Can IP Save The World?

Service providers still face the challenge of carrier-class voice over IP

Voice over IP is a hot buzzword when it comes to next generation networks. Some people make it sound as if VoIP will save the world. Puffery, yes, but the argument can be made that VoIP might just save some companies as they work to remain competitive in a marketplace with decreasing margins for long-distance services. With the continued thrust to operate in a data world, the convergence of voice and data will be the competitive difference for certain applications and customer segments. VoIP will help increase margins by decreasing start-up costs and recurring operational expenses while creating the ability to offer new, value-added services.

Building a VoIP network makes good business sense. Not only is packetizing voice a more efficient use of bandwidth, per port operating costs are lower with VoIP vs. a traditional time division multiplexing environment. Service providers also can reduce real estate and power costs by up to 75% since a VoIP gateway takes up about one-tenth the space of a traditional circuit switch. Lower costs mean companies can quickly move into new geographic markets, therefore expanding their customer base and increasing the types of services offered. Putting voice and data on one network also eliminates the high maintenance costs of multiple networks. In addition, by tying VoIP into the SS7 network, service providers can offer enhanced features that aren't available in the circuit-switched environment.

The service creation model in next gen softswitches will allow providers to introduce new products and services more quickly. The ability to use application programming interfaces to bring in external applications will drive this growth as carriers expand to new markets. This next gen network will allow carriers to focus on call center types of applications, unified messaging, integrated access services, local features, voice virtual private networks and a host of traditional local and long-distance voice services.

Making the grade Some companies tout VoIP as a way to make inexpensive or "free" international calls. For some consumers and carriers, the quality of the phone call is not nearly as important as the cost. IP is a best-effort protocol, which is acceptable for data but not for voice. If there is a one- or two-second delay in transmitting data, the receiver will not likely notice the difference.

However, we've become accustomed to talking on the telephone without latency or jitter. When carriers make the network architecture decision to move to a VoIP environment, the quality must be indistinguishable from that of the public network. As a guideline, the new networks must have a mean opinion score in the range of 4.0, which is considered toll voice quality today. The mean opinion score is a measure of voice quality and uses the absolute category rating procedure to determine the general acceptability or quality of voice communication systems or products.

The debate on the maturity and reliability of multiprotocol label switching as a way to address quality of service (QOS) issues continues. For some service providers, the answer is to offer VoIP over an existing private ATM network. This area is changing rapidly, and QOS issues will be diminished with the next generation of routers.

In addition to offering quality equal to that of the public network, the network must be reliable and redundant to be considered truly carrier-class. This means the network must have five 9s reliability. Redundancy must be built in so that no packets are dropped. No elements of the network can fail - not the ATM core, the signaling gateways, trunking gateways, billing servers or the core operations support systems. Therefore, the need for disaster recovery systems will be critical in the initial engineering and design phases so calls can be re-routed correctly.

Perhaps the most crucial prerequisite of creating a global carrier-class network is scalability. In theory, many service providers would be able to map out a scalable network on paper, but the skill is in the implementation and coordination of these efforts. This includes ordering and provisioning, A-Links, feature group D long-distance access lines, intermachine trunks, real estate space and power and vendor equipment (including new releases and maintenance upgrades). Increasing bandwidth is a temporary approach at best and is not a sustainable business model. Adding more equipment into the network is also not the best answer, for when you add more equipment, you spend more on real estate, power, capital investments and resources to manage that added equipment.

Creating an architecture that can handle billions of minutes of use per month requires a solution with high call processing capabilities. Carrier-class scalability now needs to be addressed on a global basis and considered from a packet/transaction view rather than a circuit/minutes-of-use view. Scaling a global VoIP network affects the network core - including ATM switches or IP routers, along with trunking gateways and softswitches. For instance, softswitches will need to ride on higher-end servers that can support millions of minutes per server. These considerations also impact the back-office systems in terms of hardware that must be augmented to accept this call volume. A perfect example is the ability to process call detail records from next gen platforms and process them with new mediation layers for billing purposes.

Making the right choices Service providers must select the right trunking gateway and softswitch solutions to build a scalable, reliable and high-performance VoIP network. There are as many architecture and product alternatives on the market as there are vendors. Still, it is important to keep a few key criteria in mind when evaluating these technologies. First, next gen voice infrastructure solutions must adhere to the same standards as traditional public network central office (CO) switches.

