Your browser is not supported. Please update it.

16 September 2005

IMS takes gradual steps, but for now vendors need it more than operators

The IP Multimedia Subsystem (IMS) is the most important single technology underpinning the move to next generation networks where mobile, fixed, wireline and wireless are converged to deliver a wide range of all-IP services. However, for all the headlines, few carriers are looking to deploy full IMS in the short term, and most still need convincing of the business case, as they come under increasing pressure from shareholders to cut back on capex and opex. The vendors may argue that telcos can steal a march on rivals by using IMS to move swiftly to converged networks and new services, and can use the architecture’s integrated functions to reduce overheads ‘ but the operators know that they will need to go through a great deal of justification before spending millions on an IMS and next generation services, when they are yet to see significant return on the investment on their current upgrades, especially in cellular 3G. This will favour a gradual approach in many markets, and one that could open the door to new suppliers, so that IMS could prove disruptive and lucrative to the major telecoms suppliers in equal measure.

The hesitancy of the operators to take a big bang approach is compounded by the emergence of competent technologies that provide a lower risk interim step to key IMS services and so enable the full IMS decision to be put off for a couple of years, or even shelved in favour of a piecemeal evolution. This is putting pressure on suppliers to offer relatively low cost approaches to IMS, by incorporating off the shelf technologies at an early stage, by moving quickly towards softswitches, and by supporting IP applications and structures that can be seen as a fairly low risk stepping stone to IMS.

Most tier one cellcos and wireline carriers are evaluating and even trialling IMS but few plan to deploy a full system within the next 12 months. Although IMS represent the greatest growth opportunity for many infrastructure vendors, since it spans all types of carrier including wireless, wireline and cable, their current offerings are immature and their potential customers hesitant. Also, new suppliers are coming into the market, which will become crowded in certain areas. All this is raising the challenge for vendors of getting a foothold in IMS, and creating a short term revenue stream, while buying time to get full products developed that will meet carrier business requirements.

As large telcos start to deploy IP-capable networks ‘ Verizon Wireless with CDMA2000 Rev A, for instance, or BT and France Telecom with their next generation network projects on the convergence side ‘ the demand for IMS should grow, although this will only happen where strong applications are created to justify the investment. Voice over IP and basic video will not be enough to make a carrier move swiftly to IMS ‘ early adopters will make their decisions based on gaining a competitive edge in the type of services they can deliver, and how quickly and cheaply this can be done. To date, some of the ‘first steps’ towards IMS, notably the Unlicensed Mobile Access (UMA) framework for converged voice offerings, and the evolving push to talk over cellular (PoC) standards, are the best examples of real converged applications that are starting to draw serious carrier attention. These will provide strong early positioning for their key suppliers and give the operators valuable experience in deploying converged IP services without the commitment of a full IMS. But they remain very far from the complete vision enshrined in the IMS standards, as defined primarily by the 3GPP and 3GPP2 groups, though based on classic internet standards, notably SIP.

Chad Hart, a director at research firm VDC, said: ‘IMS is the universally accepted network standard – a standard that will likely be mandated by nearly every carrier deploying new infrastructure before the end of the decade. While this is an ideal situation for IMS vendors, this will be a difficult market to capture for the next few years. IMS products have a long way to go, and it will take time before carriers accept the IMS business case.’

Carriers may be hesitant about taking the IMS plunge before products are proven and applications appealing, but few are staying entirely aloof either. For instance, the European Telecommunications Standards Institute has decided that all European next generation networks should be based on an IMS core.

Some operators are deploying IMS to meet an immediate need for one or two specific services that they believe will give them an edge, but many are taking a broader, if slower, view that IMS is the basis of transition to the next generation network. This is especially true of the wireline carriers, which are likely to be the earliest adopters of true IMS because of their motivation to converge wired and wireless networks and services. This gives them an additional impetus, over and above the key reason for any telco to consider IMS ‘ its ability to support the rapid deployment (and removal) of large numbers of multimedia applications, which can be targeted at any device and delivered over any network.

