Product Releases

Running for Coverage: A Review of Femtocells

Thu, 10/09/2008 - 6:03am
By Peter Thornycroft, Aruba Networks

The scale of femtocell deployments (should they become successful) will introduce several orders of magnitude more transmitters than today’s network of cell towers, transforming an already difficult RF management task into something significantly more challenging

Femtocells can be described as short-range radios used to extend the cellular network into homes and office buildings. While there is enormous enthusiasm over their prospects, deployment is still at a very early stage, with many operators evaluating femtocells in the lab and few are actually in early trials.

While femtocells differ from Wi-Fi access points in some ways, they seek to solve many of the same problems and are often considered to be competitive solutions. Here, we’ll examine key issues of the technology.

Femtocell Technology
A femtocell is a ‘mini base station’ that includes a radio unit like that in a cell tower for a mobile operator’s network. It is designed to provide short-range coverage, ideally extending through a house or apartment, but not radiating much outside that area. It is small, about the size and shape of a Wi-Fi access point, and is designed for self-installation by the consumer. An Ethernet connection backhauls voice and data calls across the consumer’s Internet connection to the cellular operator’s core network, allowing the mobile operator to extend its mobile network into the home by using the consumer’s Internet connection.

Most often, femtocells are deployed in order to compensate for poor cellular coverage. Mobile operators themselves estimate that perhaps 10 percent of subscribers do not get good cellular coverage at home. While femtocells may be used in this way to fill gaps in the macro cellular network, some vendors suggest that the technology will become so inexpensive that next-generation mobile networks will be made up of thousands or millions of femtocells, with only a few macro cells to cover outdoor areas.

The radio in the femtocell is controlled by the mobile operator’s core network, and it operates with standard mobile phones with no special modifications. When a roaming subscriber arrives home, their phone will sense poor macrocell coverage, and automatically hand over to the femtocell, passing across calls in progress in similar fashion to macrocell handovers.

Current Status of Femtocell Technology
A myriad of issues affect femtocells at this stage of their evolution; all operators require zero-touch installation. Consumers will buy a femtocell, take it home, power up and connect it to the Internet. At this point, the femtocell must seek out and connect to the operator’s core network, identify itself and its location, report its neighboring macrocells, and decide the RF channel it should operate on. In turn, the network must identify the subscribers this new femtocell is to serve. As with all service provider equipment, these provisioning steps must scale to thousands of new femtocells installed per day with minimal calls to customer service and even fewer technician dispatches, zero-touch installation with zero tolerance for errors.

The femtocell’s location must be determined and reported to the operator’s network to ensure emergency calls are directed correctly. This is surprisingly difficult to achieve. GPS positioning is universal and accurate, but notoriously unavailable when under a roof, which is the natural habitat of the femtocell. Triangulation on the macro network is already proven, but many femtocells will be placed in areas with no macro network coverage. Another potential solution identifies the location by the IP address of the Internet connection, but this is not completely reliable, except perhaps for combined fixed-mobile operators who control the whole Internet chain. The method most likely to avoid lawsuits when calls go astray is perhaps to ask the femtocell customer for a home address at point-of-sale, but femtocells will obviously be moved over time.

Location is also important to determine what RF channel and power settings the femtocell should use in order to fit into the macro network and avoid interference. The first consideration must be to prevent the femtocell from transmitting on a channel licensed by a competing operator, which could lead to considerable strife. To ensure this, the operator must know the femtocell’s location to ensure it selects only from channels licensed to that operator in that area. Once the legal channel range is established, a particular channel must be chosen for best fit with the macrocell network, as a femtocell on the same channel as an overlapping macrocell will lead to interference. After emergency call location, this is the second reason the femtocell’s location must be accurately known.

Operators are divided about open or closed usage models for femtocells. An open installation accepts calls from any subscriber to the network, which has the potential to overload a subscriber’s Internet connection with public freeloaders, and to leak revenue from the operator, depending on the tariff. Alternatively, if a femtocell is to be configured for use of only one subscriber or one family, administrative mechanisms must maintain a subscriber list for each femtocell, and other usage must be restricted, presumably with the exception of emergency calls for which there is a universal carriage requirement.

Inter-cell interference and frequency planning present particularly tricky problems when femtocells are used in the network. Mobile operators already perform wonders to squeeze maximum coverage and minimum interference out of a very limited slice of spectrum. Femtocells must integrate with the network’s channel plan for their location, causing difficulties because of the inevitable situations where channel shortage dictates that they must share the channel of an adjacent macrocell. Also, the scale of femtocell deployments (should they become successful) will introduce several orders of magnitude more transmitters than today’s network of cell towers, transforming an already difficult RF management task into something significantly more challenging. Additionally, closed femtocells that serve only named subscribers will appear as coverage holes to other subscribers, further exacerbating quality of service issues.

