Network Integration Approaches for Femtocells

By Steve Shaw, Marketing Director, Kineto Wireless

Introduction

The wireless industry has been searching for low-cost indoor coverage solutions since the beginning of mobile networks. For practical and cost reasons, indoor coverage is normally designed into the outdoor macro network by statistically budgeting for wall attenuation when signals propagate through external walls of buildings. While the intent is to achieve a high percentage of cases with satisfactory indoor coverage, it is cost prohibitive to design RF coverage for 100 percent of indoor scenarios.

To date, a small sub-sector of the wireless equipment industry has satisfied the indoor coverage market by offering cost-effective picocell solutions for high-traffic and high-worth locations. Unfortunately, the bulk of the indoor coverage opportunity (i.e. residential environments) has been beyond the addressable market for cost and operational reasons.

However, recent developments in 2G and 3G silicon have raised the possibility of offering low-cost femtocells to address the residential indoor licensed coverage opportunity. Technology companies such as picoChip and UbiquiSys are working on products expected to meet tough budget requirements for mass femtocell deployment. This development is encouraging for the indoor coverage market and addresses some, but not all, of the challenges for successful femtocell service deployments.

For mobile network operators (MNOs) to achieve mass adoption of a femtocell-based indoor coverage service, three key technology requirements must be met:

1. Low-cost femtocell products (under $200)

2. Reasonable approach for managing RF interference

3. Scalable, cost-effective approach for core network integration

Low-Cost Femtocell Products

The physical femtocell is the single largest-cost item in the operator business case for deploying a femtocell-based indoor coverage service. The economics of deployment depend on vendors achieving highly cost-optimized designs. MNOs evaluating this technology expect that femtocell units in volume must cost under $200 if they are to develop a successful business case. Fortunately, femtocell chipset and access point vendors are now publicly stating their ability to meet this aggressive cost target over the next several years.

Reasonable Approach for Managing RF Interference

As a simple matter of physics, femtocells operating in the same frequencies as macro cells risk interfering with the macro network. In a normal operator-controlled RF plan, the frequency allocation (and scrambling code allocations in the case of W-CDMA) are carefully planned to avoid interference between transmitters. In the case of femtocells, the idea of prescriptive RF planning for millions of devices is simply unimaginable.

Fortunately, vendors in the femtocell space are now claiming to have invented techniques to address the RF interference issues. Femtocell-macro cellular interference is a very active area of research, and a number of recent studies have analyzed the potential RF interference issues, with promising results. Lab experiments and real-world field measurements in representative environments scheduled throughout 2007 are designed to validate the efficacy of these new RF interference management techniques.

Scalable, Cost-Effective Approach for Core Network Integration

While the femtocell vendors have made significant progress in addressing cost and interference issues, less progress has been made in solving the core network integration challenge. This is primarily because conventional mobile network infrastructure is not well equipped to meet the unique challenges of femtocell-based services.

In a 3G mobile network, Radio Network Controllers (RNCs) communicate with Node Bs over private, high-capacity, dedicated links for using the Iu-b protocol to the mobile core. Mobile handsets access network infrastructure through the RNC-Node B link that controls and delivers services from the mobile network core.

Femtocells must also enable mobile handsets to access these same infrastructure service elements. However, unlike operator-owned and operated RNC - Node B links, femtocells must, for economic reasons, use the public Internet for connectivity and core network access.

This difference requires that operators planning to introduce femtocells to their mobile network choose the right network architecture. The choice can make or break the business case as well as the technical and operational viability of the femtocell concept.

Femtocell deployments rely on a handful of critical network integration requirements to be successful:

• Scalability: The solution must provide for a cost-effective, highly scalable approach for integrating hundreds of thousands of femtocell access points.
• Service Security: Network-based security to authorize specific femtocell access points (based on location, owner) as well as managing access to resources by specific IMSIs.
• Network Operations: Minimize the installation impact in the mobile core network of access points being distributed across the country, as well as the variability of a consumer product which may power on/off randomly.
• Standardized Interfaces: A well defined, industry recognized standard promotes a vibrant market for independent, third party femtocell vendors, not locked pairs of RNCs/Node Bs from traditional suppliers.

Today there are three approaches to integrating femtocells into core mobile networks: the IP-based Iu-b interface specified in 3GPP Rel.5, a new Session Initiation Protocol (SIP)-based approach, and Unlicensed Mobile Access (UMA).

The Iu-b Approach

While there is temptation to re-use the existing approach for RNC- Node b interface, the Iu-b protocol has a number of drawbacks as the basis for a femtocell integration architecture.

