may differ, this is the basic approach to time & sync in cellular networks, and it usually works well. However, it does rely on a critical assumption: that the time generated from GNSS is accurate and reliable. In reality, that assumption may not always be true.
There are many reasons why the GNSS time signal may not always be accurate, from the presence of radio frequency (RF) interference to occasional errors in the satellite system itself. There’s an excellent summary of the issues on pages 10-13 of this ITU Technical Report, but for this blog I’ll focus on just one: multipath fragmentation of the GNSS signal.
Multipath is a threat to GNSS signal reception
Multipath is the effect produced when the GNSS signal encounters obstacles in its path to the receiver’s antenna. Tall buildings, trees, vehicles, and even the ground can cause the signal to reflect or diffract, such that it takes different paths to arrive at the antenna.
These different arrival times can cause deviations in the receiver’s measurement of the incoming GNSS signals, which, if not accounted for, will cause the receiver to output an inaccurate 1PPS time signal, adding jitter and consequently degrading network timing performance. Looking to learn more about multipath,? Read our ebook Understanding Multipath and Obscuration: How to simulate GNSS signal propagation in urban environments.
With the advent of 5G networks requiring many more base stations to be ever more tightly synchronised, it’s essential for GNSS chipset developers, as well as developers and users of PRTCs, to understand and mitigate the impact of multipath on timing and synchronisation accuracy.
To help timing receiver developers and users assess the risks from multipath, Spirent and our partner Calnex have developed a complete simulation testbed to evaluate the impact of multipath on dedicated timing receivers.
Multipath effects are location-dependent and hard to measure
Measuring the impact of multipath on a PRTC is challenging because so much depends on the exact location in which the PRTC antenna is installed. An antenna in an open area with a clear view of the sky will encounter few, if any, issues. But if the antenna is mounted in a dense urban area surrounded by tall buildings, multipath is highly likely to be a factor.
The exact nature of the multipath reflections will depend on the obstructions in the antenna’s specific surroundings, and are therefore difficult to generalise. It’s also difficult to conduct live tests at a physical location, because the test requires a reliable source of GNSS time to act as a control, which is impossible to obtain if multipath is present.
The solution is to simulate the effects of multipath in the lab, using a geo-realistic model of the location, with a radio frequency constellation simulator (RFCS) generating realistic GNSS signals, environment modelling software to model a 3D environment, and ray tracing software to model realistic multipath effects. In this setup, a receiver can be tested with and without multipath, and the tests can be reliably repeated with different variables.
A complete testbed for evaluating multipath
To help timing receiver developers and users assess the risks from multipath, Spirent and our partner Calnex have developed a complete simulation testbed to evaluate the impact of multipath on dedicated timing receivers.
The setup includes our Spirent GSS7000 multi-constellation, multi-frequency RF signal simulator, our Sim3D environment modelling and ray tracing software for multipath simulation, and Calnex’s Paragon-X timing monitor to monitor the accuracy of the output 1PPS and PTP signals.
Testbed setup for 1PPS and PTP outputs from the timing receiver device under test (DUT)
Using this setup, we tested three commercial receivers marketed for use in time & sync applications. Our aim was to understand the extent to which multipath can degrade the accuracy of the 1PPS output, and the consequent risk of not meeting the ITU G.8272 standard for timing accuracy in PRTC-A reference clocks in cellular networks (+/- 100ns either side of UTC).
The units we tested included single-frequency and multiple-frequency multi-GNSS receivers and a flexible IEEE 1588 PRTC. We first tested each receiver in a simulated clear-sky environment without multipath, and then in three urban locations with realistically-simulated multipath signals: in downtown San Francisco, Manhattan and Shanghai. Each test ran over 24 hours.
Simulation of line of sight (white), reflected (red) and diffracted (blue) signals, San Francisco location
Significant vulnerabilities discovered
The tests revealed multipath vulnerabilities in all three receivers. When the tests were run under clear-sky conditions, all three DUTs performed within the PRTC-A limits. But when tested in a dense urban environment, the DUTs showed significant degradation in the 1 PPS and the PTP measurements, well outside the PRTC-A limits defined in the ITU G.8272 standard.
Timing accuracy in clear-sky conditions: timing outputs remain well within the 100ns performance threshold
Timing accuracy in multipath conditions: significant deviations exceed ITU-defined tolerances
Conclusions
Our tests clearly show that timing receivers that conform to timing accuracy standards in clear-sky conditions can experience significant degradation when installed in multipath-rich locations.
We would therefore recommend that manufacturers, integrators and purchasers of GNSS-dependent timing receivers test each receiver’s performance in the presence of multipath, and take action to mitigate any impact.
(For an example of a mitigation approach, read our case study of how NTT Furuno developed a multipath mitigation algorithm for its timing modules, tested with Spirent Sim3D.)
Spirent Sim3D software represents a major step forward in the simulation of multipath effects in real-world and geo-realistic 3D locations. The testbed described above is available for customer use in our UK lab.
