Unified communications

Thumbs up for unified communications

Unified communications can offer enterprises a number of benefits. But the market adoption of the technology is slow. A look at the business and technology benefits of this form of communication. by Larry Velez

Historically, companies have expected significant employee productivity improvements by offering employees a unified message box to access e-mail, voice mail, SMS, teletex, and fax from multiple points of interaction (POIs) The POIs are fixed phones, mobile or cellular phones, PCs, and PDAs.

Market adoption rates were projected to be billions of US dollars, yet the total market was expected to reach only $500 million at the end of 2003 (approximately), and we believe the market will not exceed $1 billion by 2007.

The market size approximations (for 2003-07) can be divided into three segments: enterprise sales ($100M-$200M), service provider products ($100M-$200M), and service provider services ($300M-$500M). Across these segments, a transformation in the technology is taking place that promises to fill the value gap that currently exists in systems.

This transformation is the evolution of Unified Messaging (UM) into Unified Communications (UC). The new features in UC are making the investment business case more compelling, but early UC experiences are reminiscent of the same buyer patterns that convinced companies to purchase UM systems. These benefits are focused on the sales and field service force, and therefore are shared by less than five percent of company personnel.

The migration to IP and the insourcing of bridging services will expand in 2003 as enterprises realize cost savings. A renewed interest in Web-based rich media will lead to a marginal increase in the deployment of enterprise content delivery networks to minimize network impacts in 2003, accelerating in 2004.

To understand unified communications, it is appropriate to review the basics of proprietary unified communications infrastructure and the complexities of scaling it out to encompass new features and functions, like adding automatic speech processing.

This will establish the foundation for discussing how UC solves scaling challenges by endorsing industry IT standards, and introduces how the UC architecture is extended to support real-time communications, including call screening and notification, presence management, multi-party Web and audio conferencing, instant messaging, automated assistant, and calendar and rule-based routing.

Unified Messaging 101

UM enables company employees to access audio or electronic messages from a single system. The system is accessible from different POIs. The most common POIs currently in use are fixed telephones, mobile or cellular phones, and the PC. In layman terms, a customer sends or leaves a message for a company employee, and the message is 'stored' in a single or virtual message box.

Depending on the implementation, the messages are either stored in a single physical location or dynamically aggregated just before they are forwarded to the employee via e-mail. The employee can retrieve his or her personal messages and either listen to voice mail messages or read text.

Advanced systems can also convert text messages into speech, enabling the employee to listen to the text message. Moreover, the employee may be able to reply to a message by speaking his message into the system; the system then converts the message into text. The text is then sent to the customer as an e-mail or another text message type. These advanced system features are provided via speech synthesis i.e., Speech To Text (STT), Text To Speech (TTS), and Automatic Speech Recogni-tion (ASR).

Impractical for Many, Useful for a Few

The advanced system features raise some practical complications due to technical and human limitations:

• TTS translation may be incomprehensible due to abbreviations, grammatical errors, jargon, language, and colloquialisms.
• STT has imperfections because it creates text with limited punctuation. In addition, very few people are able to say exactly what they mean in one take, as a medical doctor would dictate patient diagnosis.

These complications limit UM system usefulness to business environments where a community of employees has the following characteristics:

• Shares common financial objectives
• Follows a long, time-sensitive, and labor-intensive process
• Works in teams (e.g., territory, industry, customer type)
• Uses other corporate resources (e.g., finance, engineering) as efficiently as possible

This business profile would use advanced UM features to communicate with the team and customers in short bursts (messages), which incrementally moves the business process forward. The greater the multiples of these tactical bursts of communication, the greater the opportunity for increased corporate efficiency.

For example, the reduction of a sales cycle could reduce inventory shelf life, allowing greater attention to be placed on improving other elements of the internal sales to fulfillment supply chain.

Technical View

The UM system can be technically classified as a virtual store-and-forward collaboration infrastructure. The system is virtual because it is an integration of two store-and-forward systems: e-mail and voice mail. Moreover, the UM system leverages the store-and-forward architecture and applies it to other messaging types, like fax, teletext, and SMS.

The most basic UM feature will store and forward faxes as an image attachment to an e-mail. Similarly, voice mail messages can be stored as .wav or other files and forwarded as an e-mail attachment that can be played on a standard PC media player.

While these services could be useful for the mobile road warrior, downloading or replicating e-mail containing voice mail attachments would significantly increase the size of the download, lengthen the time it takes to replicate, and increase remote dial-up access costs.

Alternatively, the mail system could be hosted, but this requires the employee to be online. Another approach is to send a voice mail URL in an e-mail, which the employee can hear, when he or she goes online.

