In a class by itself

Carriers looking for the fast lane into next generation services should consider class-independent switching

The case for converging voice and data networks is compelling. Operators have calculated that moving telephony traffic to packet networks can save up to 70% in operating costs. And cost savings are just one of the benefits. A converged infrastructure makes it easier to deliver emerging services, such as unified messaging, voice-enabled e-commerce and Web-enabled call centers, that transcend the voice/data boundary.

But despite the economic benefits, a huge and still-growing voice market and many years of development of voice over IP and other media conversion technologies, voice and data networks remain largely separate. A major factor that has inhibited convergence is that the solutions appearing so far have lacked the flexibility and scalability to deliver true convergence in today's competitive and deregulated environment.

A possible answer to these challenges comes from a breakthrough technology known as "class-independent switching." By using this technology to build class-independent networks, a new generation of converged network carriers can bring huge cost improvements to voice while adding new value-added services. Converged access providers will realize the true potential of broadband access by integrating voice and data. And incumbents can leverage this technology to meet this new competition and protect their installed base.

Challenges of convergence

Some challenges that have inhibited network convergence are technical. For example, voice demands more from network quality than data. While voice-over-IP technology is mature, implementations so far have not met the levels that national and international operators require. Furthermore, voice communications take place within a fully interconnected global network - voice services cannot compete in the mainstream voice market unless users can participate fully in this global network.

Compounding these technical challenges is the difficulty in predicting what capabilities the converged network should have. Legacy telephony networks were built when competition was limited, and regulation tightly constrained the types of services each market participant could offer. As a result, they are based largely on a class hierarchy mandated by the separation of long-distance and local calling services.

In today's competitive, largely deregulated environment, every market participant wants to be free to offer a wide range of services, which means leaving behind the regulation-driven class hierarchy of the past. To be successful, a converged network carrier must be able to tailor service packages to the needs of different customer segments and to rapidly adapt its service portfolio to meet customer requirements, market opportunities and competitive threats.

Class-independent switching

Implemented in next generation telephony switches, class-independent switching has four key properties that break through the limitations of previous convergence solutions:

- It does not require media translation to deliver legacy voice services.

- It enables media conversion technology and associated services to be deployed on a very large scale.

- It provides complete integration between packet-based voice, data and fax and their telephony equivalents.

- It allows operators to offer any mix of services to any subset of users and to easily extend and adapt these service packages as needs change.

With class-independent switching, new telephony service platforms can provide all the services of a Class 4 or Class 5 telephony switch in any combination and services unknown in the legacy telephony environment (Figure 1).

Operators planning a converged solution must address the technical and operational challenges of its core applications while providing the flexibility to respond to market development. Operators also must have the tools to migrate from where they are today to the converged future. It is therefore necessary to examine some of the issues of each application type.

Long-haul applications

The long-haul telephony infrastructure is built from a network of tandem switches interconnected by dedicated trunks. The starting point for a converged network is to replicate this solution with a packet or cell infrastructure. This is accomplished by using next generation switches in place of the tandem switches of the traditional solution (Figure 2).

These next generation switches have two types of functions:

- They convert each telephony call into the appropriate packet or cell format.

- They perform the switching functions of the tandem switch.

However, meeting the needs of the emerging generation of converged network carriers brings other requirements. For example, regulations traditionally imposed restrictions on which value-added services long-distance carriers could offer.

But in the current environment, all service providers - local and long-distance - need to offer a full assembly of value-added services to remain competitive. Many of the features needed to support them, including integrated messaging, multimedia conferencing and Centrex services, are beyond the scope of a traditional Class 4 tandem switch.

Furthermore, the converged world will bring demands for more complex interactions. The major new factor is the origination and termination of voice and fax calls on PCs and other IP terminals. Examples include a voice connection from a multimedia PC or connecting calls from standard telephones to IP-based call center switches.

Another requirement is ingress switching. A significant proportion of voice and fax calls received by an IP network operator will be switched to the public network rather than transported across the operator's IP network. Until global coverage is achieved, calls destined for locations not yet served by the IP network will be switched onto low-cost bulk time division multiplexing network facilities and carried to their final destination.

Access applications

Most telephony access continues to be via conventional local loops terminated in a Class 5 switch. However, much of the growth in capacity is used by modem connections for Internet access. The goal of a convergent solution is to take this data traffic off the voice access equipment at the earliest possible point.

The approach is to drop modem traffic directly into a next generation switch that converts it to media format (Figure 3). This solution can be deployed by the local exchange carrier (LEC) or by a co-located competitive LEC offering local bypass services.

With DSL and cable modem access, operators face the opposite problem. Today, these connections are largely limited to data services. In these cases, the goal of a convergent solution is to allow this access infrastructure to be used for voice and data by interconnecting it to the voice network and providing the services required (Figure 4).

