What do enterprise IT department managers think about deploying femtocells? How does this compare with using a Wi-iFi based scheme such as UMA? We've been discussing this with a Telecom Services Planner at a medium sized US company, who is also an enthusiast for UMA. Network security, handset choice, provisioning and the ease of switching between network providers are some of the topics we've covered.
Open Access versus Restricted
Question 1: Surely UMA/Wi-Fi or femtocells would be configured as open access by default and benefit both visitors and office workers? Home workers might choose to limit access, but it’s not really a major cost problem. You can’t easily do this with the existing corporate IT network.
Setting up a separate, logical Wi-Fi access for guests is, probably, the best way to go to support UMA devices (as well as guest users on your Wireless LAN).
It can be achieved without having to set up a separate, physical system: Enterprise-grade Access Points (e.g., Cisco Aironet) allow you to set up a separate SSID for public use that is outside of the trusted network, and unable to access enterprise resources. (Each SSID is managed separately, as if it was a separate, physical Ethernet interface.
In contrast, employees must authenticate themselves to gain access to the trusted SSID to gain access to corporate resources; Guest access only gets direct access to the Internet) Effectively, the guest traffic is routed around the enterprise servers and onto the ISP.This is a common feature that an IT Technician can configure for an enterprise Wi-Fi network. In fact, this Guest access feature is even standard on Apple's consumer Wi-Fi product (Airport).
A separate LAN for Femtocells
Question 2: Would an IT department also want to implement a separate IP network for femtocells, or would they trust them enough to just plonk them onto the existing internal network and provide a single secure point of transit to the operator via VPN?
Femtocells can backhaul traffic to the MNO over the Enterprise ISP connection just like UMA traffic (as described in the previous answer): the IP traffic would be routed around the Enterprise and onto the Internet.
The IT folks would set up a separate VLAN using the existing gear (to save money), and would not need a separate physical IP network. In both cases, UMA and Femto, the MNO obtains free backhaul.
Switching between Network Operators
Question 3: How easy is it for an Enterprise with UMA or Femtocells already installed to switch between Network Operators?
The flexibility to be able to switch mobile operators is a big deal for both the operator and enterprise involved. I know that operators invest heavily in winning and retaining their large accounts, but will draw the line at some point (they have to make a profit). I see femtocells and UMA competing more for the smaller but more numerous SOHO/SME markets than larger enterprises (where the cost of specially installed picocells is justified).
The flexibility to be able to switch mobile operators is a big deal for both the operator and enterprise involved: Enterprises want the flexibility to choose service providers, and Operators want to lock the customer in (to reduce churn). In-building solutions are attractive to MNOs as a mechanism to lock in the customer for a long time; once the MNO’s network is physically entangled with the Enterprise location, it will be harder to switch out. Reduction of churn is a big deal for MNOs, economically, due to the high cost of acquiring a provisioning a new subscriber. It used to be the big topic for MNOs (when I was one, 5 to 10 years ago).
Operators will invest heavily to win and retain large accounts (that are highly profitable), but they will draw the line at some point (they have to make a profit) when it comes to costly, in-building deployments. That’s one of the promises of femtocells: far lower cost (especially for in-building solutions). Since femtocells can only serve relatively low capacity at present, I see femtocells and UMA competing more for the small to medium Enterprise (SME) market rather than larger enterprises. Only after higher-capacity femtocells come to the market can they effectively serve large business locations. UMA is an acceptable solution for large businesses today, since Wi-Fi does deliver high capacity and can readily handle the additional mobile traffic.
I agree that the pico- or femtocell will, long-term, probably serve the large market more, as the capacity goes up and cost comes down. Still, wouldn’t the current femto work well in a small enterprise? If the current state of the art is 4 or 8 simultaneous calls on a femto, then that would seem to be sufficient for many, small to medium businesses. Eventually, when the femto capacity gets up there, then they present an elegant way to deliver great capacity and coverage in a large business (and to lock them in). I’m really not sure who wins, as I can see that in the same time frame, Wi-Fi will be far more common and the enterprise can use their existing WLAN to obtain UMA service, so I see strong competition in all markets.
Customers will get a better deal with UMA than Femto, and that may sway customers to UMA. I perceive that UMA benefits the MNO more than Femto, and therefore the service plans with UMA may be more generous (free calls/data) than with Femto. From the MNO perspective, I look at it this way: UMA saves me way more than Femto, so as an Operator, I can afford to be more generous as there is more cost reduction and therefore more profit margin in the pricing. With UMA, I get free spectrum, free radio, and free backhaul. With Femto, I get free backhaul. Therefore, MNO service plans will tend to be better with UMA than Femto. Better service plans will tend to retain customers longer.
I’m not so convinced about the argument that UMA saves an MNO way more than a Femto. The MNO has already paid for the spectrum, and using femtos effectively gives ten times or more capacity than outdoor macrocells – it allows cellular to be deployed in the office at a fraction of the cost of traditional DAS or picocell solutions. UMA would likely require additional Wi-Fi hotspots to be deployed, so we’re really just comparing the cost of adding a few more boxes and still need the IT department to setup the parallel VLAN for this traffic. On the other hand, all users would need to be issued with new UMA capable handsets – surely a far more expensive task (when you include admin, training, transfer address books etc).