Quality of service (QoS) in its broadest sense of customer satisfaction is a continual concern of mobile operators. When applied to femtocells, there are two new aspects. Since standard RF interfaces are used, over-the-air performance is well understood except that the low transmit power of femtocells may mean that phones are operating at the edge of coverage much of the time. Indeed, many of the issues reported in early trials arise from coverage that is insufficient to light up the whole house, particularly difficult in multi-story dwellings with concrete floors.

The second source of QoS concern is over the broadband Internet connection. While it may prove easy for combined fixed and mobile operators to coordinate Internet QoS with femtocells, cellular-only operators are reliant on services beyond their control, with concomitant concerns about service calls and customer satisfaction. It seems clear that a successful femtocell design will need to monitor and adapt to the instantaneous bandwidth, packet loss and delay characteristics of the Internet connection.

The network interface, and indeed the overall system architecture for femtocells, is very early in the standardization process. There is general agreement that it will not be possible to tunnel femtocell traffic directly back to today’s core cellular networks; there is no interface there capable of terminating and controlling vast numbers of femtocells. The choice of architecture is complicated, as mobile networks are not yet based on IMS, but will migrate to it over the next several years, and because the interested startups have considerable intellectual property and patents in this area, and are not ready to forfeit the associated commercial opportunity. The recently founded Femtocell Forum has a goal to encourage vendors and operators to converge on a smaller number of architectural options.

The RF interface should be one of the strengths of femtocells, as the mobile phone has a well-defined over-the-air protocol. The roadmap for GSM operators includes today’s predominantly 2.5G networks, 3G which already includes a number of incremental steps, and the emerging Long Term Evolution (LTE) standard. The ideal femtocell would be software-configurable for all these, also for CDMA standards, and potentially support multiple standards simultaneously. Sadly, the cost of such processing power is currently prohibitive, even if it was feasible with today’s technology. Thus, equipment vendors must choose particular combinations, which will each appeal to a subset of global operators, shrinking their potential market and economies of scale.

Femtocells must meet significant levels of voice and data performance to be accepted by the consumer. In the voice dimension, performance equates to call quality, but mobile operators’ data services, already an over 10 percent share of revenue, must be available on femtocells just as readily as from the macro network. The newest 3G phones are advertised with data capability in excess of 2 Mbps, and as access to these services in the home is considered a primary application, operators are keen to see this level of performance is not compromised when service is provided from a femtocell. Preliminary trials indicate that this is a serious effect, as femtocell-provided data services in early trials compare unfavorably with Wi-Fi from the same Internet connection.

Wi-Fi access points work with any Internet connection and Wi-Fi client, but operator-independent femtocells do not exist. Initially each femtocell will bear a particular operator’s logo, and will work only for that operator. Until there is a de-facto standard for the network interface, a universal design will be impossible. Operators expect femtocells to provide an advantage by delivering capabilities competitors cannot easily match, even offering pricing plans to encourage whole families to use the same operator and share the same femtocell. These goals are predicated on an operator-specific femtocell, and would be negated if the consumer could purchase a femtocell from one operator, but quickly switch its service to a competitor.

Benefits to the Mobile Operator and the Consumer
The main benefit of femtocells is to improve in-home coverage, but there are many other interesting ramifications of the technology. Since 3G mobile networks use higher frequencies and different modulation than 2G, they generally have shorter range for voice calls and even worse building penetration. Femtocells can be linked to 3G either to provide subscribers another incentive to upgrade, or to compensate subscribers who find their previously marginal home coverage becomes unusable following a 3G upgrade.

The most significant reason for consumers to upgrade their mobile to 3G is for higher data rates, and the multimedia services these enable. Data services suffer the same range and building penetration effects as voice. The only way to reach large numbers of subscribers with high-data rate services is to reduce the cell size, which implies adding more base stations. Since macrocell sites are very expensive to identify and build, user-sited femtocells hold the solution.

As 3G takes off, the macro networks will serve many users simultaneously demanding data services, swamping the backhaul connection and forcing an expensive upgrade of network infrastructure. Similarly, in many areas the user population has overwhelmed the macro network with traffic. Open femtocells can be judiciously added

to relieve the pressure without requiring new cell tower infrastructure. Initial cost estimates for femtocells are also attractive: a network of femtocells could be orders of magnitude cheaper to build than a traditional macro network. It certainly appears that a 2G operator wishing to introduce 3G could do so by scattering femtocells as an overlay to the 2G network at a much lower cost than a traditional macrocell upgrade. This begs the question: do we need macrocells at all?

Femtocells Compared to Wi-Fi Access Points
When compared to Wi-Fi, it is clear that femtocells fall short in many areas, but offer a number of potential advantages. For example, the cost and pricing model for Wi-Fi is understood and successful. Chipsets are sub-$5, access points sell for $30, and consumers are happy to buy them at that price. It is unclear what value consumers will assign to femtocells.

