Upload
s
View
212
Download
0
Embed Size (px)
Citation preview
164 Computer
T his issue marks the end of theIntegrated Engineering column,which first appeared in Com-puter in 1997. From the outset,we sought to provide a forum
for researchers and practitioners to sharethoughts and experiences on synergisticengineering of complex computer-basedsystems, which span a variety of appli-cation domains such as automobiles,avionics, and telecommunications. Wenow look back at what we accomplishedand reflect on what the future holds.
SYNERGISTIC ENGINEERING REVISITED
Our first article (“Toward SynergisticEngineering of Computer Systems,” Feb.1997, pp. 126-127) outlined the beliefthat integrated computer systems requirea mixture of general-purpose and appli-cation-specific hardware and software tosuccessfully perform their intended func-tion, while balancing several attributesof interest including performance, cost,safety, and availability.
Several common themes emerge fromthe design of these systems, particularlythe need to support
• complexity management, • cooperative hardware and software
development, and • process management.
In this context, process managementencompasses the multiperson coopera-tion, distribution, and coordination oftasks essential to system development.
Using these themes as a guideline, wesolicited a wide range of articles thatemphasized the synergistic activities nec-
essary to address various aspects of sys-tem development. We have tried to blendcoverage of state-of-the-art technologywith discussions of future research oppor-tunities, emphasizing where possible theapplication and benefit of techniques toreal-world problems and situations.
As a sample of our column’s scope, wehave included manuscripts representa-tive of work in theory-based, systems-level methods; hardware, networking,and configurable computing; and appli-cations. Specifically, these manuscriptsaddressed
• integrated hardware-and-softwaredevice technologies, including tools,that support microelectrical-mechan-ical systems and systems-on-chip;
• hardware-software codesign, in-cluding smart cards, networkedembedded systems, configurablecomputing, sensing networks, andcosimulation environments;
• software technologies such as mid-dleware that support adaptive qual-ity of service, reuse, and object-oriented approaches;
• fundamental methodologies includ-ing axiomatic design, model-inte-grated computing, and systems ecol-ogy; and
• applications such as avionics sys-tems, integrated computer-aidedmanufacturing solutions, and dis-tributed systems control.
These columns provide a testimony tothe creativity of engineers and scholarswho advance the field as new needs arise.
FUTURE DIRECTIONSAlthough it would be presumptuous
to predict what specific techniques andtechnologies will emerge to support the
synergistic engineering of computer sys-tems, we can outline issues that, howeverintractable, pose interesting research anddevelopment challenges.
Computer’s November 2001 edition,focusing on the engineering of computer-based systems, addresses some of thesechallenges with articles that describe
• model-based engineering for virtualprototyping of complex systemsthrough incremental evaluation ofdesign solutions;
• the need for an integrated architec-ture that synergistically combinesthe hardware, software, and com-munication parts of the systemunder design; and
• system and tool composition tech-nologies that form the basis forbuilding integrated-development en-vironments.
In our view, designers’ continued inabil-ity to gradually refine the systems-levelmodel into hardware-and-software real-izations transcends all these issues.Effectively solving the problem of modelcontinuity is a sine qua non of successfuldesign automation at a high level of sys-tem specification abstractions.
To accomplish high-level design auto-
Looking Back,Moving ForwardJerzy W. Rozenblit, The University of ArizonaSanjaya Kumar, Motorola
I N T E G R A T E D E N G I N E E R I N G
A synergistic approach todeveloping complex computer-based systems will continue to play an important role.
mation, which should provide faster timeto market and lower design cost, specifi-cation models must
• exhibit high levels of generality—besufficiently generic to represent ageneral model of computationexhibited by heterogeneous systems,including complex timing interde-pendencies;
• support representation at variousabstraction levels;
• foster modular and hierarchical con-struction; and
• support consistency and complete-ness checks throughout the range ofabstraction levels.
The availability of such models will ulti-mately lead to model compilation. Justas we can now compile software or hard-ware languages, we must develop tech-niques for compiling system specificationlanguages into integrated “executable”physical realizations.
Future integrated-specification modelsmust also support adaptability. Designsare now emerging, especially in embeddedsystems. Conceptually, model-based tech-niques offer great potential for designingsuch systems because they facilitate imple-mentation-independent specifications.
One desirable research goal is the abil-ity to map variable model structures ontoreconfigurable platforms while preserv-ing behavioral characteristics, includingtiming constraints that are typically ver-ifiable at the realization but not themodel-specification level. Extendingmodeling techniques to capture real—asopposed to simulation—time constraintspresents another challenge.
Along with developing integrative sys-tem model specifications, we need meth-ods for design-model testing andrequirements-compliance verification. Inplace of the largely ad hoc practices thatindustry currently uses, we must define aprocess that inputs system requirementscaptured in natural-language format andoutputs a minimal set of test scenariosthat cover all requirements.
To be comprehensive, the test casesmust cause at least one state transition—possibly generating a system response—associated with each uniquely identified
system requirement. Researchers haveachieved some progress in automatingthis process, but much work remainsbefore we can include temporal con-straints in both formal requirements spec-ification languages and testing methods.
ACKNOWLEDGMENTSThe contributions that appeared in
this column would not have been possi-ble without the support of many indi-viduals. We thank the contributors fortheir patience and cooperation during theeditorial process. For those whose con-tributions we could not, regretfully,include, we appreciate your participationand encourage you to pursue otheropportunities in Computer.
We also thank our readers for theirnumerous comments on various columns—we greatly valued your feedback andparticipation. In addition, our participa-tion as coeditors would not have beenpossible without the support of our col-leagues at the University of Arizona,Honeywell Laboratories, and Motorola.
Finally, we thank those at Computerwho worked with us to produce high-quality articles during the past five years.Editor in Chief Jim Aylor provided thespark that established the IntegratedEngineering column, while the staff’sstrong attention to detail and commit-ment to producing engaging copy provedintegral to the column’s success.
W e believe that a synergistic ap-proach to developing complexcomputer-based systems will con-
tinue to play an important role in a broadrange of future military and commercialsystems. Further, we expect that the Em-bedded Computing column will expandComputer’s coverage of integrated engi-neering topics to encompass new andexciting uses for the dynamic fusion ofhardware and software components. ✸
Jerzy W. Rozenblit is a professor of elec-trical and computing engineering at theUniversity of Arizona. Contact him [email protected].
Sanjaya Kumar is a principal staff engi-neer at Motorola. Contact him at [email protected].
December 2001 165
Let your e-mailaddress showyour professionalcommitment.
An IEEE Computer Society
e-mail alias forwards e-mail
to you, even if you change
companies or ISPs.
The e-mail address of computing professionals