October 2002

VoIP Application Software Takes A New Approach

BY JEFF CRITSER

The restructuring of the telecom sector has left many industry professionals pondering how the market will reshape itself, and what new technologies and trends will invigorate the resurgence. While the VoIP space has provided some measure of protection from the overall telecom tempest, the outcome of market forces is inescapable. Embracing and deploying VoIP technology for its own sake does not go far enough; cheaper, faster, better are the mantra of today, and these tenets are inextricable and can no longer be considered in isolation. Developers need to do more with less, with a clear eye on cost reduction and technology reuse without sacrificing product differentiation.

As customers continue to drive OEMs to produce feature-rich, high-density products under these cost constraints, the lifeblood of VoIP systems -- the applications software itself -- must be reengineered to accommodate this new paradigm. This entails far more than simple tweaks to the existing body of application software and reworking commercial grade algorithms. A different design philosophy altogether is required, one that looks at VoIP platforms as complete and interoperable entities and not a mere assembly of finite components. While on the surface this premise looks simple enough, this top-down approach represents a significant departure from the way VoIP software has been conceptualized, developed, and deployed in the field today.

THE TRADITIONAL BOTTOM-UP APPROACH
Historically, VoIP software focused mainly on the DSP, primarily due to the components' high representation in the design of VoIP platforms. Not surprisingly, OEMs centered their design decisions on which DSP they intended to use, with the standard considerations of performance, size, and power dissipation following suit. The VoIP software vendors responded in kind by supplying the necessary codecs and data packaging components necessary to run on the DSP, however this bottom-up approach left manufacturers to fend for themselves with the most critical design elements, including system management, signaling, call control, gateway control, and control plane interface. Often, the integration of these disparate components was quite a difficult process, requiring the stitching together of algorithms and protocols from many different suppliers. Consequently, system efficiency was sub-optimal, and time to market was painfully slow.

Legacy VoIP offerings also made it difficult for OEMs to incorporate their own intellectual property, such as a battle-hardened echo cancellation algorithm, and since the bulk of software was written in hand-optimized assembly, portability across a spectrum of available silicon was not attainable. Without any abstraction in the VoIP software separating it from the underlying hardware, incorporating next-generation DSPs into the VoIP platform resulted in a complete redesign of the system, further exacerbating time to market and fostering reluctance by the OEMs to embrace emerging silicon products.

THE NEW TOP-DOWN APPROACH
A top-down architectural approach to VoIP software, by comparison, utilizes a modular framework to anchor the functions needed in the gateway. Each embedded microprocessor employs its own software engine responsible for specific tasks in the system, i.e., signaling, packet processing, system management and control, etc. To give the entire framework elasticity and ensure that data transfer efficiency is not sacrificed, the individual software engines are linked together by a portable transport mechanism that can be overlaid onto Ethernet, cPCI, RapidIO, or proprietary interfaces. 

The media-handling algorithms themselves -- vocoders, echo cancellers, tone detection/generation, and the like -- are treated as modules that can be added or dropped from the framework as necessary. The framework recognizes the modules through an application programmable interface (API), which the OEMs would adopt to help implement their own repository of intellectual property. By architecting VoIP software in this fashion, the entire system maintains consistency and robustness and can accommodate evolving standards from ETSI, 3GPP, and the ITU with relative ease.

THE BOTTOM LINE
For a top-down architecture to be adopted and flourish among OEM developers, substantial benefits must be clearly articulated and understood by the VoIP community. Cost savings to the OEM customer, whether measured in faster time to market, higher system efficiency, or simplified supply management and support, would likely trump all other considerations when selecting VoIP software. 

Preservation Of Intellectual Property -- Most OEMs have amassed a significant cache of proprietary algorithms and protocols that have been honed and field-tested for many years. The larger OEMs in particular have staked their credibility and success on this intellectual property delivering the reliability and quality needed for their VoIP platform. Closed VoIP software solutions cannot incorporate the OEMs� intellectual property without completely redesigning the VoIP software from the ground up. An open framework solution, on the other hand, provides a straightforward path for integrating proprietary modules through the framework's internal API.

Protection From Obsolescence -- Standards are in a constant state of flux and make system specifications increasingly difficult to pin down. Hard-coded platforms run the risk of becoming obsolete as new standards emerge and establish de facto status in the industry. This presents enormous costs risk for system developers. In a top-down architecture, however, a significant measure of future proofing can be achieved by treating the vocoders and protocols as bolt-on modules to the framework. Emerging standards can be incorporated into the system through the API, provided that they conform to the specification.

Cross-Platform Reuse -- Since top-down software products are system-aware and modular by nature, the same basic management functions needed for media processing can be leveraged across a more diverse product range. The vocoders, signaling, and media protocols can be inserted as necessary to configure the platform for the intended application. At present, there is very little commonality among software offerings from the OEMs� perspective. For example, a voice gateway may require software from vendors A, B, and C, while a media server would require software from vendors X, Y, and Z. Consequently, supply managers and program directors are forced to manage a vast network of software vendors and cannot achieve any economies of scale in their pricing proposals.

Portability -- Unencumbered by legacy assembly code, new VoIP software employing a top-down architecture is written in C/C++, made possible by the crop of good compilers available in the market today. By optimizing in C and keeping the amount of assembly optimization to a minimum, the VoIP software can support a wide range of embedded processors. It's now possible for the OEMs to standardize on a common software platform and upgrade the underlying silicon as next-generation devices become available without redesigning the entire system.

Time To Market -- As VoIP continues down its evolutionary path, more competitors will enter the sector, carving out their respective value niches. Survival will depend upon on who can produce the most cost-effective and highest performing systems and get them to market fastest. OEMs have to spend an inordinate amount of time in system integration and not enough time developing key intellectual property and performance enhancements. Complete VoIP system solutions that can integrate the customers' proprietary modules help drive the development time down dramatically without compromising the OEMs' differentiation in the marketplace.

THE WAY FORWARD
New technologies and standards continue to arise that offer considerable performance benefits, however the limitations of the legacy VoIP software have hindered the ready adoption of these technologies. Compounding the issue is the paramount requirement to reduce costs, consolidate product lines, and increase efficiencies wherever possible. To help break this cycle and provide a meaningful solution for the VoIP market, the onus is upon the software vendors to rise to the challenge and establish a new paradigm in VoIP application design.

Complete and open system software solutions that view VoIP as an integrated and interoperable platform -- not simply the sum of its functional parts -- offer the most promise for the industry. The value derived from a top-down, framework-based architecture is tangible and can be measured in terms of faster development time, higher performance under diverse scenarios, and greater protection against obsolescence. More importantly, this new approach will give the OEMs what they need most: The opportunity to innovate. By lifting the burden of integrating VoIP components, OEMs can place their R&D resources on the forefront of the technology curve and help lead the industry to meet the demands of tomorrow.

Jeff Critser is president and chief executive officer of Trinity Convergence, Inc. Trinity Convergence designs, develops, and markets embedded software solutions for developers of converged network equipment. The company's focus is to develop and deploy packet voice and fax software for telecommunications OEMs and to enable the provision of packet-based communications to customers throughout the world. For more information, visit the company online at www.trinityconvergence.com.

[ Return To The October 2002 Table Of Contents ]