Synchronization of GSM Base Stations

Timing isolation through IP

When base stations carried just voice traffic, a single T1/E1 connection typically provided enough bandwidth for the backhaul connection The rollout of third-generation (3G) data services,however, has increased the bandwidth needs for the backhaul connection significantly and moving to T3/E3 connections is simply too expensive. Transport networks are rapidly evolving to IP-rich topologies This offers mobile operators the increased backhaul capacity they require for deployment of high-bandwidth data services and the cost advantage of IP transport. However, the move to Ethernet backhaul will eliminate the option for base station clock recovery from the backhaul facility. Operators will need to move to an independent source of synchronization at the base station to meet the UMTS 50 ppb requirement as shown in figure 1.

Figure 1. The transition to high-capacity IP backhaul drives the need for stand-alone
embedded clocks in UMTS Node B base stations.

In addition to traditional span line clock recovery where the base station recovers an accurate clock from the T1/E1 backhaul feed, UMTS Node B infrastructure suppliers are introducing high-quality embedded clock options to be ready for IP backhaul. Many of these options mirror code-division multiple access (CDMA) 2000 designs where GPS clocks are embedded into the base stations to provide a time-of-day reference needed for call hand-offs. CDMA networks have always relied on embedded GPS-based clocks with precision rubidium or quartz oscillators, making them inherently prepared for the evolution to IP backhaul from a sync quality point of view.

Using rubidium-based oscillators is the most robust solution for independent synchronization of UMTS base stations, as rubidium oscillators are proven to meet the 50 ppb requirement over the full service life of the equipment. Quartz oscillators, on the other hand, are subject to higher native aging rates and warm-up/restabilization characteristics that make it difficult to assure compliance to the 50 ppb requirement for more than a few years.

This exposes network operators to QoS degradation and potentially high maintenance costs associated with manually calibrating quartz oscillators to bring them back on frequency after only a few years in the field. The danger to the operator is that this type of failure is undetectable until QoS issues reach a critical threshold.

Figure 2. Satellite backhaul providers rely on GPS-based retimers to insert quality
synchronization onto the T1/E1 feed to the base station.

GPS-based timing
A problem occurs for legacy GSM base stations that have relied on recovering synchronization from traditional T1/E1 backhaul lines provided by the incumbent local exchange carrier/public telephone andtelegraph (ILEC/PTTs) to maintain their 50 ppb frequency accuracy requirement. Changing the backhaul to circuit emulation services (CES), such as the satellite transport as shown in Figure 2, requires that a local source of synchronization be placed at the base station to deliver an accurate clock reference since the satellite network cannot support the 50 ppb requirement on its own. There are two ways to achieve this: 1) install an external GPS clock to externally time the base station equipment, or 2) use a GPS-based retimer as shown in Figure 2.

Figure 3. Circuit emulation and IP encapsulation introduce instability to network
timing signals.

A GPS-based retimer buffers incoming traffic and clocks it back out with PRS level accuracy (Figure 3). A retimer can be transparently introduced to an existing base station as the retimer is placed on the backhaul feed directly before the base station. The timing signal the base station receives is reclocked to be precise and stable, enabling accurate synchronization between base stations. Note that the retimer itself requires access to a PRS-based clock. Typically, this is achieved using an integrated GPS receiver.

Such a retimer should implement a cut-through mechanism to preserve communications when the GPS signal is not available, eliminating the retimer as a point of failure. When cut-through is enabled, the base station will revert to the original backhaul timing scheme and its original dropped call rate without the retimer. In this way, retimers can only improve

quality, never reduce it. Retimers can result in a dramatic reduction in dropped calls. A five-cell field trial conducted in September 2004 with a major GSM operator resulted in a 25.5% reduction in dropped calls when the backhaul synchronization signal was retimed. The five base stations involved in this test were experiencing relatively high dropped

call rates, and synchronization impairments on the backhaul lines were suspected as the root cause Note that these measurements include call hand-offs with base stations not using retimer synchronization feeds. While there was still significant improvement in these cases, the most substantial improvement was realized when both base stations involved in the call hand-off were retimed.

THE AUTHOR By Barry Dropping holds a BS in Electrical Engineering Technology from the DeVry Institute.