Furthermore, consumers are comfortable with Wi-Fi. Regarding femtocells, it is uncertain whether consumers will care or understand that femtocells and Wi-Fi access points use different technology, and whether this will affect consumer acceptance of the newcomer technology. Consumers may also be concerned about being locked into one operator with their femtocell.

Another key point is that Wi-Fi access point interfaces are clearly defined, allowing multi-vendor interoperability and driving innovation as designers seek novel and profitable new uses for Wi-Fi. It is difficult to see this competitive, innovative model translating to the femtocell world where each device will be specific to, and sold via a particular operator; the operators become gatekeepers in dictating new features, an economically inferior business model.

Also, Voice over Wi-Fi (VoWi-Fi) is already a mission-critical service for many organizations, and video and IPTV over Wi-Fi are in the fine-tuning stages. Meanwhile the best mobile network video service, DVB-H, does not scale beyond a tiny screen.

While operators’ exclusive use of licensed spectrum is usually seen as a benefit, it becomes a disadvantage in the home. As previously noted, femtocells must learn exactly where they are located, and transmit only on the channels licensed to the parent operator, one of the most significant unsolved problems. Indeed, while femtocells are restricted to a single location, Wi-Fi access points can be mobile, so a personal access point can be set up almost anywhere in the world. Further, Wi-Fi QoS is proven even under conditions of RF interference that exist in the real world, whereas early trials show the performance of femtocells overlapping with macrocells on the same channel is problematic.

Mobility is another area where Wi-Fi lags cellular, but the gap is closing. Fixed-Mobile Convergence (FMC) research over a number of years is culminating in large-scale trials of solutions such as Unlicensed Mobile Access (UMA), a GSM standard for supporting dual-mode Wi-Fi/cellular phones and handing over calls. UMA is in commercial service world-wide with operators such as BT, Orange and T-Mobile. The most significant barrier to this technology, the availability of dual-mode handsets, has already been overcome.

Wi-Fi covers a much broader range of products than just phone devices. Wi-Fi is embedded in an extraordinary range of consumer devices, from PCs where it is now the default communications interface to digital cameras, game consoles and printers, among others. This is due to the ubiquity of Wi-Fi home access points, interoperability guaranteed by strong standards, and the low cost of silicon due to huge volume production. Wi-Fi is used by the nascent home media center market, connecting devices throughout the home for photo, music, video and TV services. Many DSL and cable TV vendors are integrating Wi-Fi with their termination and set top boxes.

The ongoing cost models are very different. With Wi-Fi, there is nothing to pay after purchase. But any device accessing an operator’s licensed femtocell must be registered in some way, and will attract a monthly subscription fee. Choosing a particular operator’s femtocell will immediately reduce the range of clients to those compatible with that operator’s network.

The Future of Femtocells and the Home Communications Network
The femtocell concept, and the considerable research and development that has resulted in early trials of femtocell technology, offers some exciting glimpses of one version of the mobile operator’s future network.

Wi-Fi is indisputably the technology for the in-home wireless network. A number of FMC architectures use the home Wi-Fi access point to deliver the same benefits as femtocells, and Wi-Fi is already entrenched, has high production volumes, low prices and good consumer acceptance, and is likely to grow in time into this wider role of communicating photos, music, TV and voice around the home. Wi-Fi already has a significant network effect, and mobile operators must choose whether to be part of that network, or to maintain their separate technology island.

But the mobile operator still controls two vital levers with the subscriber. First, the principal phone number is key to the subscriber’s identity. Non-mobile forms of FMC cannot replace this number, although when the mobile network supports reliable, high-bandwidth data services, it will become possible for voice service providers to tunnel through it with VoIP and offer an alternative. The second key advantage is that only mobile operators can provide wide-area coverage.

Analysts and others are suggesting that various combinations of Wi-Fi and femtocell technology may give consumers the best mix of services, quality and price. A simple device might combine a Wi-Fi access point and femtocell , although more interesting integration is possible. When an operator also provides wired Internet service, the combined unit could add a DSL termination. There is scope for these integrated service providers to deliver a complete Internet, phone and TV service to the consumer.

It will take some years, but the possibilities exposed by inexpensive femtocells in conjunction or competition with Wi-Fi access points promise considerable disruption to existing business models, to the ultimate benefit of consumers.

While femtocells differ from Wi-Fi access points in some ways, they seek to solve many of the same problems and are often considered to be competitive solutions.The network interface, and indeed the overall system architecture for femtocells, is very early in the standardization process.

Peter Thornycroft is a technologist with Aruba Networks,


Share this Story

You may login with either your assigned username or your e-mail address.
The password field is case sensitive.