January 1999

Embedded CTI - The Best Of Both Worlds

BY BROUGH TURNER

I'm an evangelist for open telecommunications, that is, for leveraging the power and flexibility of open, mass-market computer platforms to solve problems in telecommunications. Among other things, I preach PCs and PC technology to telecom equipment vendors who have traditionally built systems using proprietary software running on special-purpose embedded computers. So what could I possibly mean by "embedded CTI"? There is no contradiction. Let me explain.

There are many advantages to PC technology, not the least of which is cost-effective processing power and first access to new technology. But the most important feature of mass-market computing platforms (PCs) is that they support the world's best software development environments. Software development is the bottleneck for most new product developments, even for products where the software content is not apparent to the buyer. But how do we apply the advantages of open telecommunications to systems at other scales - larger or smaller than a PC?

For large and/or highly available telecom systems, we have CompactPCI. Although very new, at least for telecommunications, CompactPCI has already acquired momentum. Indeed, acceptance of CompactPCI surpasses that of all prior industrial computer standards - VME for example. Why? Because CompactPCI is based on PC technology. While it provides rugged industrial packaging and the ability to hot-swap individual boards, CompactPCI's real advantage is that it inherits PC software and PC software development tools.

What about the other direction? What happens if you are trying to build a small, low-cost system with a target selling price that is less than the cost of a PC? Embedded systems offer a solution.

EMBEDDED SYSTEMS
As a result of the current furor over the "Year 2K" problem, even the general public is becoming aware that embedded microprocessors are used inside all kinds of equipment. Today, even a $100 consumer appliance is likely to be controlled by a small microprocessor. But typically, these embedded computer devices are highly specialized. And, their software is developed in a highly constrained environment.

There are embedded approaches, however, that can bring the advantages of open telecommunications systems to at least part of the low-end telecommunications market. The recent announcement by Microsoft of their release of Windows NT Embedded 4.0 highlights the approach. But the opportunity is much broader than just Windows NT. Today, we have component hardware (PC functions on a single chip), embedded versions of mainstream operating systems (UNIX and NT), flexible technology licensing, and support services that allow CTI solution developers to take advantage of the best of both of worlds - open and embedded.

WINDOWS NT AND OPEN TELECOMMUNICATIONS
If you take a look at the servers that MIS directors deploy, or talk to CTI developers creating open telecommunications systems for the enterprise, you find the majority using Windows NT. Windows NT has won in the enterprise. However, booting Windows NT not only takes time, but also requires a PC with hundreds of megabytes of disk storage, plus a display, a keyboard, and a mouse to run the user interaction. It is not well-suited for small-footprint, low-cost solutions. Windows NT Embedded 4.0, however, provides a solution to take open telecommunications solutions into embedded systems.

In September 1998, an early alpha version of Windows NT Embedded 4.0 was released for technical evaluation to a small group of companies, including Natural MicroSystems. A broad beta version is scheduled for the first quarter of 1999. While the software is still in beta, real product developments are possible today. The same open telecommunications tools that are available at the enterprise level, and up into the network/central office level, can now be applied to small systems. One must still provide the embedded PC, but a wide variety of embedded PC components are available. In addition, appropriate support services are available in the open telecommunications industry. If you can write telephony applications today using board-vendor APIs, or using TAPI 2.1, Wave APIs, and ActiveX controls under Windows NT on a PC, then you will also be able to develop an embedded solution.

EMBEDDED CTI SOLUTIONS - PAST, PRESENT, AND FUTURE
In the past, embedded CTI solutions were typically custom designs. Development was highly specialized and relatively inefficient because it did not take advantage of the software tools and infrastructure of mass-market PCs - in short, the Wintel platform.

Today, there are some CTI solutions using embedded Intel processors running versions of DOS in ROM, with various custom features added on. While these solutions are at least leveraging the Intel environment and leveraging over 15 years of DOS experience, DOS is rather dated. DOS doesn't support the advanced system programming features of UNIX or the flexible tools available with Windows.

For new designs, there is PC-on-a-chip technology and viable embedded versions of both UNIX and Windows NT. In the near future, we can expect embedded CTI solutions at low price points and in small form factors. These systems will leverage the full complement of UNIX or Windows NT features - and the advantages of open telecommunications such as reduced time to market, lower development costs, and flexibility. Innumerable configurations are possible, ranging from small systems with no disk, no keyboard, and no display up to solutions that include the full UNIX or Windows NT feature set, but on an embedded computer.

