Timing issues have been a concern of mobile network architects ever since Ethernet - an inherently asynchronous protocol - became important for mobile backhaul. The reasons are obvious enough: without exquisitely maintained timing relationships, user experience degrades or calls terminate because they are not handed over properly from one cell to the next.

These days, it is equally important to maintain Internet access sessions and synchronous services across cells using asynchronous backhaul.

With the advent of small cells, the challenges have grown larger. Where in the past the primary issue was maintaining synchronization between neighboring macrocells, mobile operators now must contend with handoffs between small cells operating “underneath” macrocells in roughly the same area.

Also, as small cell networks are laid “underneath” the existing macrocell network, frequency coordination has to be augmented with signal phase alignment. The former is fairly easy; the latter not so.

The reason is that distance matters for signal phase relationships in a way not crucial for the frequency domain. And distance now matters in a new way. “Where” base station controllers sit in the network is a new issue. More controllers are remote and centralized.

And that has direct implications for where clocks are located. Traditionally, timing synchronization has used macrocell GPS or master-slave clock networks. That was one sort of challenge for macrocell networks.

But the very proliferation of sites (nestled underneath existing macrocells) requires active spectrum coordination and interference management. And asymmetric transmission delays make it harder to maintain and recover timing.

To achieve successful call signal handoff over the IP network that connects base stations, and transport real-time data, each base station must support timing accuracy of +/- 50 parts per billion, as a rule of thumb.

That wasn’t as big an issue when mobile backhaul primarily used T1 or E1 synchronous TDM networks for backhaul.

Today’s mobile backhaul networks, using Carrier Ethernet, require new methods to preserve timing. Up to this point, timing synchronization over Ethernet has used either the  ITU-T Synchronous Ethernet (Sync-E) G.8261 protocol or IEEE1588v2 Precision Time Protocol.

Sync-E uses a reference clock that synchronizes all the slave ports. So does PTP.

But timing issues arise because the locations where small cells operate do not always have reliable access to the GPS signals traditionally used to provide the highly-accurate frequency and time-of-day synchronization information required by Time-Domain (TD)-LTE and LTE-A networks. Deploying GPS at each small cell location is also not economical.

The IEEE1588v2 Precision Timing Protocol and PTP for packet time-stamping can provide small cells with nanosecond-accurate time-of-day synchronization across the backhaul network.

But the 1588 protocol passes its timing information between network nodes by encapsulating it within standard Ethernet or IP data packets.

The same packet delay variations that produce jitter and delay asymmetries in packet networks also introduce timing inaccuracies in the recovered clock. PTP assumes symmetric latency across communication paths. And latency in the real world of backhaul is not symmetric, and has a static as well as a dynamic component.

In other words, timing relationships between a master clock and any single base station recovering the timing can vary significantly, especially when microwave backhaul is used for parts of a backhaul network.

All Long Term Evolution base stations and small cells therefore require accurate phase and time-of-day synchronization. GPS, a traditional solution, is fine so long as it is available at every base station, cost and line of sight are not issues, and additional complexity is not an issue.

Weather conditions, reflection at tall buildings and the increasing challenge of GPS jamming make a GPS-only solution however not practical for a large scale rollout of public small cells. Maintaining phase synchronization in the range of +-1µsec also constitutes a big challenge during holdover periods where tracing the GPS reference is not possible.

If a network relies on macrocell time of day and phase control, latency and ability to receive macrocell signals indoors are issues. Additionally, few operators are able to upgrade any network ubiquitously, so a mix of controllers, base stations and backhaul must be assumed.

Timing has been an issue for backhaul networks ever since Ethernet became part of the mobile backhaul network. Now small cell architectures are pushing timing issues to the forefront again.