Cutting the IP cord

An all-IP wireless network can help carriers provide mobile data services There is a stark simplicity about the telecommunications networks of yesteryear, especially when compared with what many industry observers believe is happening now and will continue to occur: growing complexity.

In the 1970s, networks were analog and hierarchical, with well-defined layers between local access and long-haul and international transport. Traffic was overwhelmingly wireline voice, and volume increased in steady and predictable increments.

Networks of the 1980s were marked by the advent of digital technology. Besides that, however, not much changed. Aside from X.25 and frame relay for relatively specialized corporate applications, the emphasis of networks was squarely on circuit-switching. Initially, installed technology managed to accommodate the boom in non-voice traffic such as fax transmissions.

Fast-forward to the early 1990s and the first stirrings of true complexity. Fax traffic began growing into a roar. Adding to the wireline network traffic jam was proliferating cellular traffic terminating on or originating from the public network.

Now it's 2000, and things in the public network are more complex than ever. The popularity of the Internet is skyrocketing and so is the resulting network traffic. Corporate data applications are multiplying. Cellular traffic is growing exponentially, and voice and data applications in the wireless world are rapidly converging.

Most existing networks were never designed to accommodate this level of complexity or this amount of data. They were designed, instead, to handle voice traffic. To add to the challenge for wireless network planners, more complexity and more data loom on the horizon as customers' communications needs grow increasingly mobile.

Craving convenience To satisfy the demand for mobility, wireless service providers face a particularly difficult challenge.

Next generation wireless data services are one of the major drivers behind increasing network complexity. They are also one of the most important because these services are appropriate for today's convenience- and speed-craving end user (Figure 1).

Industry observers expect a future explosion of entertainment-related wireless services, which will allow subscribers to download music, videos or interactive games. People also will be able to take advantage of location-specific services such as finding a specific type of restaurant located within 100 yards or 100 miles.

New features offered in wireless data services are causing end-user demand to accelerate. The 3G access technologies, for example, offer increased bandwidth for wireless data users. New 3G devices also promise upgrades such as better use of the scant space available on the LCD screen and location-sensitive services.

There also are benefits for carriers in 3G. Because 3G uses packet-switched technologies, carriers will be able to make more efficient use of allocated bandwidth spectrum. They'll be able to serve more customers with less bandwidth while providing more reliable and more attractive services.

Overall, the shift from a voice-centric to a data-centric traffic model has implications for network capacity and architecture. Today's circuit-switched networks are optimized for voice traffic, and upgrading them to accommodate data applications can be prohibitively expensive. Also, they are not designed to handle complex, next generation data services.

The trouble with data Most service providers today rely on circuit-switched network technologies for wireless services and features such as wireless roaming and mobile subscriber authentication, as well as basic call setup, management and tear-down. Many also rely on the same infrastructure for data applications.

The trouble is that circuit-switching equipment is an expensive and inefficient way to handle data traffic and the changing trends in service creation. Traditional voice traffic is characterized by point-to-point calls of short duration. Data calls, on the other hand, are characterized by multipoint sessions of long duration and large, variable-sized packets that flow asymmetrically.

With increasing data traffic from Internet use and other data-intensive applications such as multimedia services, circuit switches are becoming increasingly overloaded. As data sessions occupy ports on telephony switching gear in the central office, service providers are forced to purchase more switching gear. Traditional switching ports are relatively expensive and, therefore, a scarce resource.

The circuit-switched model employed today is optimized for voice communications. Consequently, it is designed to offer reliable service at peak demand times. As a result, much of the network capacity lies idle during non-peak times. Circuit-switched networks dedicate channels of bandwidth to communications sessions for the entire duration of each session.

In a circuit-switched network, wireless or otherwise, it is not possible for sessions to use statistical multiplexing to take advantage of the inherently bursty nature of most data applications. As a result, the available bandwidth of a circuit-switched network is rarely maximized. And because customers are billed for the duration of the connection rather than the amount of data transferred, service providers make less money on short, bursty data transfers such as e-mail.

Circuit-switched infrastructure components, because of their proprietary nature and centralized architecture, are difficult and costly to program. Upgrading or reprogramming a circuit-switched network to accommodate new kinds of value-added services can be time consuming and expensive, slowing time to market for popular new services that carriers increasingly rely on to differentiate themselves from competitors.

