Showing results for 
Search instead for 
Did you mean: 

Is Femto GPS a self-inflicted pain?


It appears that US Mobile SPs are very eager to put GPS modules into their Femtocells (aka FAP). In Europe and elsewhere, however, this is not the case.  Why?

Everybody knows that getting reliable GPS lock inside buildings is very hard.  The best GPS reception is at the window.  But for the best Femto coverage you want it in the middle of a typical McMansion.   When SPs embed Femto into other devices such as STB, they won't be by the window.  External antennas are pain. Trees interfere with line of sight.  And GPS modules are not free.  So, why do US providers want to inflict this pain on themselves and their customers?

Let's review why folks *think* they need GPS in a FAP:

  1. FAP location verification. This ensures:

         (a) FAP does not radiate on the wrong frequency and does not violate FCC rules; this is relevant if an SP has different spectrum channels in different coverage areas;

         (b) FAP can provide data to e911 services for UE (phone)

  2. FAP oscillator disciplining.  Accurate and stable frequency requires expensive crystal hardware - too expensive for FAP.  One of the techniques used to sync clock of the cheaper crystals is to continuously obtain precise time from GPS.  But there are viable alternatives, such as sync'ing from neighbor macro broadcasts or via NTP from a quality source. These two mechanisms can provide sufficient accuracy for WCDMA and are much cheaper.  Certainly, European SPs targeting deployments without GPS think so.

It is possible that 1.a can be more of a problem in the US with SPs there having grown by a lot by acquisitions and covering many different spectrum ranges varying not only by state but also by country and even sub-county level.  However,  there are likely easier solutions than introducing the GPS for this problem.

For example, the FAP could periodically scan for transmitters on the target channel(s).  If it decodes a foreign neighbor (wrong MCC/MNC), we know it is not where it should be or someone else is violating the rules.  In the few cases where it finds no neighbors, the FAP may violate FCC rules and transmit on the wrong frequency, but this would require a customer who lied about the placement of their FAP. Therefore, such incidents should be minimal. The damage is limited because the FAP ensures there are no other transmitters around on the same channel (and PSC) and the FAP transmits at relatively small power.  But would FCC agree with such interpretation?

Now, e911 gets even more interesting.  FCC wireless e911 rules mandated a two phase roadmap.  In Phase I, the SPs were required to provide the location of the base station to the emergency services.  In Phase II, they were required to provide the location of the UE itself within 50-300 meter accuracy.

For Phase I, the location of FAP would need to be provided.  For macro, this is based on its registered address.  The idea of movable base station was probably not considered back in 1996 when the regulation was first issued.  A more recent precedent in FCC is the e911 rules for VoIP issued in 2005.  These rules require SPs to transmit customer's "Registered Location".  The customer must maintain his latest location in the SP database.  This recognized the fact that it may not always be technically feasible for SPs to automatically determine the location of their end users without users' active cooperation.  The same should hold for FAPs IMHO.

For Phase II, we need UE location.  The solution has to work on macro as well as on Femto.  A number of approaches for UE location determination have been investigated including network-based (such as triangulation of signals), UE based (such as with GPS) and hybrid.  None of them are 100% reliable.  The current winner appears to be a hybrid approach with GPS and assistance data (AGPS) from the network.  Knowledge of the FAP location may help in providing assistance data to UE.  But again, it is not clear why registered location won't suffice.

To sum up... the pain was inflicted by FCC & SPs' conservative interpretation of the current rules:

  1.  The e911 requirements could be addressed with user registering their FAP address with the SP.  However, it is not clear if FCC's current rules allow this.

  2.  The non-interference requirements could be met in the same fashion, plus enforced by having the FAP periodically scan for conflicts and shut itself down.

Should SPs take action at FCC to clarify the rules?  I think so.   The current FCC rules are ambiguous as they relate to Femto.  This leaves SPs to incur extra costs, while providing a less ideal experience to users, some of which have to place Femto devices close to a window, use external antennas, etc. And some customers will never get a GPS lock anyway because of external barriers such as trees and fences.

My take is that FCC rules will eventually be relaxed to be "best-effort only" GPS in Femto.  But in the meantime, it is up to each SP to interpret the rules as they see fit.

What do you think?


