Just as mobile carriers that once dismissed Wi-Fi now embrace it, so mobile WiMAX is becoming more important as a complementary technology to 3G than an alternative, certainly in markets where cellcos are well established and there is limited opportunity for start-up or wireline players to penetrate the purely mobile sector. But talk of hybrid 3G/WiMAX networks is almost always over-ambitious. With hand-off between Wi-Fi and cellular networks only in its early stages, despite all the hype, WiMAX is even further behind. And while the deployment of the IMS (IP Multimedia Subsystem) by large carriers will help deliver IP services uniformly over multiple networks, there are still many complex issues of integration, roaming and security to be addressed at the network level.
In most WiMAX deployments of any scale, particularly in the mobile world, the technology will be adopted alongside other networks. It will be used to enhance the capabilities of a wireline system by extending coverage and adding portability, or to support full fixed/mobile convergence, and it will offer a broadband wireless data option for cellular and satellite providers to use to complement their core systems. As interest in such cases rises, it is becoming a matter of urgency that WiMAX should be able to integrate effectively at all levels with other networks. This will increase its appeal to large providers and will help drive adoption, rather than hostility, in the 3G community.
The priority is to find relatively simple and cost effective ways to integrate WiMAX access into existing networks, often before an operator has adopted a full-blown convergence technology such as an IP Multimedia Subsystem (IMS).
Cisco is demonstrating a similar pragmatic approach to WiMAX-3G integration as it is to IMS, offering step-by-step convergence rather than an expensive big bang, and as we are already seeing in Wi-Fi/3G models, this caution is likely to prove popular with capex- and risk-averse cellcos, even if it achieves less dramatic results in the early stages. Cisco is offering the Wireless Integration Platform (WIP) and Intelligent Packet Solution (IPS) as the keystones of its strategy and looks set to use such products to gain partnerships with WiMAX and 3G equipment vendors, such as Siemens, rather than seeking to compete in WiMAX hardware itself.
As in many other areas (see Wireless Watch January 23 edition), Cisco is anticipating industry standards work and no doubt hopes, thereby, to influence those standards bodies to adopt its technologies. Often criticized for sitting on the fence over WiMAX, it is increasingly clear that the company is actually far ahead of many other players ‘ it may not have hardware close to market, but it is making early strides in an area that will give it far more influence and revenue, the software frameworks that will enable WiMAX to play a full role in an all-IP network. Along with Qualcomm, Cisco has some of the most advanced knowledge and intellectual property in mobile IP and can be expected to contribute significantly on this front, bolstering a standard that has been conceived for the lower two layers of the network and is not always sophisticated in the area of routing. Work on this from WiMAX supporters will be essential, since Qualcomm has the chance to assert its own advantages in this respect, possibly to promote an alternative to 802.16 such as its own Flash-OFDM.
Cisco, while it protects its huge market share in its traditional markets, such as enterprise Lans, is aware that, in its newer territories, it will do better to keep a guiding hand on middleware technologies ‘ as the world moves towards multivendor IP ‘ than trying to build up 70%-plus shares of hardware markets with increasingly low margins. It has taken this approach in IMS and the same is seen in WiMAX integration. In the former, Cisco was almost alone among the networking majors in staying ambivalent about the standards-based control architecture, and when it finally released it own platform late last year (see Wireless Watch December 12 2005 edition), it remained the voice of caution. While it will support more and more IMS functions, it will differentiate itself by providing a gradual approach that positions IMS as a platform for future new services, without the need to migrate legacy systems any time soon. The company chose to rework certain existing products so it can support non-IMS and fully standard services, with the structure based around Cisco’s Call Session Control Platform (CSCP), a family of proxies for SIP (Session Initiation Protocol, the keystone of IMS) that were designed to be aligned architecturally with the IMS standard Core Session Control Functions. The idea is that Cisco’s own SIP elements would map to standard IMS functions, so that the latter could be introduced relatively simply, and one by one, as the market requires, without major architecture reworking. Therefore operators are promised a lengthy grace period in which they can move at their own speed from current to IMS functions, mixing and matching the two.
A parallel approach is now visible in WiMAX-3G integration. While the WiMAX Forum is currently working to define a standard way to map radio QoS (quality of service) to IP transport QoS, in order to support mobile/wireless IP and integration with cellular networks, Cisco has leapt ahead with its own mapping method for 802.16 and supports all the QoS classes defined in 802.16 and by the 3GPP. Cisco says: ‘A simple non-standards based (but very functional) QoS mapping can be applied immediately’, a comment that represents the Cisco approach neatly, attracting customers with a simple solution that pre-empts the standard. This then creates a body of user support and increases Cisco’s chances of driving standards efforts. ‘Siemens and Cisco are leading the effort of the WiMAX Forum to standardize WiMAX to 3GPP/3GPP2 interconnectivity,’ Cisco explains in a white paper. ‘Given the above described service classes and the Cisco comprehensive QoS implementation, any service class may be supported within the physical limitations of the deployed network.’