On the gateway side, the hardware should comply with the rigorous NEBS standards for CO equipment and support full redundancy and hot-swap capability. It should support the high-density interfaces, such as T-1, T-3, OC-3c and OC-12c, needed by carriers to support high call volumes. The gateway must provide toll-quality voice, using next gen echo cancellation and voice compression algorithms to minimize latency. The trunking gateway should be designed to scale to the large configurations required by major carriers, supporting hundreds of thousands of calls or more.

From the softswitch perspective, carriers require a solution that provides basic telephony features while allowing them to create and deploy new, enhanced applications easily and quickly. The softswitch maintains network policy information on services, routing and provisioning and applies these policies to determine how calls should be handled. The softswitch supports features such as pre-subscribed customer calls, flexible routing, subscriber-specific features and more. In addition, the softswitch model provides an open approach through which new services can be created in "Internet time." The softswitch interacts with standard IP-based application servers so carriers can develop applications themselves or purchase third-party applications and implement them in a matter of weeks, not months or years.

Finally, the scalability and reliability of the softswitch is critical to the overall network scalability. Service providers should select a softswitch that delivers an extremely high call-processing performance, to support millions of busy hour call attempts (BHCAs). It's also critical that carriers consider not just BHCAs, but calls per second. BHCAs reflect a number equally distributed across a full hour, but phone calls are not equally distributed during that time. Carriers will see spikes in usage from second to second, and they must be prepared to meet those peak demands. BHCAs, while adding up to heavy traffic time, do not account for those seconds or minutes of true peak usage.

These criteria must be met to build scalable networks that support billions of minutes per month. The softswitch also must be fully redundant and provide failover times that are less than one second, so no call is dropped in the event of failure.

Global reach Next gen network providers deploying VoIP at carrier-grade must be concerned with having global reach. When building a global VoIP network, companies now must consider how cost of access systems are affected. Different currencies and their fluctuations, tax issues per operating units and exchange rates come into play when determining the effect on billing, customer care and order management, credit and collection systems. Global reach also will affect trouble ticketing, element management servers and performance test tools at the network operations centers (NOCs) and regional network points. Consistency in terms of standards and best practices from country to country also must be developed for identifying trunking ports, facility management and inventory management.

Decisions must be made as to whether legacy systems such as credit and collection systems, trouble ticketing and element management servers still are effective on a global basis. Perhaps they need to be augmented or even replaced to support voice services and then "converged services." Because we are moving toward a packet world, the systems must be applicable to both voice and data packets. When expanding the network geographically into greenfield areas, service providers must consider whether platforms currently in use will need to be scaled to support new customers and voice traffic.

As new distribution channels open, service level agreements will become increasingly important. Support tools for measuring network performance will be essential and will need to be provided on a real-time basis. The continued evolution of customer portals will be required to operate effectively and competitively in the marketplace. The need for carriers to provide real-time data on traffic statistics, port provisioning, invoice and billing information as well as doing their own moves, adds and changes will be critical. Carriers and retail customers want mission-critical data at their fingertips. That means customers must be able to access the information via an Internet portal, giving them the ability to manage their telecom needs right from their desktops.

The global footprint also requires service providers to build out global NOCs centered on providing regional support and host support. This will involve building training programs for field operations, engineers and support personnel. Each region will have different requirements in terms of providing Tier 1, Tier 2 and Tier 3 support.

Creating the seamless global connection To provide global voice services, carriers must meet certain key infrastructure requirements - such as the ability to offer a global signaling platform to support interconnection agreements with both PTTs and other local operators. Each country may have a slightly different signaling variant that will need to be developed and physical assets that will need to be deployed. The process for completing the testing within each country can be a time-consuming effort and must be coordinated carefully. Next gen suppliers must be experts in developing signaling variants. The ability to originate and terminate voice traffic anywhere in the world will be critical in providing a seamless connection.

Carriers also must address least cost routing, customer-specific routing and global routing. The information needs to be defined based on specific dialing plans; carrier traffic patterns, which may vary from country to country; flexible pricing and costing plans; real-time reporting; and manipulation of data to reflect market dynamics. Creating a global routing table and supporting infrastructure will prove to be a major challenge for carriers as they expand their networks worldwide.

Deciding to enter into the VoIP world seems to be a relatively easy choice to make. Mapping out a strategy and network architecture on paper takes time and careful planning. Creating an architecture robust enough to be considered carrier-class yet simple enough to manage effectively is the true test of building a global next gen network.