The advanced UM system integrates speech synthesis, and automatic speech recognition. In this case, speech synthesis is used to convert text messages into speech and voice mail into text. ASR is used as the command-and-control interface that helps the employee navigate though the system, offering basic commands such as read, delete, pause, purge, and write.

To write a message, the message is first stored as a voice mail. It then passes through the STT engine and finally back through the TTS engine to read back the 'spoken' text message.

Technical complications

In our experience, short messages are spoken back with acceptable accuracy. The impracticality occurs, when a long time is spent to determine whether the message is acceptable. In addition, the IT Organization (ITO) should note that there are often several technical complications:

• An increase in use of database and storage resources, and consequently costs: Each voice mail converts into an e-mail, and potentially e-mails are stored as voice mail. The ITO should assess this carefully and seek systems that reduce consumption of these resources.
• High cost of acquiring speech engine licenses for the sole use of the UM system: The ITO should instead seek to use the speech engines as a corporate resource of speech processing power for multiple purposes.
• 'N x integration': The greater the number of POIs and messaging systems that must be integrated, the greater the cost of ownership to keep the virtual UM functioning correctly. This is especially the case when enterprises want to integrate with mobile operator voice mail systems, where the access price is significant.
• Integrating speech control, with existing voice mail and e-mail: There are no open standards to protect the integrity and longevity of the integration work. This implies that product changes to the store-and-forward infrastructure could render previous integration work unusable, necessitating costly reintegration.
• Small degrees of inaccuracy in the automatic speech recognition and speech synthesis engines: These inaccuracies can lead to a cumbersome and frustrating experience for the user.
• Immature programming and configuration interfaces for the virtual UM system: Changes to the system that could be done by a skilled administrator are limited. We often find that the UM vendor is hired to implement changes at high costs.

Despite these impracticalities and technical challenges, numerous companies still value the UM investment because it has led to the collapse of the total time needed to meet a business objective (e.g., obtaining product discount approval), even if the system is not speech-enabled.

Yet market adoption of UM is stagnant, and vendors seek to put a new face on the technology to solve technical challenges, create new value, and stimulate adoption. The research and development investments to increase market adoption have spawned UC.

Technical View

Technically, UC is a combination of the UM virtual store-and-forward infrastructure and real-time communications infrastructure. The use of open industry standards and standard application server platforms provide the foundation to deliver on the UC feature promise.

The key standard ingredients for UC are as follows:

• Session Initiation Protocol (SIP): This protocol establishes communications between two IP end points (e.g., IP telephone, IP soft phone) and provides presence information to be managed by user set or company rules. The vendor migrations from UM to UC use SIP as the fundamental standard, implying that a converged local-area network (LAN) is a pre-requisite for using SIP.
• Lightweight Directory Access Protocol (LDAP): This protocol is used to authenticate users on the system and determine their privileges for communications services. All user details from other systems could be accessed via this directory structure.
• Extensible Markup Language (XML): This protocol is used to embed UC features in corporate applications. For example, clicking on a client's name from a CRM application can prompt various message types that could be sent to the client.
• Voice XML (VXML): This standard protocol is used to build speech-enabled applications. These applications are short scripts for how to treat incoming calls. The scripts can be as simple as playing a greeting or as complex as setting up a dialog with the calling party to arrange an appointment on the employee's calendar.
• J2EE or .Net application development framework: This is the general framework upon which the UC features and functions are built. The vendor community wholeheartedly believes that UC features must minimize the use of proprietary components, and use of these application server frameworks represents the initial forays of vendors building software-only applications.

UC systems provide the ITO with certain advantages. The speech engine can be shared by other systems and is not locked into one system. And an industry standards-based interface will insulate integration investments from application changes.

Yet UC does little for increased database and storage requirements, though VoIP stored messages consume less space than traditional voice mail systems.

Summary

UC is a necessary evolution of UM. The evolution makes the integration of multiple store-and-forward systems more robust. The robustness is the result of several years of experience building complex single message box systems and significantly reduces integration and operational complexities.

The use of industry open standards and standard infrastructure increases confidence in interoperability with other IT systems. The UC system also addresses the value question behind the unified message box by providing real-time communications features. These features promise to improve individual and team productivity.

Early business-case successes indicate that this can be valuable for specific departments within a company. We expect the vendor community to increase market spending to discover where the benefits of unified communications can be applied to other departments within a company. We believe the infrastructure agility of the UC system will assist in the search for new users, and the market will conservatively grow at a compound annual growth rate of 19 percent between 2003 and 2007.

Larry Velez is program director, Infrastructure, META Group. E-mail him at Larry.Velez@metagroup.com. www.metagroup.com This article first appeared in Network Computing Asia