By John Rhea
SAN DIEGO — The acceptance and implementation of standards for commercial off-the-shelf (COTS) electronic components continued to make substantial progress in the past year, says Joe Chapman, the former Texas Instruments executive and now consultant based in Midland, Texas.
Chapman reported the latest findings of a two-year study conducted for Defense Standardization Program Office (DSPO) during the COTScon West 2000 conference Dec. 12-13 in San Diego.
Chapman ticked off three accomplishments and one partial accomplishment. Among the achievements were:
- the continuation of the integrated product team approach involving users and suppliers from the beginning of the design cycle;
- widespread recognition among military and commercial users that their product life cycles are different; and
- industry participation in global activities such as the Avionics Working Group.
He listed as a partial success the efforts of component manufacturers and original equipment manufacturers, or OEMs, to coordinate their efforts early in the definition phase of new programs.
Chapman stressed that COTS means different things to different people and that the concept should be understood as merely embracing catalog products available to all buyers regardless of their level of ruggedization. He also emphasized that COTS should be considered "the right part for the application."
Chapman also updated the status of the declining government and aerospace market for electronic components, a factor that has limited the aerospace industry's clout with the chipmakers.
Based on the latest figures from the Semiconductor Industry Association in San Jose, Calif., he noted that the market segment has fallen from about $1 billion to less than $400 million in the past year.
The resulting switch to commercial-grade parts upscreened to meet military specifications has been a matter of continuing concern for leaders of the DSPO. Still, Chapman reported that these parts have been operating safely in military and commercial aircraft in cockpit applications and in such flight-critical systems as engine controls.
For example, he noted that Boeing alone uses more than $50 million worth of commercial microprocessors a year. No single standard or specification exists for these parts at present, he said, but work is in progress within the Avionics Working Group sponsored by the aircraft makers and suppliers.
COTS electronic components are designed into virtually every weapon system, Chapman says, although upscreened parts appear to account for less than 10 percent of the total parts count.
The uprated parts operate slightly outside their commercial specifications for junction temperatures — 92 degrees Celsius vs. the 85-degree C commercial spec. — but the OEMs are qualifying the boxes and other subsystems and accordingly considering the components qualified by extension, he said.
"If COTS is the answer, what was the question?" Ed Hennessy, technical marketing director at Sky Computers Inc., Chelmsford, Mass., asked the COTScon audience. In his view successful COTS implementation requires planned technology insertion throughout the life cycle of a weapon system.
These extraordinarily long cycles — for example, more than 40 years for the SR-71 "Blackbird" reconnaissance aircraft — greatly exceed the constant upgrades of the electronic components. A typical digital signal processor (DSP) design is about 12 months, Hennessy noted.
The COTS savings are therefore principally realized at the front end of a system program, but Hennessy said that with planned technology insertion COTS could cut operations and maintenance costs by 40 percent.
Hennessy's proposed solution is to substitute several short design cycles for one long design cycle, and he listed goals of a four-fold improvement for the time required for design concept to field prototypes or to upgrade existing products.
Implicit in this approach is the use of an open-system architecture, and Hennessy urged, "You have to look under the hood." Stress the software, he concluded, and regard the hardware as "commodity-like."
Rodger Hosking, vice president at Pentek Inc. in Upper Saddle River, N.J., reported a successful application of COTS in the use of field-programmable gate arrays, or FPGAs, to replace DSPs in software-based radios. There has been an explosion in the wireless communications market, he noted, and this presents an opportunity to use cell phone technology based on COTS.
The real value of a software radio is what he called "information property," or IP, and the hardware implementation can be achieved with a mixture of application-specific integrated circuits (ASICs) and DSPs. The ASICs tend to dominate the computationally intensive functions such as filtering and demodulation while DSPs offer additional flexibility for analyses and decisions.
Where FPGAs enter the picture, in Hosking's scheme, is the bridging of applications such as decoding and fast Fourier transforms. The maintainability and portability of FPGA cores allow migration to future generations of the radios.
Just as Chapman and Hennessy urged continuing upgrades throughout the system life cycle, Hosking noted that these cores could migrate from generation to generation because of the IP libraries maintained by FPGA vendors and third party affiliates.
A last minute addition to the COTScon speakers lineup, Richard Jaenicke, director of product marketing at Mercury Computer Systems in Chelmsford, Mass., raised the issue of the software standards that will be necessary to implement COTS and made the distinction between open and mainstream standards — both of which will be needed for COTS-based systems, in his opinion.
Windows NT, for example, is not an open standard, but it is so widely used that it makes sense in many applications. Conversely, Defense Department-unique open standards, such as the controversial Ada programming language, have few users outside the military community and therefore little impetus for product improvement, Jaenicke said.
Now the Defense Department's Defense Advanced Research Projects Agency (DARPA) is launching a new standard for military users known as the Vector Signal and Image Processing Library, or VSIPL.
Another, Common Object Request Bridge Architecture (CORBA), is due to be available by mid-year for signal processing applications. This is also a DARPA-driven effort, and Jaenicke suggested it could become the standard for enterprise computing and transaction processing.
As in other applications of COTS, according to Jaenicke, the telecommunications industry is driving the standards for embedded computing. The defense market itself is too small to drive the standards, he noted.