BT is an advanced case but typical of large carrier thinking ‘ prepare for IMS, but be cautious about implementing too soon. Few want to be pioneers in this massively complex and expensive platform. In the BT super-project to migrate to all-IP, 21CN, Ericsson is a key prime contractor. The carrier said that, while IMS was not an immediate selection criterion, Ericsson’s experience in this area was important. ‘It was important that we could show a clear evolution path from what we are providing now to where BT wants to move to in the future. BT has evaluated that migration,’ said the company. Ericsson is in the position here that all its rivals are after in the large operator sites ‘ taking an early foot in the door that should lead to major IMS business later.

While convergence is a major accelerant for IMS projects, another is the rise of the MVNO (Mobile Virtual Network Operator), which leases capacity from a cellular network owner in order to deliver mobile services under its own brand, often highly targeted to the customer base for its core products. The flexibility to deliver highly tailored services to ever more granular customer segments, over multiple networks, devices and user interfaces, and with a common link to billing and the customer databases, is what the IMS promises, combined with other technologies to simplify the creation and provisioning of applications to the levels of ease familiar in the PC internet world. MVNOs are often far more skilful than the traditional cellcos at targeting specific customer groups with strong brands and high ARPU offerings ‘ Virgin Mobile’s success in the teen market with its tie-up with music channel MTV was an early example. But the host network will need to be able to support more and more services and customer groups, and this will be a key factor in winning MVNO business in future.

MVNOs with active IMS projects include Virgin Mobile, Finland’s Saunalahti, BT, Disney, Amp’d and eAccess. These are MVNOs that are heavily focused on stealing a march through multimedia services, and on targeting selected customer groups to reinforce their core brands, and they are early adopters of 3G services.

Potential candidates to adopt IMS also include the VoIP providers, such as Vonage and AT&T, since their services are already SIP-based, but few are evaluating full IMS as yet; and the cable operators, which are currently evaluating UMA and SIP applications, but waiting for CableLabs’s promised PacketCable Multimedia specification before making broader plans.

Most of the early activity will come from the major operators however. While about 70 vendors now claim IMS solutions, only a dozen carriers have gone public with specific plans and launch dates, despite a host of trials and evaluations, so gaining early revenue is a tough battle. As the chart shows, the most common applications for first stage IMS deployments are VoIP, push to talk and UMA convergence.

Among the wireline carriers, most tier one players are evaluating IMS, with the most advanced triallists including BellSouth, BT, NTT of Japan, France Telecom and MCI (soon to be part of Verizon). Among the cellcos that are furthest along with IMS trials are MMO2 of the UK, TeliaSonera from Scandinavia, Orange, Telecom Italia Mobile, Sprint Nextel and China Unicom.

Apart from unifying networks and delivering a broader range of services more quickly ‘ all of which should boost ARPU and customer loyalty significantly – the other role of the emerging software architectures is dramatically to lower operating costs, particularly for multi-network carriers converging a variety of infrastructure into one IP framework. We have analyzed many times how the potential return on investment in 3G networks, especially in Europe, is under threat by new expectations of wireless data pricing, with users accustomed to Wi-Fi demanding flat rate schemes at low costs, and the prospect of WiMAX only accelerating the trend. The only way that carriers will get ROI on 3G networks is to slash the costs of delivering 3G services, as Vodafone chief Arun Sarin recognized last autumn when he pledged to cut capex and opex by £1.4bn a year.

One of the key costs is making new applications available. Currently, a new service costs an average of $5m to deploy and large cellcos typically release four a year. Motorola believes the target should be one per week at around $150,000 each. To achieve this, it offers an IMS framework with a full Java object repository to create, test, deliver and execute new services. These services can then be targeted individually to different networks and devices, but this requires minimal effort and expense compared with creating each service individually for each platform.

However, to arrive at these savings and justify the cost of an IMS on opex grounds, carriers will need to be deploying a fairly large number of applications that are all likely to improve ARPU quickly, which will not be the case for small players.

