Predictions For 2002 Special Feature

VoIP Addresses Availability, Cost, And Quality Of Service

By Bob Massad

We believe this will be true provided end users are confident that high service quality is achievable and sustainable, and that the instances of good and bad quality calls -- as perceived by the end users -- are noted and accounted for in an objective way. These issues have been generally accepted as the key factors that have limited the growth of VoIP in the past.

Several events and technologies are converging to enable this VoIP explosion. As is usually the case, some of those factors are price related, but we believe that the most significant factors do not have much to do with price at all. We believe that while price isn't everything, a balanced, reliable, useful, and high-quality solution is.

Falling Costs Of IP Phones
The first factor to mention is an expected enormous drop in the price of IP phones. Not more than two years ago, IP phones took a big bite out of a customer's wallet, on the order of $500. In 2002, due to volume increases, improvements in semiconductor and DSP technology, improvements in soft phone functionality, and significant competition by independent IP phone vendors against the captive system suppliers, we expect to see IP phone prices trending towards $100. In addition to breaking a key psychological price barrier, the drop will make IP phones immensely competitive with the traditional PSTN desktop digital phone. The kicker for IP telephony will be the ability to support a much richer application set and a richer user interface with the IP phone than is possible with the traditional desktop phone. An example is the killer application for IP telephony: unified messaging.

Emergence Of Bridging Apps
Another factor that will spur VoIP growth will be the broad emergence of low cost "bridging" applications that allow for the effective VoIP-enabling of standard PSTN digital telephones. It is generally considered a good thing to allow users to adopt new technologies at their own speed, rather than requiring "forklift" upgrades. To enable smooth upgrades, vendors will be providing low cost migration tools that allow the continued use of traditional PSTN telephones on the desktop, and transparently provide the IP packetization of their digital signals for transmission over an IP infrastructure. This in turn will allow traditional PSTN desktop telephones to access many of the advanced and value-added features of the next-generation network service providers, such as Web sites' "click to call" buttons.

Windows XP Spurs VoIP Growth
A major factor in VoIP's growth will be the emergence of the Microsoft's Windows XP operating system. Microsoft's operating system ubiquity will drive use among the masses and get attention from waves of application developers. Microsoft is making significant moves into the communications space with its messenger functionality. Both the messenger client -- featuring call quality enhancing functions such improved forward error correction, context-sensitive codec selection, and echo cancellation technology -- and the inclusion of a Session Initiation Protocol (SIP) protocol stack are the key VoIP enablers.

Adoption Of SIP
SIP could help drive the industry forward all on its own, as it greatly simplifies the development of VoIP applications due to its simpler textual encoding methods and modular approach, as compared to the complexities involved with monolithic H.323 implementations where the encoding is binary. SIP is modeled after HTTP, and we all know what HTTP did for the World Wide Web. It's also noteworthy that SIP seems to natively support instant messaging applications, whereas H.323 does not, hence SIP's inclusion in Microsoft's XP. Aligning VoIP applications like click-to-talk with the current pervasive use of instant messaging should prove hard to resist. XP will get SIP deployed everywhere, and SIP-based applications will be sure to follow.

Greater Control Over Service Levels
In concert with an expected drop in IP phone prices and the ubiquity guaranteed by Windows XP, there will be changes in both VoIP end system management and device designs, as well as a long-desired change in the focus on data management.

Network operations managers and end users alike -- concerned about performance and quality of service (QoS) -- will have access to low-cost embedded, objective call rating technology that will allow them to monitor and grade the quality of every call they either transport or participate in, in real time.

The availability of this technology will change the internals of VoIP end systems as they strive to accommodate these non-intrusive agents. Mirroring the LAN switch rush to embedded RMON agents in the early 90s, VoIP gateways and media servers will begin to implement conceptually similar agents; but CPU, memory, and cost limitations will force these new voice monitoring agents to be more streamlined and more computationally efficient than the compute intensive RMON.

Similarly, high-end IP phones will also take on metering (and possibly active agent) functions as service providers, captive IT groups and third party providers are held to high standards of service delivery and objective verification. To maximize use of available processing power and to differentiate them from the pack, DSP providers will follow suit and will be embedding the compute intensive portions of these agents into their DSPs.

Quality Levels Focus On The End User Experience
A very interesting thing about this new agent technology is that they will be specifically engineered to measure and report on the end user experience, instead of the typical laundry list of uncorrelated counts, rates, and stats of every kind so prevalent in the SNMP world. Voice, video, and similar applications require a new qualitative metric that expresses the end user experience in a simple and clear way. Applications such as voice and video are more sense-organ oriented, more qualitative or subjective in nature than traditional applications such as file transfer and e-mail. Therefore, a new paradigm is needed to assess quality.

The technology will correlate natural IP network-originated impairments such as delay, jitter, and -- most importantly -- burst packet loss, and provide a single quality score, much like we got when we were in school. Moreover, the agents will need to model how humans recognize transitions in quality, and how they remember events. The agents themselves will be based upon sophisticated statistical modeling techniques that require a minimal amount of host system resources, while providing a very accurate assessment of the end user's opinion of call quality.

Consumers and providers alike will use this technology to verify Service Level Agreements (SLAs) and determine which calls are "billable" and which are not. The same information will be used by network operations staff to alert them of network problems;  marketing departments will also use them to understand the opinions of their customer base and designing messages and programs around such insights.

Further down the line, the call quality monitoring technology will extend to active QoS control in the form of a closed-loop feedback system that has the ability to improve the quality of calls in process, in real time, by interacting with the infrastructure. If one assumes that there is a lag time between the instantaneous occurrence of a network impairment and the end user's recognition of that impairment, and our research shows that lag time to be on the order of five seconds, then it becomes possible for technologies operating on a millisecond scale to moderate or rectify those impairments before they are recognized (or before they become annoying enough to spur a premature termination of the call).

These last two technologies are going to lead to higher usage rates and longer call durations for VoIP. That's a win for both service provider and service user. It costs each party when quality is poor. Real-time QoS adjustments will also allow providers to "right-size" their networks as they uncover what network infrastructure configurations provide acceptable quality at given levels of capacity.

Conclusion
2002 will see VoIP-enabling technology become generally affordable and ubiquitously available. The coming year will also see the most significant obstacle to VoIP use -- quality of service -- come under control. The jumping of this last hurdle -- long the bane of voice of data deployments -- ensures VoIP's widespread adoption.

Bob Massad is vice president, marketing for Telchemy. Telchemy develops intelligent network Quality of Service (QoS) management applications that help to provide high quality service delivery over packetized networks such as the Internet. The company�s initial focus is the Voice over IP (VoIP) market where service providers and major corporations are concerned about potential QoS issues resulting from packet loss, latency, and jitter. Telchemy�s adaptive management applications extend existing standards, implementing advanced statistical models that relate network performance to end-user perceived voice quality.