Anton, very interesting comments. Completely agree that the center of the "McMansion"  is where most people would want to place their femto, and this is precisely where indoor GPS coverage can be challenging. This can compromise the user experience. In my opinion, this however does not take away from the fact that there is a need for a service provider to determine the current location of the FAP.  As you pointed out there is "can my femto use this spectrum in this area?" issue. I think it needs to be resolved. Otherwise can lead to lot of conflicts (lawsuits ?) between operators. The 911 issue is also valid. I believe in this case the operators will provide a better service if they can provide the civil address associated with the femto through which the call is being made. Kind of like the landline 911 experience. These are two drivers for GPS. However for these two problems there are also alternatives to GPS. The third driver for GPS is slightly more difficult to work around  and that is synchronization. The need for depends on the network.

Finally on the question about US vs. European operators, I believe the emergency call system does not work in the same manner over there. Also the operators there may have more homogeneous spectrum allocations. So all they need to look out for  are FAP in foreign country scenarios.

In summary, agree with a lot what you said, except that I feel that GPS is not a self inflicted pain by US operators. GPS merely represents a solution to some requirements for which the femto community has to find alternate solutions, if GPS is not proving to be a practical solution for all cases.


Ajay, thanks for comments.  My fear is that US SPs may not feel like they have a choice given the current regulatory environment.  Or at least they may not want to take the risk of going down the wrong road given the ambiguity.  If the FCC rules were clarified and relaxed, the SPs could make the decision based purely on technical reasons.  Then, some may decide that using GPS is worth the troubles for them and others that it is not.  But the FCC rules aren't going to clarify themselves unless SPs take action at FCC.  At the moment, it looks like GPS in FAP is probably here to stay (in the US). At least for a while.


I tried moving my Vodafone Access Gateway femtocell to a new location several miles away from the original installation, and found that the device started up and worked fine.  It seems that there is no location lock (at least for moving within the UK).  Vodafone simply states in the VAG Ts & Cs that “the emergency services may use the details you register for the Gateway in order to identify your location.  It is therefore important that you keep your contact details up to date and immediately notify us of any change.”


(This comment is made on behalf of Todd Young, VP Business Development, Rosum Corporation).

I really enjoyed your article.  A few comments:

1. The timing requirements for UMTS are indeed precise frequency stability (100ppb or 250 ppb depending on the release), but the requirements for CDMA and WiMAX are much more challenging - absolute timing to 1us and (for WiMAX) 20ppb frequency stability!  NTP and macro cell don't cut it for these requirements.

2. The reality of E911 is that as people go wireless-only and cut their land-lines (and replace with femtos in some cases), people will continue to expect that when they dial 911 that things will happen.  Did you know that the FCC requires no indoor testing for E911 today?

3. I think you are right that the solution for E911 includes user registration of their address or carrier address registration via the provisioning process.  However, this address must be validated automatically on start-up to avoid all the problems that Vonage has had with people moving their VoIP boxes and first responders being sent to the wrong place (even the wrong country).  The other reason why registration and auto-validation is the right solution is the "which door to kick down" problem.  Let's say that you live in an apartment and you dial 911 on your handset in a bad coverage area, but you have a femto and it serves your call.  If the femto is placed beside a shared wall, then if you locate that femto with GPS (or Rosum's TV+GPS), with an accuracy of say, 10m (which would be excellent indoors in an apartment building), then you still probably don't know which apartment the caller is in.  That 10m circle could span several apartments.  However, if you have an address and validate that the location is consistent with the address, then you have actionable information for first responders.  If the femto shows up on the other side of town, or in another city then you know the registered address is no good.

4  Finally, I also like you point about putting femtos near the window. If femtocells really take off, then in order to avoid femto-to-femto interference, you really want the femtos to be in the RF center of the residence - where coverage for the residence can be delivered with the minimum power, and where THE BUILDING ATTENUATION CAN BE USED TO REDUCE INTERFERENCE WITH OTHER FEMTOS AND MACRO CELLS.  If femtos are driven to the windows (by GPS and macro cell timing), then interference will be problematic.  Building attenuation should be our friend.

Rosum was founded by original architects of the GPS to extend timing and location into indoor and urban areas where GPS fails.  We use advanced navigation and timing technologies to fuse together multiple sensors including broadcast TV-based timing (50dB better stronger than GPS indoors), Assisted GPS, Macrocell sniffing, network timing.  We integrate these sensors to improve availability and accuracy of frequency, timing, and location.  Or technology, Rosum ALLOY, is being adopted by major femtocell manufacturers.