‘It’s very hard to find a single application that you can roll out using IMS that you couldn’t implement without it,’ Alan Stoddard, general manager of converged networks at Nortel, said in a recent interview. ‘It only makes financial sense when you get to the third or fourth application; it doesn’t make any sense for just one.’

This, then, also favors a halfway house approach whereby operators deploy some IMS-type services without the full framework.

Some vendors doubt whether the big bang IMS approach will ever take off, believing that, while it has huge attractions for large suppliers, most operators will make their way to IMS piecemeal, as part of a gradually evolving next generation network. Ironically, the king of IP, Cisco, is one company that seems to sit in this camp. Many Cisco executives have recently criticized IMS as too complex andcostly, and with few benefits for end users, and Cisco is reworking existing products rather than creating a true IMS. Cisco describes these products, which center on its Call Session Control Platform (CSCP), ‘a family of SIP Proxies (including the CSCP) that are aligned architecturally with the IMS Core Session Control Functions (CSCF) and support specific CSCF interfaces and protocols’. It will add individual IMS interfaces as the market demands them rather than plugging them all into the platform from day one.

The key stepping stone to IMS remains UMA, explaining the interest of the most IMS-focused vendors, notably Motorola. UMA is a 3GPP standard for the GSM operator community that allows hand-offs between the GSM/GPRS network and a landline or Wi-Fi network. UMA is a step on the IMS road, although it lacks full support for SIP as yet. It is the basis of converged services such as BT Fusion that allow subscribers to use a single handset to access the wireline network (via Bluetooth or, in future, Wi-Fi) plus WLans and cellular networks, depending on availability.

Steve Shaw, marketing director at Kineto, the start-up that provided key technology for UMA, says the standard will be a first step for many GSM carriers into IMS services, even if UMA lacks SIP.

An example of how UMA might be a forerunner of IMS is Nokia’s recent contract with the Finnish MVNO Saunalahti. The carrier will use both UMA and IMS from Nokia, which also is testing its UMA solution with several other operators and provides handsets as well as network infrastructure. Chris Ebert, director of 3G convergence and applications for Nokia, said that some operators will use UMA as a standalone technology, but many will see it as a stepping stone to IMS because it provides the access technology that IMS currently doesn’t have.

There are problems for UMA in its current form, though, notably its lack of support for 3G networks and for handset-PC communication. Also, there are security concerns, centered on whether Wi-Fi access could open a tunnel into the cellco’s network, and although UMA has some security built in, some cellcos are waiting for more robust solutions to be added, such as the UMA Security Gateway, recently launched by Reef Point Systems. This claims to handle 500,000 simultaneous connections with intrusion detection, denial of service attack prevention, custom firewall filtering and dynamic virtual routing.

However, the major vendors are aware that, despite some shortcomings, these early stage IP platforms risk placing the early moves to all-IP in the hands of new suppliers, and of prolonging the evaluation period for full IMS, thus delaying their opportunities to make major revenue gains. This is why Motorola and others have become so interested in pre-IMS technologies such as UMA and PoC. Lucent, one of the most advanced strategists in terms of placing IMS and multi-network convergence at the heart of its approach, is keenly aware of the difficult balance between gaining market awareness through engaging in trials, and sustaining the possibly long delay before full scale roll-outs.

In her last quarterly results briefing, Lucent CEO Pat Russo highlighted the dilemma that faces all her rivals too. ‘IMS is a footprint game,’ she said. ‘We need to be in trials, we need to be working with customers. But it’s a very complex space, as it’s a reference architecture that covers many network elements, and these trials can last up to a year.’
She added that Lucent was now in 50 trials with 14 customers, an increase from the second quarter, but ‘there’s not much spending in this space at the moment’.

‘We have some revenues from products that are considered IMS elements, such as our super distributed HLR [home location register], but these aren’t strictly IMS revenues. We won’t see spending from customers deploying IMS on a mass scale until 2006 and 2007. But it’s important to remember that IMS is about delivering applications that will drive traffic, and that in turn will drive further demand for network capacity,’ she summed up ‘ as well as creating significant systems integration requirements for the large telco suppliers and for companies like IBM.

