This guest post from Oscilloquartz, explains why synchronisation and timing will be increasingly important, and the options to support it in the network.
Synchronisation is typically required to optimise the spectrum bought by network operators and deliver the highest-quality of service experience to their users. Devices interconnected across a network must be synchronised to some degree of measured precision and across one or more dimension, enabling the scarce, shared resource of spectrum to be consumed efficiently (and, in some cases, regulatory requirements to be met). Synchronisation allows network operators to wrest the maximum throughput out of the spectrum that they have.
There are different layers of synchronisation, and the first question that an operator must answer in cost-efficiently designing its network is whether frequency, phase and/or time synchronisation is required for the applications to be delivered. Ultra-precise synchronisation is an increasingly important capability for interconnection of mobile small cells and other applications, and there are different strategies for deploying the capability across a network.
Dimensions of synchronisation
The most common and basic type of synchronisation is frequency only. There is no notion of time; the challenge simply is to get a frequency clock ticking at the same rate across the appliances within a network. Frequency synchronisation is a basic requirement for all wireless technologies, from GSM all the way through each generation of mobile technologies.
Basic frequency synchronisation is needed to make sure, for example, when someone is driving a car and the person’s smartphone receives data on a given frequency, that the data, in fact, comes in from the base station at the expected frequency. The Doppler effect can make the observed frequency higher or lower depending on the velocity and direction at which the car is being driven, so the signal is synchronised in frequency to account for this and ensure coordination across the network.
There are specifications for how precise the frequency must be, and it is measured in a metric called fractional frequency offset (FFO). A network operator is typically required to synchronise base stations to a FFO under 16 parts per billion, allowing for a margin of error.
The next layer of synchronisation, then, is phase. One way to think of phase synchronisation is like a heartbeat ticking across the network. If the heartbeat is to be one pulse per second ticking at all base stations, a capability for phase synchronisation allows all of the base stations across the network to coordinate around the same understanding of the beginning of that second. This effectively enables all of the base stations to know when to carry out certain process in a coordinated way.
Finally, there is time synchronisation. For specific applications, not only must the base stations be aware of when the beginning of a second occurs, but they also must understand to which time of day that second corresponds. For example, the application might need the base station to comprehend whether 12 o’clock means midnight or noon. This type of synchronisation is needed in more advanced features. Time and phase synchronisation is measured using a metric called time error (TE) and the typical requirement in high-end applications such as LTE-TDD and small cells is to synchronise network elements and bound the time error under 1.1 microseconds.
Transactional financial trading, some internet of things (IoT) gateways and smart power grids are among the applications demanding ultra-precise time synchronisation, but one of the most prominent among them is the radio access network (RAN). For the sake of efficient utilisation of radio spectrum, as well as optimal user experience, different mobile cells must be tightly aligned in phase, in order for transmission and reception of data across base stations, small cells and other mobile devices to be coordinated.
Existing networks typically cannot deliver the precise phase synchronisation that these leading-edge applications require. Strategies around two different approaches have emerged in the marketplace for building a capability for synchronisation at sub-microsecond precision of individual instances and applications in a distributed network environment.
The first approach leverages Global Positioning System (GPS) and Global Navigation Satellite Systems (GNSS). The network operator synchronises devices to the clocks in GNSS satellites which are UTC traceable. With GNSS receivers deployed at every location, a time alignment with an accuracy of better than 100 nanoseconds of UTC can be delivered across network elements.
GNSS alone is not a sufficient solution when interconnecting indoor small cells
Of course, deploying all of those GNSS antennas across a large, distributed system environment can be quite expensive. Also, the GNSS “everywhere” approach can result in frequent service outage since the GNSS signals are very weak and therefore easy to jam or spoof. So, while it is easy to use and delivers a very accurate degree of synchronisation, it is also very sensitive and can be quite costly. And GNSS alone is not a sufficient solution when interconnecting indoor small cells, which are becoming more and more prominent for enabling much higher throughput indoors, where GNSS signal-to-noise ratio is very poor.
Well-designed network-based timing distribution typically entails less operational complexity and can deliver a timing precision in the range of several hundred nanoseconds
GNSS-based timing distribution is a widely used approach today, but most network operators also must look to a backup solution for the synchronisation of performance-critical applications. The alternative approach is to distribute timing information via the underlying data network. Well-designed network-based timing distribution typically entails less operational complexity and can deliver a timing precision in the range of several hundred nanoseconds.
IEEE 1588™-2008, “IEEE Standard for a Precision Clock synchronisation Protocol for Networked Measurement and Control Systems,” supports delivery of frequency, phase and time synchronisation across heterogeneous systems via switched, packet networks. In Precision Time Protocol (PTP), time-stamped packets are traded between a master and a slave clock.
You could have GNSS at aggregation sites, connecting to PTP grandmasters, forwarding packets to small cells to deliver time.
The tradeoff with IEEE 1588 PTP is that maximum precision relies on minimal network delay asymmetries and packet delay variation. In this sense, the quality of the clock is not determined by the protocol itself; rather, it depends on the architecture and dynamics of the network. The use of PTP-aware network elements such as boundary clocks and transparent clocks can help control network asymmetry and packet delay variation.
In truth, the optimal solution may involve some combination of the two approaches. You could have GNSS at aggregation sites, connecting to PTP grandmasters, forwarding packets to small cells to deliver time. Solutions integrating GNSS antennas, GNSS receivers and PTP engines have emerged in the marketplace to provide accurate and affordable phase synchronisation for the rapidly growing small-cell market. This is crucial because small cells figure to play a larger and larger role with IoT proliferation and 5G on the horizon.
Ultimately, operators require the flexibility to reliably and affordably enable synchronisation in any environment and with no restrictions in terms of technology, form factor, etc. A thorough understanding of the new synchronisation landscape is needed today to make wise design choices for tomorrow’s requirements.
About the author:
Responsible for the development of synchronisation delivery and assurance solutions, Nir Laufer is director of product line management at Oscilloquartz, an ADVA Optical Networking company. He also represents ADVA in ITU-T SG15 Q.13 and regularly contributes to the development of network timing standards.