Design Article
Comment
hm
Model-driven development forestalls “11th hour” design problems
Darrell Teegarden, Business Unit Director, System Modeling & Analysis, Mentor Graphics Corp.
7/23/2010 4:17 AM EDT
The
development of a large-scale system or mil/aero platform, be it consumer
minivan or combat helicopter, is a long-term project fraught with complexity.
The penalties for a miscalculation often manifest as cost overruns, schedule
delays, reliability problems, and even dangerous product failures. Every
engineer and manager worries about risks like these and takes measures to avoid
them. But
even in a disciplined and rigorous program, critical issues may not become
apparent until the “11th hour” of a project’s life cycle. Moreover,
certain realities in today’s mil/aero development ecosystem are making it even
more challenging to achieve consistent quality and cost-effectiveness
throughout the development life cycle, particularly at the time of system
integration. As
systems become more complex, contractors who once specialized in narrow technical
areas are being forced to act as systems integrators and in turn, are contracting
subsystems to a global network of subcontractors. It adds up to significant
potential for misunderstandings, errors, and omissions due to communication
challenges across language and cultural boundaries. New
technology options add further complications and require diversified skills
from system developers. Designers must deal with embedded software, electronic
control units (ECUs), electro-mechanical subsystems (mechatronics), and the networks
through which they communicate. And of course, the complexity of the wiring
systems (harnesses) that connect all these elements has multiplied accordingly. Yet
another factor is compliance with DO-254 and other aviation standards and regulatory
mandates as well as modeling-related recommended practices in ARP4761. The
purpose of DO-254 is to ensure that airborne electronics perform their intended
function under all foreseeable conditions. DO-254 oversees design assurance for
in-flight hardware, and its full implementation will affect every level of
electronic hardware: system, LRU, board, and device. ARP4761 covers a different
but relevant discipline: modeling and its use in assessing the safety of an
aircraft or other system. While neither document explicitly increases the
complexity of the end product, they may complicate the design process. An
ideal solution would have teams of experts working independently throughout the
technical, system, and process levels, with designers concentrating on their
specific tasks using the best available tools. Ideas would be tested using
software modeling techniques and, once proven, would pass from step to step in
the process. Monitoring of compliance with regulatory standards would be
integral to every step. Ultimately all the pieces would be efficiently
integrated into a system that works. It is this vision that gave rise to
innovative model-driven development (MDD) tools and processes that are fast
gaining favor among system developers. MDD lays the groundwork for an
integrated design flow that addresses the complexity challenge once and for
all.
System integration challenges
increase in proportion to complexity
A typical aerospace platform is a complex hierarchy of
subsystems and components that includes multiple development programs. To
summarize these programs in general terms:
1. Dedicated
IC devices are developed and assembled with other components to create printed
circuit board-based functions.
2. Software
is integrated with these functions to create stand-alone electrical subsystems.
3. The
subsystems are assembled into complete electrical systems supporting unique
platform capabilities, such as power distribution, flight control, and braking.
4. Compatibility,
communication, and integrated operation are established between these complete systems.
5.
Finally, the systems are brought
together into a functioning, producible, maintainable, and highly reliable platform.
Most
programs begin with a concept development and refinement phase, after which the
system requirements and constraints are defined and reviewed. The system is then
functionally and physically partitioned into subsystems and components and the
requirements for the interfaces between these elements are allocated and
specified. With these definitions completed, engineers create a preliminary
design of each subsystem or component to validate the approach. Then they
proceed with a detailed design phase in which the engineers refine the
subsystem and component designs to the point where it appears they will meet
performance requirements and are specified sufficiently to be built.
At this
point, the system integration process begins. Subsystems and components must be
built and assembled into a system prototype; this is tested and debugged prior
to being demonstrated to the customer. Once the customer accepts the prototype,
a production readiness program is initiated and the system is released to
production, after which it can be operationally deployed.
Perhaps
system integration sounds like an elaborate and costly process. In fact, it is,
and it can be a bottleneck. Bringing subsystems and components together into a
single system is trouble-prone and unpredictable. Problems encountered at this
point in the program may require a subsystem or component redesign, or possibly
even refinement of the system requirements. As the yellow arrows in Figure
1 imply, a flaw detected late in process can send the project back to
“Square One.” Repeated iterations through the earlier stages of the design
cycle significantly delay a project.
Historically,
rigorous testing and verification have not taken place until the second half of
the product life cycle. Components and subsystems are first tested individually
and then integrated into a system which then undergoes architectural,
functional, and requirements verification. However, when system integration and verification is
postponed until subsystems are physically produced and ready for assembly, the
risk for costly rework and delays goes up dramatically.
Figure 1: A traditional
system development life cycle
Problems
found during systems integration are very expensive to fix because time, effort,
and money have already been committed to detailed design and production
tooling. Equally important, a project that is behind schedule and over budget can
damage credibility with the customer and impact long-term business
relationships.
Next: Title-1




hm
8/23/2010 5:32 PM EDT
I may like to know following:
- What is typical learning curve for this MDD?
- How much cost is involved for these tool modules and what additional man power is required? Please explain with example.
- Can this tool chain be integrated in new projects in gradual process? This is require to convince both management and engineers.
Sign in to Reply