Timing and synchronisation are critical components of an efficient mobile network architecture. Get it wrong and you can dramatically affect the efficiency and capacity of the network, not to mention dropped calls or data sessions during handovers. There are two significant differences for public access small cells that require different solutions than residential 3G femtocells. Here we look at the options and best practices.
What makes for a good timing and synchronisation solution?
The main parameters of a good timing and synchronisation solution include:
- How fast it takes to "lock on" to the correct frequency within tolerance. Nobody wants to have to wait hours before they can use their latest femtocell at home.
- Holdover time: How long the base station can maintain the correct frequency if it loses its connection from the master clock. For lower cost products this might be minutes; for large cellsites it might be days. This all comes at a price.
- Cost: Particularly of the small cell but also for master clocks at the central sites.
- Adaptability: Flexibility to operate across a variety of different transmission technologies, wireless/wireline, circuit/packet, etc.
3G Residential Femtocells were easier
There are two common approaches to 3G residential femtocells:
a) GPS, which provides very accurate frequency and phase alignment but requires a little extra hardware. As a bonus, it can also determine the location of the device. CDMA femtocells use this because of the need for synchronised phase alignment. North American operators mandate it because of the location information used for 911 emergency calls. Few networks outside North America require this option because of the extra cost. A downside is the need for an external antenna positioned to receive the GPS signal which typically won't penetrate indoors.
b) NTP (Network Timing Protocol), which operates across domestic quality broadband Internet, such as DSL. The low cost of this solution is attractive. The downside is the potentially long time to acquire and lock-on to frequency within the tolerance, which can range between a few minutes to more than an hour.
c) Network sniffing, where nearby macrocells (and even other small cells) can be used as an alternative clock source. This is particularly helpful when small cells are used for additional capacity, ensuring that they are tightly synchronised with neighbouring cells helps mitigate interference issues.
By contrast, the common approaches to 3G macrocells are:
a) SONET/SDH: The most popular scheme which embeds synchronisation within the transmission network, but this is being replaced by high speed Ethernet.
b) GPS, as for residential femtocells above. Essential for CDMA networks, it can be a quick and easy solution but risks the network going awry if the GPS signal was to go offline for any reason.
c) Synchronous Ethernet (SyncE): This is similar to SONET/SDH for Carrier grade Ethernet services – you won't find it in your office or domestic Ethernet, but is becoming more widespread across commercial networks. It distributes a master clock throughout the network and provides frequency (not phase) synchronisation. There is a limit to the number of "hops" it can traverse and the network must provide an end-to-end SyncE connection across every "hop." It's less often found in corporate networks or for residential customers.
d) Precision Time Protocol (PTP) and specified in IEEE 1588v2: A packet timing scheme (similar to NTP) which provides both frequency and phase timing. It's common to combine SyncE and PTP, which is slowly starting to displace SONET/SDH in many networks.
The demands for timing are becoming greater because:
- Network operators typically mandate tighter timing requirements for public access small cells, even more than the standards demand and better than often specified for residential femtocells. The 250ppb limit for 3G femtocells is often tightened to the 50ppb used in macrocells.
- LTE-FDD requires 16ppb frequency tolerance.
- LTE-TDD requires phase alignment as well as frequency accuracy, with a tolerance of 1.5uS.
- Specific LTE-Advanced features, such as Co-Ordinated Multi-Point (CoMP), require phase alignment of 0.5uS. This may become popular when working as part of a HetNet (a closely knit mix of large and small cells).
- Public access metro-cells and higher capacity enterprise picocells are now being used with other transmission schemes, such as Carrier Ethernet.
Some choices to be made
The industry will need to agree which of these various options to adopt. The choices will include which techniques to use, what tolerance to aim for and holdover time.
Symmetricom, who supply the master timing clocks for the majority of mobile networks today, recommend that at least two out of these four options (GPS, SyncE, PTP, NTP) are installed for any public access small cell. This would provide fallback in case any one method becomes unavailable. For LTE, this might be a mix of GPS and PTP. A soft-client would be required to choose between the available sources.
Rakon, the leading supplier of oscillators in femtocells today, are more cautious about the need for LTE phase accuracy and don't recommend using NTP for an LTE small cell. They expect a combination of GPS and SyncE/PTP to be popular. From their perspective, higher spec oscillators than those used for 3G residential femtocells are likely to be in demand for LTE.
What this means for small cell designers
We should see 3G public access small cells incorporate SyncE and PTP for timing instead of NTP, because this will enable faster "lock-on" timing and provide the higher tolerances to work within a HetNet.
Residential LTE femtocells are likely to require a GPS receiver as part of the standard build. Public access small cells connected using SyncE would not. This is true for both FDD-LTE and TDD-LTE modes.
It will be up to the network operators to decide how good they want the holdover time to be – the longer required, then the better and higher cost of the onboard oscillator it will need. I'd expect this to be a few hours for the more critical public access metro cells, i.e. more than a residential femtocell but less than the 1-3 days for a full size macrocell.
As always, a tradeoff between cost and performance needs to be made.