Let's look at an example of what this might mean for a specific solution, such as a unified messaging system. Today, using open telecommunications components, you could build a unified messaging system for an enterprise-based solution using a PC running Windows NT, taking advantage of its networking and Microsoft Exchange Server integration features. But such a system would cost more than a PC and take up at least the space of a PC - after all, it's based on a PC.

But what if you were to use embedded Windows NT? You could, for example, create a unified messaging solution on a single board that plugs into a proprietary PBX. There would be a one-time special engineering effort to spin the custom board, but even this could be subcontracted - to the CTI board vendor for example. The important point is the resulting board would run all the same software as the PC-based solution. Moreover, as the software evolved, new features could be downloaded to the board.

The resulting CTI system could be manufactured for far less cost than the full PC approach, even taking into account license fees to the operating system software vendor and to the telecom board vendor who provides the telecom interface and the voice, fax, and call processing capability. The operating system provides flexibility about where a networked open telecommunications system stores its files, so this single-board messaging system could be designed with or without an on-board disk drive.

Another example would be a four-port Internet telephony gateway for an enterprise WAN access box. Again, one solution would be to build this gateway using a Wintel PC. PC-based technology has led the Internet telephony gateway market so far. However, there are many existing WAN access routers with built-in proprietary form factors. Adding new capabilities to these boxes is typically tedious and expensive. Now it is reasonable to design a single plug-in board with the functionality of a PC-based 4-port Internet telephony gateway. With no display, no keyboard, no mouse, and no disk, the PC CPU and memory is no more expensive than other embedded CPUs. And, the circuitry of the PC-based Internet telephony board is already an extremely efficient DSP implementation. The combination has costs similar to a from-scratch design, but with the advantage of leveraging open telecommunications software.

CTI developers can now plan to add the functionality of an entire PC - in a form factor substantially less than a PC - to proprietary systems, for a fraction of the cost of a PC. The advantages of open telecommunications are available to vendors' previously proprietary solutions! Expect significant new capabilities in this market in the next two years.

A MORE DISTANT HORIZON
If you look back over the past 45 years of computer industry evolution, it's easy to guess that embedded Wintel is just the first step. Commercial mainframe computers were introduced in the early 1950s. They are still sold today. Their capabilities have grown enormously, and they continue to evolve. But minicomputers emerged in the mid-1960s, and, by the early 1970s, minicomputers had pulled ahead of mainframes, first in unit volumes and then in total dollars sold. Then, in the mid- to late-1970s, personal computers and workstations emerged. Minicomputers are still sold as high-end servers today, but they were superseded in the market by workstations and PCs in the mid-1980s.

You may notice a pattern here. Every couple of decades we have a new generation of programmable computer platform with costs one-tenth that of the preceding generation, and unit volumes a hundred times larger. We would be na�ve if we did not assume that PCs were going to be superseded by something, and soon. It is not totally clear yet what it will look like, but today's best bet is some kind of programmable, networked, IP appliance.

The key question for the open telecommunications industry is what will the new software operating and development environments look like? Will it be Windows-based, perhaps built on Microsoft's Windows CE operating system, Win 32 APIs, and TAPI 4 or 5, as Microsoft might like? Or will it be based on Internet protocols with Java/Beans/JINI, as Sun Microsystems might like? Perhaps it will be a newcomer - maybe the Palm OS!

The next five years will be very interesting for our industry. But right now, embedded UNIX and embedded Windows NT allow CTI developers to create some very interesting, very powerful solutions that leverage today's open telecommunications environment - and continue the trend in the computer industry toward smaller, lower-cost solutions.

OPEN TELECOMMUNICATIONS' BROAD REACH
The open telecommunications approach continues to extend its reach - taking solutions from the world of computers and using them to change the way people develop equipment and applications for the telecommunications industry. Advances, such as CompactPCI, allowed PC technology to move up from the enterprise into the central office with multi-thousand channel public network systems. Recent developments in embedded technology and embedded operating systems now allow open telecommunications to move down to the small end of the scale. By staying in sync with these advances, open telecommunications developers can take advantage of new technologies to take the industry in new directions - and make good money at the same time!

Brough Turner is senior vice president of technology at Natural MicroSystems, a leading provider of hardware and software technologies for developers of high-value telecommunications solutions. For more information, call Natural MicroSystems at 508-620-9300 or visit the company's Web site at www.nmss.com. E-mail to the author (rbt@nmss.com) is also welcome.