Back at the operator’s core network, Cisco offers WIP and IPS to integrate WiMAX base stations and access networks into a mobile or fixed service provider environment. These can be applied to any IP technology including Wi-Fi or, presumably, Flarion Flash-OFDM, although we can expect Qualcomm to be working on some advanced 3GPP2/IP integration techniques of its own in readiness for its acquisition of Flarion being finalized.
The WIP is the hub round which all the main core network functions are organized. These include authentication, billing, content management, roaming and other modules. User traffic reaches the WIP through its access gateway and is then fed into these various functions, which are linked in turn to back end databases and infrastructure such as accounts systems. Mobile IP (MIP) based handover will be used when the WiMAX user is in portable mode, and to enable data session continuity between Wi-Fi, WiMAX, and UMTS/GPRS. Although WIP is generic technology, it has been specially adapted for operators that want to incorporate WiMAX into existing 3GPP networks, with the specific features modelled on those already created for WLan/3GPP.
Again, Cisco offers a short term, ‘quick and easy’ option as well as a longer term, more robust one. For operators wanting to move quickly, loosely coupled interworking of 3G and WiMAX can be employed, similar to existing Wi-Fi solutions. Loose coupling does not require 3GPP protocols to be terminated at the WiMAX base station and so avoids changes to WiMAX standards and minimizes changes in existing 3GPP specifications.
This clearly fits the current WiMAX Forum objective of accelerating integration with 3G networks in order to stimulate uptake and reduce time and cost to market for vendors. Cisco’s offering for loose coupling is IPS. Until now this has been implemented to integrate 2G/3G access technologies.
In tightly coupled interworking the WiMAX gateway is connected to the 3GPP core network (SGSN) in the same way as the 3GPP access network elements and the 3GPP protocol stack must be run on top of the 802.16 network interface cards, switching between physical layers as required. The 3GPP and WiMAX networks share the same authentication, signaling, transport, and billing infrastructures, but exposing the 3GPP interface can imply security threats, while injecting WiMAX traffic directly into the 3GPP core could upset the carefully engineered core network elements ‘ so there will be extensive re-engineering to adapt to the increased load and differing traffic characteristics.
While Cisco drives efforts at the core, at the client end, integration will be driven partly by multi-network devices such as dual-mode handsets, or by subscribers requiring services and user interfaces to be consistent across multiple devices ‘ from set-tops to cellphones ‘ which may actually be attached to different networks. In this quest for convergence, providers will need to attend to the access portion of the system as well as core interworking ‘ and for several years the key access technology will remain Wi-Fi, even if it is working as a front end for WiMAX.
There are several access-oreinted standards in the offing, covering hand-off and other aspects, but in many cases these are being pre-empted by proprietary solutions that the more urgent operators are keen to adopt in order to gain market edge. The main emerging standards to address handover fall into two groups – homogeneous handover within a single network, and heterogeneous handover across dissimilar networks and with multi-radio clients. For the former, there will be 802.16e for mobile WiMAX, 802.11k (load balancing) and 802.11r (fast handover) for mobility in Wi-Fi systems, and, of course, the well established 3GPP/2 specifications. Most Wi-Fi hardware already incorporates the more limited 802.11f, which allows roaming among access points on the same network segment. The main aim of 802.11r is to speed up handover to the levels required by voice, less than 20 milliseconds. In particular, there needs to be pre-authentication of users before they reach an access point, as the requirement of the 802.11i security standard, to reauthenticate each time a subscriber roams to a new AP, slows down handover intolerably for voice and video. Until 802.11r is complete, enterprises will have to use proprietary hardware for fast roaming, or the mobile IP standard, in which the home IP address is static but is redirected to the changing IP address as the device moves between base stations. The ‘r’ standard should be finalized by April 2007, with a harmonized proposal now achieved from an original list of eight, the successful specification largely based on Cisco/Airespace technology.
A key standard for heterogeneous handover, between different networks and client radios, will be the emerging 802.21 Media Independent Handover standard. This aims to formulate procedures allowing mobile devices to recognize and analyze the radio environment around them; different layers in the software will then suggest, and eventually trigger, a switch from one network to another. The 802.21 work is in its early stages but will eventually provide link layer intelligence and other related network information to upper layers to optimize handovers between heterogeneous media, whether users are mobile or stationary. The latter becomes important for ‘intelligent’ services where cognitive radios automatically adjust to the best connection available (as defined by various criteria such as signal strength or tariff). Even when stationary, the environment around the user may change to make one network more attractive than another, or the user may switch from maing a voice call to downloading a large file and so require a different network for optimum performance.
Under 802.21, the mobile client is capable of detecting available networks, and the infrastructure can store required network information, such as neighborhood cell lists and the location of mobile devices. In general, both the client device and the base stations can support multiple radio standards and sometimes more than one interface simultaneously.
