I am currently wrapping up a project in which I have worked over two years. I have been working in interference mitigation and cancellation for OFDMA-based femtocell networks. One of the widely used main assumptions for these kind of networks is that the femtos do not have GPS capabilities – after all, they are by definition located indoors where, most likely, they won’t have GPS coverage – so one assumes that femtocells are not synchronized from the rest of the network -macrocells -.

This lack of synchronization arises many challenges that are the main area of investigation of many current projects, including mine. Handover between layers – femto to macro or viceversa – and interference are two of the main one.

When I started working on this topic, femtocells were just starting to be produced and sold, but nowadays there is multiple options available, such as Verizon’s Network Extender (made by Samsung), AT&T’s 3G MicroCell and Sprint’s AirAve. I was curious to know how do these products deal with the synchronization problem and I was very surprised – and pleased – to find out that most of the current products are equipped with a GPS receiver.

A colleague posted recently on LinkedIn some results and insights he got after using AT&T’s 3G MicroCell. In this case, the GPS s used, among others, to make sure that the customer is radiating in licensed spectrum where the operator – AT&T in this case – owns the spectrum. So, if the femto is taken to another address different to the one it is configured for, it will not radiate – so, it doesn’t work -. Another function is the GPS is to provide location to 911 calls. Finally, it is used for timing – synchronization -. The problem is that one has to place the femtocell at most a few feet away from a window, otherwise it will not work.

I asked him about the roaming and he reported the following. He tried 25 times and only had one dropped call when doing handover between the macrocell and the femtocell.

After hearing this I thought that vendors had solved the synchronization problems by simply adding GPS to the femtocells. However, another colleague mentioned that the system works as follows. After locking the GPS for location, one has 15 minutes to later establishing a wire-line connection – i.e. connecting it to your cable/DSL router somewhere else perhaps not close to any window -. Subsequent location verifications to keep the femto radiating can be done by the network listen feature detecting other cells/femtos that you would expect to see in the designated location.

On the other hand, for CDMA/EVDO femtocells, these networks are time-synchronized, so they are likely to require GPS signal the entire time. However, there is probably other ways to achieve time synchronization. Note that, for WCDMA networks – like AT&T’s MicroCell – there are existing techniques to achieve synchronization via the broadband connection without GPS, such as IEEE 15xx.

In my case, we have been working with OFDMA-based networks, where a very good synchronization is necessary. Assuming that the system lacks of this synchronization and femtos can only listen to neighboring femtos, we chose a distributed approach applying game-theoretic concepts.

As final comment, I was curious to see if Sprint’s Airave is a 4G femtocell and I found the following:

Built-in GPS antenna for network timing updates and E911 services.

The current GPS location (provided by the GPS antenna) is used to help the base station quickly locate GPS satellites for faster GPS acquisition. If the GPS signal is ever interrupted or too weak because of the location, install the external GPS antenna and place it in an open area for better reception (See “Installing the External GPS Antenna” on page 18.)

So it does require GPS as well.