Timing and Synchronizatio n

• For RNC – Node b communication, the Iu-b protocol is run over a private IP or dedicated ATM link to ensure consistent, managed performance
• Supporting Iu-b over the public internet requires a more robust and tolerant implementation of Iu-b to overcome potential pack loss, delay and jitter.
  

Mass Scalability

• RNCs are designed to handle hundreds of high-usage Node Bs, not millions of low-usage femtocell access points.
• Treating each femtocell as a Node B interfaced to the RNC is cost prohibitive. Even if many femtocells can be aggregated to appear as a single Node B, RNCs have the wrong cost structure and scalability to serve as the access interface to the mobile network.

Service Security

• Running the Iu-b interface over the Internet requires a new security layer to protect the RNC and mobile network from Internet attacks.
• There are no defined mechanisms for controlling access to a femtocell-based service. Both femtocells and handsets accessing service through those cells need to be properly authenticated and individually authorized for access.

Network Operations

• Macro network radio planning and systems are not designed to support the consumer behavior of potentially millions of unplanned femtocell moves/adds/changes.

Standardized Interface

• The Iu-b interface is not 100 percent interoperable between different Node B and RNC vendors. A true mass market can only develop with a fully standardized and interoperable interface between femtocells and the mobile core network.

Network Offload of IP traffic

• A requirement has developed from operators looking to deploy femtocells to bypass the GSN infrastructure for non-operator IP traffic. Support for this feature in an Iu-b approach requires a significant amount of effort, including modification to the specification.

The SIP Approach

The idea of using SIP technology for integrating femtocell access points into the mobile network is appealing on the surface because it is the foundation of IP Multimedia Subsystem (IMS) technology. However, the SIP aggregation solution being promoted today for femtocells bears little resemblance to a true IMS architecture.

In this case, SIP is used as the protocol between the mobile core network and the SIP client on the femtocell device. But in the mobile core, a new SIP-enabled Mobile Switching Center (MSC) is required to operate the translation from SIP into existing network interfaces. In fact, the solution has more similarities to conventional UMTS core network functionality than to a true IMS implementation.

This new SIP-enabled MSC needs to support many of the existing UMTS MSC functions and interfaces to, for example, provide a single phone number (MSISDN) for each handset, offer a unified set of supplementary services, and support femtocell-macro cell handover.

The SIP switch would need to support UMTS Visitor Location Registry (VLR) functionality to route calls and it would also have to communicate subscriber-activated supplementary services such as Call Forward All Calls to the UMTS Home Location Register (HLR) so they will not be de-activated. As a result, using the SIP protocol to aggregate femtocells leads not to a new IMS core but to the purchase of an MSC dressed in SIP clothing.

Another issue with a SIP-based femtocell implementation is that it lacks feature transparency. The femtocell must translate every UMTS call-control procedure into an equivalent SIP procedure. This leads to a number of problems:

• The femtocells control the handsets, not the MSCs. Therefore, the femtocells must replicate every regular and supplementary service normally managed in the switch. Beyond the technical complexities, control of each service would not likely be the identical to that of MSCs, consequently the subscribers' experience may change as they moved in or out of the house.
• Each new feature for the mobile network must be supported in both the MSC and femtocells.
• Network-based features such as ringback tones must be implemented on the SIP MSC, and some may also required the SIP MSC to support Customized Applications for Mobile Network Enhanced Logic (CAMEL).
• None of the advantages of the SIP service control model extend to the UEs, which remain standard UMTS devices. And there is an additional dependency for each service: the femtocell SIP to UMTS service mapping function.
• At the other end of the session, the SIP MCS must stitch together SIP control with UMTS service controls for calls to other mobiles and to the PSTN.

The UMA Approach

Clearly, the existing RAN access infrastructure and SIP-based infrastrastructure are both ill-suited for femtocell deployments. The 3GPP UMA standard, originally defined to enable millions of dual-mode cellular/Wi-Fi mobile handsets to access mobile services over the Internet, can be directly leveraged to address this access network challenge. UMA provides a standard, scalable and cost-effective IP-based access infrastructure that can be leveraged by femtocells in the same manner as it is currently by used by dual-mode handsets and Wi-Fi access points.

Leveraging UMA for Femtocell Integration into Core Networks

The 3GPP UMA standard provides a standard, scalable and cost-effective method for end-user devices to access mobile network services over any IP-based access network, including the Internet.

As is often the case with a new standard technology, innovative companies learn to apply it in new and innovative ways. Recently, it has been demonstrated that UMA can address the core network integration challenge of femtocell-based service by providing a standard, scalable, IP-based interface into mobile core networks.

