Synchronising Cellular Networks

The mobile operator market is a commercially aggressive and changing environment, customer retention is critical to the success, sustainment and growth of a business.

Operators have worked hard to improve efficiency of customer support, service packages and network performance. However aside from the obvious commercial considerations such as price and content there are still technical problems that customers will commonly quote for changing operator.

  • Call drop out
  • Poor or no network coverage in certain areas
  • Quality of Service

Whilst  network coverage is and has improved greatly (We are nearing total coverage),  the biggest problem is call drop out and this is often the most visible frustrating quality of service issue a customer will encounter.
 
How can poor Synchronisation cause a call drop out?

The basis of any robust and reliable GSM / UMTS network is good quality synchronisation references. This is particularly applicable to base stations, an area often overlooked in terms of synchronisation.

The accuracy of the synchronisation at the base station is critical to the call hand over (when a subscriber moves from one cell/base station to another) . This level of accuracy is required to ensure hand over between base stations without drop out is +/- 50 parts per billion ppb.
 When a hand over occurs between two base stations the potential for frequency difference can be as high as 100ppb. This in turn can be equated to a Doppler shift of 100kph in vehicle speed. If the mobile subscriber device i.e. the handset, data-card etc. cannot react quickly enough to this Doppler shift the call will be dropped.

The Problems

As highlighted above without effective synchronisation a mobile network will not function correctly due to frequency disparities between base stations.

The historical method of achieving synchronisation quality within the GSM environment (in particular at the base stations) was to  co-locate a low quality oscillator at the base station site  which would be retimed using a  timing reference derived  from either a T1 or E1 backhaul line. This was known as recovered clocking.

However this approach creates another set of issues which Mobile Network Operators must address: Using leased lines (which are normally sourced from fixed line operators) the mobile network operator will have no direct control over the leased line reference, therefore the quality of the synchronisation reference cannot be guaranteed (remember we are dealing with an accuracy tolerance of -/+50 ppb!).

Also if the leased line synchronisation reference is compromised the fault resolution time can be slow, difficult and costly due to reliance on the provider of the leased line to effect the resolution. While this is happening the base station will at best only have the co-located low quality oscillator in holdover mode. This will only provide frequency accuracy of around 10 x -5 meaning regular call drop outs at the user end.

This problem for mobile operators is now also being compounded by the demand of modern subscribers for the increased and more flexible bandwidth that is required to support today’s rich media content.  In order to cope with this network operators are moving away from traditional TDM infrastructure for backhaul as well as access. As most data and content is now IP based the cost effective solution of choice is based around Ethernet which delivers high bandwidth cost effectively.GSM / UMTS markets and networks are not an exception to this.

Hence relying on a leased fixed line for the provision of synchronisation is no longer feasible.

The Solutions

As the operator’s migrate from TDM based E1 backhaul to packet based IP backhaul historical GSM / UMTS synchronisation methods such as recovered clocking cannot be employed,  However this can be counteracted by a range of solutions that are now available.

In order for the mobile operator to control the handover quality determined through base station synchronisation within their network they need to directly control the synchronisation quality. Therefore an independent synchronisation method must be utilised at the actual base station.
 
In field trials is has been demonstrated that the use of operator controlled synchronisation at the base station reduce call drop out by up to 35%.
 
There are a number of ways to achieve this:
 

  • Direct base station synchronisation via local Primary Reference Source (PRS).
  • Use a GPS based retiming device
  • The use of IEEE1588 V2 synchronisation delivery over packet

Direct base station synchronisation via local Primary Reference Source (PRS)

Where either existing TDM based E1 back haul is used or the operator has already migrated to packet based IP back haul the operator can resolve the requirement for synchronisation delivery by deployment of a Primary Reference Source (PRS) at the actual base station, in this example the PRS receives its reference from a GPS signal (The GPS clock reference adhere’s to G.811 and as such meet’s the required bit loss tolerance). This provides a direct reference to the base station as shown in figure 1.
 
Figure 1 – Direct Base Station Synchronisation via local Primary Reference Source

Additionally the PRS unit fed by the GPS Reference  can have a high grade built in oscillators to provide holdover protection in the event of loss of GPS reception, thus giving the mobile operator direct control of the synchronisation source and time to fix, without compromising Q.O.S to the user.
 
Retiming E1 Traffic lines

Where the mobile operator still has access to an incoming E1/T! signal A simple and very effective GPS based retiming solution can be used.

Figure 2 – GPS Based Retiming of E1 transmission line

The re-timer device  which is also fed by a GPS reference retimes the reference signal received from the leased lines removing any transient movement in the synchronisation reference before it can cause a call drop out at the subscriber device. As with the above example The GPS based devices are traceable to ITU-T Recommendation G.811 quality and adhere to that of the 50ppb quality required at the base station.

Additionally the GPS Retimer unit could also have a high grade built in oscillators to provide holdover protection in the event of loss of GPS reception, again  giving the mobile operator direct control of the synchronisation source and time to fix, without compromising Q.O.S to the user.

PTP IEEE 1588V2 Base Station Synchronisation

Differential Clock Recovery (DCR) is a technique, recently  ratified by the IEEE, in a standard known as  PTP/IEEE1588v2. PTP/IEEE1588v2 can exceed G.823 Stratum 2 accuracy for phase and sync <50pbb freq drift and 1-3µs phase/absolute time Again ensuring tolerance requirements are exceeded).

This offers a real  alternative to the deployment of GPS(which can be difficult/costly in cases where the mobile operator does not own the physical site).  
 
Figure 3 – Base Station Synchronisation via IEEE1588V2

Synchronisation is derived from a traditional primary reference source (PRS), such as GPS(although it must be noted that one only GPS reference would be needed per network rather than the multiple references required in the above examples), For added resilience the GPS reference source could also be  be backed up by other Primary Reference Sources such as a cesium reference clock.

The synchronisation is then distributed from a Grand Master device to slaves deployed at the Node-B/Base Station via the IP/Ethernet backhaul. The slave extracts timing and provides a synchronisation reference signal to the base station external synchronisation port.

Its is clear that it is vital that the synchronisation plan of existing and evolving mobile/cellular networks is revisited and assessed, with a changing and more demanding arena on the horizon synchronisation devices will be critical to customer satisfaction and control of churn.