Integration will increasingly become the focus for these large vendors, as they can no longer hope to provide the whole IMS solution from end to end as they did in previous telecoms generations. The telco suppliers have been used to operating in a closed world where they controlled the whole infrastructure system including the control layer and applications layer. Now they not only have to make a complex system, developed with massive R&D cost, financially attractive to their key customers, but also fend off the entry of new players. Since nearly all telcos now mandate that the standard ISC interface, which sits between control and application layers, is supported, third party applications server providers and software developers can grab a piece of the pie ‘ although in most cases, the major vendors remain prime contractors at least.

Initially, application servers will be the key component of IMS and this is opening a huge opportunity for companies with an enterprise heritage to transport their products to telco land – IBM WebSphere and BEA WebLogic being good examples.

The step-by-step approach being adopted by many operators also creates chances for small software houses that can provide a spot solution that is sufficiently attractive to telcos looking for their first convergence move. Examples of early IMS-type approaches where start-ups will, temporarily at least, thrive include WLan-cellular roaming, favoring companies like Kineto and Bridgeport; and interactive gaming, with companies like Brooktrout and Terraplay providing enabling technology.

The importance of small specialists in the early days of IMS is illustrated by the top rank of IBM’s software partners in its own IMS initiative, which are hardly household names – AudioCodes, BayPackets, Brooktrout Technology, Leapstone Systems, NexTone Communications, Qualphone, Terraplay Systems and Ubiquity Software are the most prominent. Yet all provide an important component of an IMS picture ‘ AudioCodes, for instance, makes a Media Gateway for seamless VoIP connectivity between existing IP PBX and PSTN and mobile networks; Brooktrout’s Media Server will underpin an IBM SIP-based prepaid interactive gaming platform that spans fixed and wireless connections; NexTone specializes in VoIP-to-IMS interconnectivity; and Ubiquity provides an SIP Application Server for creation of new services using IBM tools such as Rational Unified Service Creation Environment.

Such approaches may make the road to IMS a painfully long one, from the point of view of vendor cashflows, but they also increase the likelihood of IMS deployments actually delivering the results that telcos are after, by reducing risk and allowing operators to experiment with one or two key services before taking the plunge.

But this will not stop the suppliers seeking to accelerate the process as much as they can, and one way will be to reduce the cost of IMS, particularly by turning to off the shelf hardware components. The key standard here is AdvancedTCA, a backplane design based on commodity components and heavily supported by Intel. Makers of communications servers, such as Alcatel, Motorola and Sun, are supporting ATCA, while IBM is working with Motorola’s Avantellis server architecture. ATCA, which has emerged from the world of the PCI computer bus, provides a standardized chassis and blade server architecture that supports very high rates ‘ 2.4 terabits per second on the backplane ‘ for next generation multimedia applications. It holds out the promise for infrastructure vendors of upgrading the chassis using commoditized components and mass market silicon, greatly reducing their costs as they come under heavy pressure from operators to provide platforms with lower associated capex and opex.

Motorola and Alcatel are taking leading roles in applying this to IMS at an early stage to make the underlying infrastructure costs more attractive. Last month, Motorola announced an important joint venture with Intel, reflecting a similar one by the chip giant and Alcatel, to to create a proof of concept for IMS network functions based on Motorola’s Avantellis server. Motorola said that using standards-based communications servers as the foundation for IMS network elements will speed the deployment of the infrastructure.

Such moves will help to reassure carriers, as will the opportunity to trial IMS-like services on a limited scale for 2006-7, but the most ambitious convergence projects will require bolder IMS moves, and we should see the leaders, such as BT, Sprint and France Telecom, making major investments from about a year from now. Well implemented, an IMS could clearly give these early movers significant market advantage in terms of the services they can deliver at reasonable cost, but we have seen many major telecoms architectures before that promised enormous benefits in theory, but failed to find the applications to exploit those benefits (arguably, 3G still falls into this category). All eyes will be on the applications community, from start-ups to majors, to come up with the services that will justify investment in IMS, and make the new networks ‘next generation’ from the end user’s point of view, not just that of the underlying technology.