High spectral efficiency and smooth performance of mobile networks is aided by accurate and reliable timing synchronisation. This is even more important in an LTE HetNet environment which co-ordinates transmissions across macro, micro and small cell layers.
Leading oscillator vendor Rakon has published a white paper discussing this topic and we've extracted the salient points below.
The key timing requirements
GSM, 3G and LTE-FDD networks need transmissions to be frequency aligned to within 50ppb (parts per billion) mainly to cope with the Doppler shift effect of fast moving users (e.g. inside automobiles or trains). Few of us drive that fast indoors, so indoor residential small cells have a relaxed tolerance of 250ppb.
LTE-TDD mode systems must also time align transmissions to an accuracy of +/-1.5us, known as phase synchronisation. This capability is also essential for some of the LTE-Advanced features such as eICIC, CoMP and eMBMS. Where LTE-A is being used for location finding/position reporting, requirements for <0.5us are being requested.
The industry is actively debating just how tight that phase tolerance really needs to be. While 0.5us might be worthwhile for the most demanding and difficult locations (e.g. large football stadium), typical indoor office environments using LTE may be adequately served using frequency sync alone.
There are few phase synchronised mobile networks in operation today apart from those using TDD. As we move towards more sophisticated HetNets, operators need to have a clear strategy and roadmap to meet their anticipated needs. Retrofits could be a very expensive!
A choice of timing sources
A small cell timing solution comprises:
- one or more external timing sources
- some intelligent method of acquiring a stable clock and determining that it is fit for purpose
- an onboard oscillator capable of sustaining accuracy and linearity during inevitable short term variations, glitches and outages throughout the equipment's lifetime.
There are three main sources of synchronisation signals
- Satellite GNSS, such as GPS: These offer excellent phase tolerance of around 0.03us but usually require visible sight of the sky and can be susceptible to jamming. Highly sensitive GNSS receiver modules such as from u-Blox (which incorporate a Rakon oscillator) offer improved indoor performance.
- RF Sniffing: It's very likely that small cells in congested areas can receive RF signals from nearby macrocells and slave to their heartbeat. It does require briefly scanning/receiving on the transmit frequency band and sometimes even on other frequencies not actively used by the small cell itself. Almost all small cells can do this while it's less common in DAS or Remote Radio Head architectures.
- Backhaul: A combination of Synchronous Ethernet (SyncE) and IEEE 1588 v2 is a popular solution. This delivers both frequency and phase synchronisation, with the SyncE helping to increase the stability and performance that IEEE 1588 provides standalone.
Generally speaking, operators are looking to design their networks with at least two of the above options available at every cellsite. They want to ensure that the network is resilient and won't fail if any single timing source goes offline. A fallback option is even more important for phase synchronisation because the holdover time achievable for any given oscillator is very much shorter than that for frequency sync. Lower cost parts may achieve phase sync measured in minutes rather than weeks or months.
Phase noise radically affects highest achievable throughput
Quite separate from achieving the right frequency and phase alignment is the need for the clock signal to be extremely stable. Phase noise affects how much data can be modulated onto the RF signal. Just as signal to noise ratio is important to determine if the data can be received, low phase noise allows higher modulation rates to be used. LTE defines modulation schemes up to 64 QAM and LTE-A up to 256 QAM. This means there are many more constellations to be decoded and they must be as error free as possible.
The bottom line is that the headline data rates of LTE-Advanced can only be achieved with low phase noise clocks. In network architectures where the radio signal is sent to remote radio heads using CPRI or other techniques, this can be a problem although there are workarounds.
Age before beauty
Oscillator performance degrades over months and years due to aging, resulting from chemical changes in the crystal structure and its internal environment. When this becomes too far out of specification, the equipment must be replaced or repaired. Aged components can develop increased phase noise and non-linearity, significantly affecting overall network performance. Replacing the small cell doesn't just affect the capital equipment budget - maintenance visits to replace or repair small cells can be costly too.
This provides a strong business case to select higher quality components at the design stage which result in longer equipment lifetimes. Rakon has implemented a patented polynomial compensation technique which extends the working lifetime of their products beyond generic oscillators.
Outages, glitches and changeovers are inevitable from time to time. Holdover is the ability of the system to maintain the synchronization within acceptable limits when the source of synchronization is lost. Once the holdover time has expired, the frequency and timing sync may need to be re-acquired – this can prolong a short term outage into a lengthy period of downtime.
Phase sync is much more difficult to maintain that frequency sync alone. It's not unusual for phase timing holdover to be a factor of 1,000 times shorter (see table below).
External timing sources may be lost for many reasons, from localized GNSS jamming, backhaul connection interruptions, external equipment reboots.
Sync may be quickly substituted from another source which is why most operators look for redundancy by using at least two sync sources. They need to balance the cost of providing multiple external sync methods against that of demanding higher performance and more costly oscillators.
A smorgasbord to suit every taste
Rakon's approach has been to offer a portfolio of oscillators for designers to choose from, balancing budget, performance and physical format. It's sometimes surprising to see different designers making quite different component choices for products destined for the same end user and to meet the same product specification. It isn't quite as straightforward to balance design cost against performance, so perhaps operators should take a closer look at the underlying component choices made.
The choice of timing solution and oscillator component is critical for both the performance and longevity of every small cell product.
Low phase noise and linearity of the timing solution is critical to achieve the highest datarates that LTE-Advanced offers. This requires a high quality oscillator at every radio node.
Several LTE-Advanced features and all LTE-TDD systems require phase synchronisation, which has 1000x shorter holdover times and leads to the need for multiple timing sources.
Small cell designers have to make some hard choices to balance cost against performance, especially for the latest LTE small cells. The range of oscillator products commercially available today, such as those shown from Rakon above, offer options to meet both budget and technical performance. It's now down to the industry itself to make the right choices.
The white paper can be downloaded directly from the Rakon website (No registration required)
Disclaimer: Rakon is a sponsor of ThinkSmallCell