The standard supports transparent service continuity while a mobile node (MN) switches
between different link layer technologies. This architecture relies on identifying a mobility management protocol stack within the network elements. This protocol includes the Media Independent Handover (MIH) function.
Higher up the system, supporting fixed/mobile convergence of services over different networks, there are two key technologies emerging ‘ the IP Multimedia Subsystem and UMA (Unlicensed Mobile Access). Although these are often positioned as competitors, in fact both have very different characteristics and the latter is far more limited in functionality.
While both IMS and UMA are 3GPP standards, there are various proprietary attempts to offer integration between Wi-Fi/WiMAX and 3G or fixed networks, all claiming to be pre-empting the standards and offering superior performance. One example is Azaire Networks, which joined the WiMAX Forum at the start of the year. Its core product is the IP Converged Network Platform, which integrates Wi-Fi with 2.5G/3G networks and will now add WiMAX support.
Two other small vendors represent the IMS and UMA viewpoints in the convergence picture. Kineto Wireless was a major contributor to UMA before the technology wasa turned over to the standards community, and works closely with major UMA backers such as Motorola. On the other side of the fence, BridgePort Networks, whose platform is supported by IBM and Cisco as part of the MobileIgnite Alliance, criticizes UMA for not using the SIP protocol, on which IMS and most IP services are based. This is a real flaw in UMA, since the logical strategy of implementing that technology in the first stage, for relatively simple, voice-centric integration of networks, and then moving more gradually to the full-blown IMS, is complicated by the different protocol base. BridgePort claims its products are a true stepping stone to IMS because of its SIP orientation. Chief technology officer Sanjay Jhawar says that, now that the 3GPP and 3GPP2 have finalized technical requirements for IMS-based hand-off between cellular and Wi-Fi networks, ‘there’s enough information on the technical requirements to go ahead and begin building products’we’ve been waiting for this moment’.
As a result, BridgePort has upgraded its key product, NomadicOne Network Convergence Gateway, to support 3GPP and 3GPP2 IMS requirements, launching the NomadicOne IMS Convergence Server, and application server which works with IMS. Features include, as well as simple handover, support for mobile IP Centrex, providing access to the same phone number and the same services over both the mobile phone and the hosted IP telephony system in the office.
BridgePort has also announced SoftMobile, a back end system that combines the NomadicONE Network Convergence Gateway with a standard mobile phone and a soft client to enable a broad range of voice and messaging solutions. “It’s a fast-to-market solution for convergence that doesn’t require a new kind of phone that has Wi-Fi and cellular in it,” Jhawar explained, and is designed to allow cellcos to compete with services like Skype, allowing users to receive VoIP calls at their mobile number.
It is important that vendors do not start creating another religious war out of UMA and IMS however. Nokia and Motorola both insist that the two have related roles and will often be implemented in tandem, or at different stages, within the same carrier. There is rising demand, however, for a more closely integrated hybrid platform and Cingular is already known to be trialling such a system, created for it by Kineto. A hybrid would also benefit larger vendors that have put large investment into IMS but have taken less interest in UMA ‘ which to date has mainly seen trials just in Europe – such as Nortel and Lucent. Nortel has been fairly dismissive of UMA, claiming it is too complex to make it work flawlessly, especially for voice. The company says it will OEM a UMA product if required but will not develop one, as it sees it already as a commodity. ‘UMA is simply an access technology, reusing whatever services are available in a G/U network ‘ MSC voice services, GGSN services, IMS services. Every component of UMA is already a commodity ‘ Wlan, DSL, IPsec gateways.’
However, Nortel points out that IMS services can be delivered over UMA access, as well as over 3G or GSM, and so the two should not be seen as incompatible.
In the end, though, the key appeal of UMA is to prolong vendor and carrier investment in legacy networks, bringing traffic to the traditional mobile switch for a longer period while delaying migration to new models supported by IMS and VoIP. UMA’s support for VoIP, while relatively straightforward, is matched or bettered by that of the IMS ‘ more commonly associated with excellence in multimedia. As well as SIP-controlled media services, IMS supports Voice Call Continuity, which offers similar functions to UMA but with full SIP integration.
While UMA has a market today, its uptake will be limited by perception that it is an interim technology and by the fact that it is not fully formed ‘ handover algorithms are still not defined to the satisfaction of large users and there are gaps such as measurement reporting tools.
This is just one example of the complex decisions facing service providers as they move towards the integrated all-IP network, supporting fixed/mobile convergence and a combination of wireless and wireline technologies. Two things are certain but often ignored ‘ the road will be long and expensive, and there will be many interim steps along the way, which may seem a long way from the integrated IP dream, but will provide real functionality boost without the economic and technical risk of achieving a ‘big bang’ move to all-IP using immature systems and half-formed standards.