140 likes | 812 Views
Federal Aviation Administration. Complex Integrated Avionic Systems and System Safety. Presentation to: Europe/U.S. International Aviation Safety Conference Name: Ali Bahrami Date: June 9, 2005. Electronic flight inst. Ex. 757/767. Integrated display system
E N D
Federal Aviation Administration Complex Integrated Avionic Systems and System Safety Presentation to: Europe/U.S. International Aviation Safety ConferenceName: Ali Bahrami Date: June 9, 2005
Electronic flight inst. Ex. 757/767 Integrated display system Ex. 747-400 Expanded IMA Ex. Falcon EASy, ERJ-170 Integrated Mod Avionics (IMA) Ex. 777 • Integration within closely related functions • Most functionality in hardware/firmware • Integration of most display-related avionics functions • Most functionality re-programmable • Integration of avionics + some flt. control and airplane systems • More generic processors & software-based functionality Trends in Avionics: Integration and Complexity 2000 1980 1990 • Integration of many avionics functions • Card-based processors in cabinet racks
Trends in Avionics: Architectures • Huge increases in: • Functional integration. • Software size and complexity. • Shift in techniques for isolation/independence: • Traditionally, redundant features were completely isolated – now they communicate with each other. • High/low criticality functions traditionally physically isolated from each other – now share computing and databus resources. • Mix of new and reused (“legacy”) software.
Trends in Avionics: TSO • TSOs: • Traditionally, TSOs were used for simple equipment (e.g. seat belts) and well-defined “stand-alone” functions (e.g. air speed indicator). Installation issues were minimal. • Now, TSO requirements cover only a small fraction of the designed functionality. • TSO functionality may be embedded in an integrated avionics suite (“functional TSO”). • Vendors need TSOA to ship “brain-dead” hardware which doesn’t comply with the full TSO requirements until installed and software is loaded.
Trends in Avionics: Engineering and Business Practices • Increasing dependence on Commercial Off-the-Shelf (COTS) hardware and software. Examples: • Microprocessors (from PC industry). • Operating systems (e.g. Windows). • Graphic processors (from video game industry). • Changes in manufacturer-vendor relationships and responsibilities. • Global design and manufacturing of highly integrated avionics functions. • Shift from airframe manufacturer as “designer/builder” to “integrator/assembler.”
Certification Challenges • Integration and complexity: • Current processes (e.g. DO-178B/ED-12B for software) were developed with much simpler architectures in mind. • Experience is showing that there are complex and often unexpected “connections” between traditionally unrelated or independent functions, especially during failures. • Failures become more difficult to predict and diagnose. • It becomes less and less feasible to test all inter-related failure modes. • Fully integrated test facilities become more challenging and expensive to build and operate.
Certification Challenges • Software: • Software-based isolation and independence is much more “fluid” and difficult to assure than relying on hardware. • Mixing of COTS, reused, and new software – all developed by different processes and to different standards – makes assessing the safety issues much more difficult, especially in standardized ways.
Certification Challenges • “Functional” TSO: • Difficult to separate TSO issues from installation issues • TSO’d function may be part of the software that resides on a circuit card. • TSO compliance can only be assessed when installed in the host system. • Even simple issues like part marking become complicated. • TSO change processes were not developed with these complex TSO “packages” in mind. • Engineering and Business practices: • COTS products are not developed to traditional aviation standards. • Detailed certification data and knowledge often resides at vendor rather than manufacturer.
How the Authorities Have Responded • The authorities have already taken a number of actions to support recent IMA trends and specific projects, including: • Development of IMA AC and TSO. • Development of an Order on software reuse. • Approval of functional TSOs. • Numerous DO-178B/ED-12B “workarounds.” • Additional relevant guidance is in work. • However, continued industry support is needed…
What is Needed to Support the Trend? • Current software certification methods did not envision modern IMA architectures, so we need new methods… • That are equally effective in ensuring safety… • While supporting the certification of IMA. • The current TSO process is not well-suited for embedded software functions, so we need new approaches to TSOA… • Which allow design and production approval for traditional TSO functions in IMA architectures… • While protecting the level of safety provided by type certification processes.
What is Needed to Support the Trend? • When manufacturers out-source development and test: • New processes for authorities/manufacturer/vendor communication are needed. • Testing: • Testing of the IMA “pieces” will not find integration problems. • The actual airplane is not an adequate test environment for many IMA issues. • Full-scale integration test facilities may not be commercially viable. • Industry needs to help develop new approaches to integration testing that will find and characterize IMA problems before certification.
Authority-Industry Partnership • Cooperation is needed more than ever. • Traditional certification processes were developed to match past commercial practices • The pace of change is increasing • Industry will need to lead the effort to develop new methods of compliance. • New methods cannot just “do less” – they MUST preserve, and where possible, improve the level of safety. • Focus on safety-related issues while with IMA, it is more difficult to separate what is or is not “safety-related.”
Summary and Future Perspectives • The authorities support industry’s efforts to advance the technology • Historic cooperation between the authorities and industry has been essential in developing viable and effective methods of compliance and safety assurance. • Cooperation is even more critical as we collectively support rapid technological advances while at the same time increase the level of safety. • Potential broader issue: Does the overall safety assessment process need to be revisited, to account for the migration of functionality (and failure conditions) from hardware to software?