Subscribe to our Newsletter

Keep informed of small cell developments with our free monthly newsletter and articles. Sign up and receive a free eBook. Your email address will not be shared with 3rd parties.
View past editions

RSS Feed

rss_icon_small Blog updates via RSS
or emailed to your inbox

You are here: Home Technology System Clock Frequency Accuracy within Femtocells for timing and location
Clock Frequency Accuracy within Femtocells for timing and location
Written by David Chambers   
Thursday, 05 February 2009 19:42
Clock Frequency Accuracy One of the potentially most difficult aspects of a femtocell design is providing the accurate timing clock for all the digital circuitry. It’s got to be good enough to meet the tight tolerances, but inexpensive to fit the budget. We discuss the requirements and innovative ways in which various vendors address this issue.

Why?

Mobile networks handover calls between basestations as you move around. Handsets derive the accurate frequency that they transmit and receive on from the basestations. If the transmission frequencies are not very closely matched between adjacent cellsites, then “pops and clicks” can occur when the call hands over (switches) between basestations. In the worst case, the handset would not be able to immediately lock onto and acquire the new signal and your call would drop.

There can also be issues of interference between networks, typically reducing the call quality and network capacity.

Timing accuracy also allows the femtocell to “sniff out” adjacent cellsites, ensuring good behaviour to reduce interference.

What’s required

We should first establish the timing accuracy requirements for femtocells. This has traditionally been very demanding – 3G UMTS systems required a tolerance of as little as 50 parts per billion in long term frequency accuracy.

Referring to the official specification from 3GPP, TS25.104 Release 8 (December 2008) quotes:

“The modulated carrier frequency of the BS shall be accurate to within the accuracy range given in Table 6.0 observed over a period of one timeslot.”

Table 6.0: Frequency error minimum requirement

BS class
Accuracy
Wide Area BS  ±0.05 ppm
Medium Range BS  ±0.1 ppm
Local Area BS  ±0.1 ppm
Home BS  ±0.25 ppm

For example, if transmitting at a nominal 1920 Mhz, then the permitted range is 1,919,904 Hz to 1,920,096 Hz. The wide area cellsite must stick within these limits for as long as it operates, which may be as much as 10 years after production.

Femtocells(stated here as Home Basestations) are permitted five times this deviation, so from 1,919,520 Hz to 1,920,480 Hz. This allows the products to use cheaper components and keep the costs down.

You can compare this with the frequency tolerance required of handsets, which is 0.1 parts per million – four times wider than femtocells.

Other radio technologies have more demanding requirements:

  • CDMA, used by operators such as Sprint and Verizon in the US, is s synchronous system. This is quite unlike UMTS, where each cellsite must transmit at the right frequency but not necessarily in phase with others. The frequency clock and phase must be kept in step with the rest of the network.
  • WiMAX, the new mobile broadband system offered by Clearwire/Xohm and many other operators around the world, requires a synchronised clock with a tolerance of some 1 microsecond.

 

How is this frequency accuracy achieved?

Various techniques have been used to check and maintain this long term frequency accuracy. Very accurate (and expensive) crystal oscillators, calibrated in the factory when the cellsite is manufactured, are used to provide high initial accuracy. These degrade over time as the crystal ages.

Crystal frequencies vary with temperature, so temperature controlled crystals are used (TCXO), maintained at the same temperature to avoid frequency drift.

Regular site visits have been common in the past, where field engineers carry highly accurate (and very expensive) frequency standards to compare and adjust longer term frequency drift. You could compare this approach with tuning a piano (mechanical rather than electronic).

The cheapest and more common approach is to derive a long term accurate clock and use this to correct the internal crystal frequency as it slowly drifts over time. Various techniques are used to do this:

  • Deriving the frequency from the PDH clock timing of the 2 Mbit/s transmission links
  • Deriving from a central accurate clock over IP using techniques such as IEEE 1588
  • Using a GPS receiver – the GPS system is based on a very accurate caesium clock reference
  • Deriving a clock signal from the terrestrial TV broadcast signal (analog or digital)

The CDMA femtocells launched to date have used GPS, which meets the dual need to provide the highly accurate frequency clock as well as location. This allows the femtocell to work out which licenced spectrum frequency to use for the area it’s in, and to switch off when outside the country. A third benefit is accurate location information for emergency calls.

WiMAX femtocells also use GPS for similar reasons.

3G UMTS femtocells are typically not designed with GPS in mind. This is partly due to the cost of adding a GPS chip, and partly because the same level of timing and location accuracy is not required.

Rosum innovated using an interesting technique, using terrestrial TV signals combined with GPS. As they point out, these are available almost everywhere and are some 50db stronger than GPS. Since they are transmitted at lower frequencies (hundreds of MHz), they penetrate in-building and are easily detected. Unlike GPS satellites, TV transmitters don’t move, so there’s no problem when satellites disappear over the horizon – a problem known as “holdover” where the femtocell must maintain accuracy whilst the GPS satellites are out of range. And although these TV signals don’t provide accurate location, receiving them shouldn’t require an outdoor aerial (or the femtocell to be positioned next the window). [Update 2011: Rosum no longer appear to be in business, suggesting they did not gain support for their approach]

Do these requirements need to be as demanding for femtocells?

3G UMTS femtocells have an advantage over their CDMA and WiMAX cousins. Since femtocells are very low power devices, and used in controlled ways, the technical committees have recently determined that the tight frequency tolerances can be relaxed.

One of the changes to the 3GPP standards introduced in the Release 8 version (published at end 2008) is the introduction of a new class of device. Femtocells are part of these low power home network devices, and the frequency tolerance has been relaxed to only 250 parts per billion (yes, that’s still quite demanding). External macrocells are still required to be accurate to 50 parts per billion (for 3G UMTS anyway).

Impact of lower frequency tolerance requirements

Each femtocell vendor may have a slightly different approach to meeting these requirements. All will want to reduce their product costs and use the cheapest possible reference crystal. CDMA and WiMAX femtocells are likely to incorporate GPS or TV receivers to meet their specific needs.

Comments (0)add comment

Write comment
smaller | bigger

security image
Write the displayed characters


busy
 

Keep informed of femtocell thinking. Signup to our FREE monthly newsletter and articles and get a FREE ebook!

Buy a Femtocell

Where can I buy a Femtocell?

You must buy these
from your mobile network
operator in your country.

Click here for a list of current
operators offering products
today and links to
their product pages.

Buy a Femtocell

Buy our Book

2nd Edition now available
Still less than £30
Buy online now

femtocell-primer-2nd-edition-web-cover
More info

"an excellent introduction to femtocell technology and business issues." - Andy Tiller [ip.access]