Question 4: Surely there is additional work in configuring handsets for UMA access (which requires Wi-Fi security key exchange/config, loading preferred Wi-Fi access points etc) than for straightforward femto access (which requires none)?
Yes, there is a small bit of additional work to use UMA (whereas there is no work that is required of the End User to use a Femtocell): the UMA handset must be provisioned with (a) the name of the Enterprise WLAN (manually or automatically, such as broadcasted from a service console, such as the BlackBerry Enterprise Server). This no big deal—very similar to connecting to a Wi-Fi network from your laptop, which everyone is familiar with. Optionally, if you wish for the UMA device to use a protected network, then you’ll also need to deliver (b) credentials to the handset to allow access the Enterprise WLAN (again, manually, from a web site, or using an enterprise-wide server). We believe that the availability of an automated tool ensures that this can be done in a simple, cost-efficient manner.
Manual provisioning of this data is not scalable or reliable, as you pointed out.
Question 5: You’ve suggested that switching between UMA operators would be easier than switching between femto operators. I’d agree that there would be no need to change the Wi-Fi access points or phones for UMA, but the femtocells would need replacement to change femto operators. In the US, switching could also easily involve a change in technology (so new phones all round between GSM/CDMA/iDEN). Perhaps a larger issue is the lack of operators offering a coverage solution, although all the main US operators will have something this year (3 out of 4 do today).
Yes, when a MNO installs an in-building pico- or femtocell system, it will be hard to switch out (good for MNO, bad for Enterprise). The MNO will probably require a multi-year contract to ensure that their investment is recovered. After the installation of an in-building Femto solution, the Enterprise will have great solution, but one that the are contractually and physically locked into for some time. As you pointed out, if the Enterprise wanted to switch operators, they’d have a harder time ripping out the old MNO femto solution and having another installed. In contrast, the UMA solution requires no change to the WLAN—this suggests that UMA can be attractive for larger Enterprises. In-building Femto is restrictive; UMA allows choice.
Personal Handset Choice
With femtocells, surely there is no need to replace any of the handsets; with UMA you would need to replace most of them – causing issues with staff who like to have their personal choice of pink/Android etc. model.
Question 6: Wouldn’t limiting the choice of handset to UMA models cause an issue with some workers?
Handset choice may not be as big an issue for Enterprises as you might imagine, since there is little real choice to day and UMA-compatible devices are becoming more common. Handset choice is not a common benefit for the corporate customer today, so I doubt that they’d miss something that they never had. The IT and Telecom department decide which Operator to use and which handset portfolio they prefer and what handsets they choose to support at the Enterprise End Users can “choose” only from a limited set of phones (much as they can “choose” from a very limited set of PCs that are supported). If the set of available UMA devices is sufficiently large to gain approval from the IT/Telecom department (and it is today), then UMA devices are what the IT department will offer, and workers will be happy with what they receive ;)
Requiring a special handset is a death sentence for a service, so it is a concern that there are a limited selection of UMA devices today … but this problem is going away due to the widespread adoption of Wi-Fi in mobile phones. Basically, Wi-Fi is becoming quite common in high-function, business phones, and is expected to be in most smartphones in a couple of years. And UMA is just a client that is securely installed for the Operator in the Wi-Fi capable phone before it is distributed. So Wi-Fi adoption will ensure that UMA devices are widely available, and handset choice will be less restricted, over time.
Only in rare instances (small company, strong-willed exec) can the End User buck the system and demand a specific handset. In such cases, the IT department is creating an exception to their rules about which types of devices to support. Their standardization on a few devices is an effective cost control. When an exception is made, then the one-off handset may not have the use of the in-building system. But for execs, special solutions can be provided, such as a femto in their office (just like the cell site that Verizon installed for John McCain).
I’d disagree from my personal experience. It’s pretty common for users to buy their own handsets and switch the SIM card across. I’ve also seen many use two phones – say a blackberry and something smaller for voice only. This is even more prevalent in smaller companies. If there were an adequate range, why is Marc Fossier (EVP Orange) quoted saying that the operator has just 20 handsets that support UMA and would like to have more.
"It's sad that UMA clients aren't widely deployed across devices. We would plead for device vendors to deploy as widely as possible."
Combined Femto Unit versus Standalone
Question 6: Are enterprise femtocells likely to be standalone units, or be integrated with other features? Market studies are showing that domestic customers want fully integrated boxes with broadband modem/firewall/router/Wi-Fi/Femtocell.
I think enterprise customers will prefer a standalone femtocell, probably with greater capacity – hence Huawei’s 16 channel femto (or is it a pico), and the chip vendors stating they can support 8 channel (picoChip) or 16 channel (Percello) on a single chipset.
Agreed: Enterprise femtocell solutions will require a lot more capacity than current reference designs, and they’ll want them separate from—not integrated with—their Wi-Fi Access Points (that they already have in place, thank you very much). So the Home market may be like the trend to Femtocells integrated with Wi-Fi gateways (an “all-in-one” solution), but Enterprise customers will want Femtocells completely independent of their WLAN.
Thanks to Robert Duncan for sharing his thoughts with us.