Both of these access applications require the integration of packet and telephony facilities and demand that the next generation switch be media-agnostic. Also, to avoid performance penalties, each traffic type must be switched into its native format.

The ultimate goal in these local access applications is to eliminate the Class 5 telephony switch for voice traffic entering the network through digital access connections such as DSL and cable modem links. The only way operators can meet low-cost and high-performance goals while enabling the delivery of high-margin value-added services is by using switches with class-independent switching.

Value-added services

Core telecom services provide little opportunity for differentiation, resulting in intense price pressure, reduced margins and high customer churn. In the case of telephony services, these pressures will be exacerbated as convergence enables even more carriers to compete.

The best defense against commoditization is the introduction of value-added services such as messaging (voice or unified), one-number services, personal assistant services, three-way calling and content services. Value-added services can generate incremental revenue, stimulate additional network usage, slow price erosion and build customer loyalty.

As the number of services grows, so does the overall complexity and the potential impact on scalability. With so many variables, growth in customer demand for each service is nearly impossible to predict, making flexibility in provisioning essential. And large numbers of services must be easy to manage and maintain.

Marketing presents another challenge, requiring the deployment of value-added services in the form of tailored packages for each group of users, with the user groups defined by need rather than physical location and not limited by trunk group or subnet. For example, it must be possible to offer employees of a large corporation a customized package of services whether they are located at an enterprise site, are telecommuting from home or traveling on business.

Services in converged networks are defined by softswitches and media switches or by gateways that can come from an array of companies. Class-independent switches interface to softswitches via a service provisioning interface. The key to meeting the scalability and flexibility requirements of value-added services is this interface (Figure 5).

To avoid scalability and flexibility traps, the next generation switch must support the key interface standards used today and be extensible to support new standards as they emerge. Vendor-specific or application-specific interface implementations should be avoided because they cannot adapt to changing needs. And implementations in which the next generation switch provides an "open" application programming interface are risky because the service provider must believe that softswitch and next generation switch developers will choose to support it.

Operational requirements

In addition to the application-specific requirements described previously, several requirements are common to all converged applications.

Scalability is perhaps the defining requirement of telecom networks. No other information system supports tens of millions of users or shares a common infrastructure with hundreds of millions of others. Over the years, telcos and their suppliers have developed ways of delivering solutions on a very large scale with utility-class reliability.

One of the biggest challenges of convergence is moving solutions from the lab or small-scale pilot to a telco-scale deployment. Technical choices that are acceptable in a private data network often will fail at the scalability hurdle.

A critical test of scalability is how a system grows as users or services are added. A good solution scales linearly; a bad solution scales exponentially. One example is the choice some next generation switches have made to use dual buses, one for each type of traffic. Another is the need to dedicate each module to a specific set of services in most first generation solutions. These solutions work fine in a small environment but do not pass the scalability test.

Migration is another factor to consider. When moving toward a converged network, operators must preserve their investment in the existing infrastructure. The converged solution must interoperate fully with legacy telephony and packet services - not just at a transport level, but also at a service level. It also must be flexible and easily adapt to the new functions required as the migration progresses.

In planning their infrastructure, service providers must prepare for change driven by customer requirements, market developments and competitive pressures without knowing exactly what form it will take. Applications that are virtually unknown today will be in common use in five years. Internet-based application service providers may revolutionize corporate computing, and unified messaging might transform personal communications.

The best course is to choose modular solutions linked by standard interfaces, to choose open architectures not locked to the specific applications that are built on them and to choose solutions that will scale as new requirements bring demands beyond those anticipated today.

Finally, the industry must settle on a common solution. For all the challenges they have faced in the past, most network operators have operated a fairly homogeneous service. However, in the developing competitive service market, homogeneity will be the exception. A growing number of services must be packaged for different user groups, and complexity is increased further as operators move into each other's territory.

But operators cannot afford to have different solutions to meet each application need. Each separate solution incurs a substantial overhead in training, developing operating procedures, integrating management and billing systems and sparing. To keep these overheads from spiraling out of control, operators should introduce a common solution to address the broadest possible range of applications.

The road ahead

Pulled by customer demand for Internet-based multimedia services and pushed by deregulation and technology, most carriers are planning for multiservice networks that can handle fast-changing patterns of voice and data traffic. But although media conversion technologies are mature, many packet telephony solutions today have critical limitations.

Class-independent switching overcomes these limitations for the first time, bringing network operators a range of market opportunities and financial benefits. New entrants can easily offer voice and data applications with a rich array of services. Local carriers can move into long-distance applications, and long-distance carriers can deliver local access, all from a common platform. Converged network carriers will have the flexibility to define service packages to meet their customer, competitive and business requirements.

Class-independent switching brings the flexibility to interoperate with existing equipment and to address convergence opportunities from a single platform. Network operators will no longer have to deal with the costs and operational headaches of using different classes of switches in different parts of their network. Class-independent switching clears some major obstacles from the road to next generation services.