The functional diagram shows UMA supporting 2G and 3G Femtocell access points.

A UMA-based femtocell architecture offers numerous advantages over both an Iu-b-based architecture and a SIP-based architecture.

• UMA implements a collapsed architecture that pushes radio management functions to the femtocells, rather than splitting them between the RNC and the femto-NodeB.
• The collapsed architecture increases scalability, as do the control functions enabled on the UNC for support of intelligent femtocell access points.
• Because UMA integrates with the existing core network, it simplifies infrastructure rollout, reduces operational costs, and avoids network complexity.
• UMA provides full service transparency; all UMTS services are supported by the same core elements that service the surrounding macro cell network.
• Because it uses the existing network core, UMA architecture leverages existing emergency call support and lawful intercept functionality.
• Flexible service access controls are enabled by mechanisms built into the UNC that allow the operator to decide when, where, how, and under what conditions a device is allowed on the network,
• UMA discovery and registration techniques mean femtocell access points can be shipped with a single fixed image; they do not need to be individually provisioned.
• The UNC learns any information it needs to serve each FAP during the UMA registration procedure; it does not need to store FAP configuration or data.
• UMA provides a clear path to standardization; the 3GPP Generic Access Network (GAN)/UMA standard already defines the Up interface and the addition of Iu mode support is underway.

Femtocell Operation

The minimum functional elements for a UMA-enabled femtocell are a 2G or 3G air interface module, a UMA client module, a standard UMTS/GSM Subscriber Identity Module (SIM), and an IP networking interface.

The UMA client functions reside in the femtocell rather than the mobile handset, as is the case with UMA-based dual-mode cellular/Wi-Fi handset services. As a result, any standard, off-the-shelf UMTS/GSM handset can attach to the UMA-enabled femtocell's air interface module. The femtocell performs the interworking function between the UMTS/GSM air interface and the interface to the UMA Network Controller (UNC) in the core network.

Upon power up, the femtocell uses a standard SIM/User Services Identity Module (SIM/U-SIM) and the EAP-SIM protocol defined in the UMA specification to authenticate to the mobile network and to create a secure IPSec tunnel between the femtocell and the UNC security gateway.

The femtocell then uses existing standard UMA procedures to discover and register with the appropriate UNC. This ensures that handsets attached to the femtocell are always connected to the correct serving MSC and Serving GRPS Support Node (SGSN). This important and automated step minimizes the femtocell-macro cell planning and ensures seamless handover between the two access points. If the UNC accepts the femtocell UMA registration, the UNC provides the system information needed to go into service over the air interface.

Handset Operation

Handsets, upon arriving in the vicinity of a UMA-based femtocell, detect its presence through normal GSM and UMTS radio procedures. When the handset attaches, it triggers a UMA registration by the femtocell on behalf of the handset. The femtocell must successfully register the handset with the UNC, and the handset must be authenticated by the mobile core network to be authorized for service access.

This is a standard feature of the UMA registration procedure for each visiting handset and allows the UNC to provide network-based service access control per-subscriber and per-device. Operators have complete control of which subscribers and devices can access which femtocells, and can control access to specific regions or countries around the world. This capability also means MNOs can control access policies from the network without depending on femtocell-based access controls to be trustworthy.

UNC Operation

The UNC terminates the Upi interface from the femtocell. Almost all operations over the Upi interface are common between dual-mode handset services and femtocell AP services. Since both dual-mode UMA handsets and femtocells use the same Upi interface, the UNC will be able to support both types of access concurrently. For MNOs, this means a single UMA UNC investment supports both dual-mode handsets and femtocell access applications.

Conclusion

While much of the debate around femtocells today centers on the cost and RF issues of deployment, it is clear that the current Iu-b protocol is too costly and difficult to deploy for a massively successful femtocell roll-out. And a SIP-based architecture not only requires an overlay network but also offers few of SIP's benefits.

Only UMA meets the key requirements for the deployment of hundreds of thousands of femtocell access points:

• Enables cost-effective scalability
• Has low operational/deployment impact
• Leverages existing core network elements
• Provides full circuit and packet service transparency
• Has automated discovery/registration procedures for access points
• Includes integrated, proven IP security and encryption
• Provides network enforced security and access controls
• Has open, standardized 3GPP interfaces for femtocells

UMA is clearly the right technology at the right time to make commercial femtocell services a huge success.

Author Biography

Steve Shaw is the marketing director for Kineto Wireless and a leading evangelist for UMA technology. He can be reached at [e-mail] or at [phone].