Today, a carrier will be successful if it can move away from commodity services into those that provide a much higher return on investment. While almost all service providers offer three-way calling and caller ID, consumers are beginning to expect high-value services such as the ability to make travel plans, access unified messaging accounts or play interactive real-time games.

New services, quickly The bottom line is that today's wireless service providers require data networks that provide not only higher bandwidth but also the means for quickly and cost-effectively deploying emerging services and applications across multiple access networks. The question is how to get there.

Ultimately, the best way to prepare wireless access infrastructures for next generation application delivery is using a network architecture that provides voice and data access over a single, shared IP core. This network core can transfer all types of traffic, including data, voice and video, regardless of bandwidth, session length and other variables, over a ubiquitous packet-switched infrastructure. This all-IP network concept is the basis for next generation wireless network architectures (Figure 2).

An all-IP network can offer several advantages. Because it is based on open standards, IP-based networking equipment can be easily, quickly and cost-effectively programmed and upgraded - usually in-house and without expensive outside contractors. And with an all-IP network, these open standards can be implemented throughout various layers of the provider's network. Broadband, satellite or traditional wireless access can be independent from call or session control, which itself is independent from services and billing layers and so on. Each layer interfaces with the others using commonly understood open network protocols.

With this network strategy, service providers can reduce costs by using a best-of-breed strategy, maximizing their buying power for the optimum equipment they need at each layer.

Phasing it in Of course, the move to an all-IP network won't happen tomorrow. Most wireless carriers are taking a careful, phased approach that leverages current investments in circuit-switched networks.

The first and easiest migration step is converting long-distance transport to use IP technologies. With an IP trunking solution, carriers still use switches to handle "intelligent" call control requirements - such as mobility, three-way calling or 800-number services - with existing mobile switching centers. But long-distance transport is accomplished through IP telephony gateways, opening several opportunities for carriers to save money and make money.

For example, instead of relying on another long-distance provider's infrastructure, a carrier can use its own IP backbone to carry long-distance traffic. The carrier could even open a new revenue stream by leasing its IP backbone to other carriers.

The next logical step is bringing IP-based equipment closer to the subscriber to provide intelligent network services such as three-way calling and call waiting. In this scenario, the radio equipment would communicate with IP access gateways to handle voice and data traffic, while the control node manages call control functions such as setup and tear-down. Instead of having call control and media handling bundled into one switch that is expensive to upgrade and program, these network services can easily be decentralized into functional entities that perform a specific set of tasks. The media gateway and signaling control node can interface via standard IP protocols.

This network design enables service providers to send all traffic, whether packetized voice or data, to the Internet or an intranet over a fully managed core IP network and add enhanced services as needed.

The benefit of this approach is twofold. Operating costs can be reduced because the process does not involve expensive proprietary hardware and software, and time-to-market for new services can be accelerated. IP-based equipment can be programmed and upgraded to accommodate value-added services in a fraction of the time and effort required for proprietary equipment. In the future, this solution also will allow seamless connectivity between services such as wireless voice over IP or data, wireless LAN/WAN or broadband, and it will provide interoperability between virtually all types of networks and equipment.

The third - and most difficult - step is taking IP to the wireless subscriber. Analysts predict that packet-switched devices won't be available for a few years. But most agree that an end-to-end IP network - from the transport carrier to the end-user device - is on the way. This phase will have the biggest payoff for carriers, enabling them to choose from a host of hardware and software suppliers for any part of the network. They also will be able to avoid the high cost and effort of maintaining two separate networks, one for voice, the other for data.

Another strategy to consider when migrating toward an all-IP network revolves around expansion. Carriers may want to consider trade-offs when making major upgrades or considering rolling out new services. Will it be more cost-effective to upgrade an existing circuit-switched infrastructure or to invest in new IP-based infrastructure components instead? The new infrastructure may have a lower cost overall, considering the cost of upgrading and programming circuit-switched equipment to accommodate value-added services (Figure 3).

While carriers have obtained a high comfort level with circuit-based networks based on more than 100 years of operation, they will have two choices in the near future: either maintain two radically different networks - one circuit-switched, the other packet-switched - or converge the two networks into one cost-effective infrastructure for both voice and data. For cost and other reasons, most carriers and service providers will opt for convergence.