Guest lecture: Scalable Design and Analysis of Mixed-critical Systems, Application to Automotive and Avionic architectures

Abdeldjalil Boudjadar, postdoctoral researcher and a project investigator at Linköping University, Sweden.

15.12.2016 | Jens Kargaard Madsen

Dato tir 03 jan
Tid 13:00 14:00
Sted Room 120, building 5125, Finlandsgade 22, 8200 Aarhus N

Abstract
The need to integrate large and complex functions into today's avionic and automotive electronic control systems (cyber-physical) requires complex and powerful computing platforms. Such systems are usually built by integrating different subsystems, potentially provided by different vendors, to enable the incremental Design and Certification (iD&C), recommended by the Integrated Modular Avionics (IMA) and automotive standard ISO26262. System components can have different criticality levels. To achieve fault containment and isolation between the different criticality levels, classical architectures rely on the physical separation between subsystems. Such an approach is reliable but leads to a poor utilization of the platform resources. Recent architectures, such as IMA and AUTOSAR, support functions related to different subsystems to share the same computing platform with an efficient use of the hardware. This makes the analysis task very challenging where non critical behavior interleaves with critical behavior where non-critical components/processes can be sacrificed to satisfy the critical ones. However, sacrificing non critical components may lead to a degradation of the quality of service.

In this talk, a model-based framework for the design and formal analysis of mixed critical systems will be presented. The framework is being developed in collaboration with industrial partners and targets avionic and automotive architectures. Our framework enables flexible scheduling of mixed critical subsystems while accounts for fault tolerance. To scale up our analysis process for industrial systems, we combine abstraction-based and compositional techniques. Our analysis technique is applicable for both qualitative (safety, timeliness, predictability) and quantitative (performance, degradation of the QoS) properties.

Abdeldjalil Boudjadar is a Postdoctoral researcher and a project investigator at Linköping University, Sweden. He obtained his PhD degree in Computer Science from Toulouse University France in December 2012. After graduating, he has been working at Aalborg University for 2 years and has visited Queen’s University Canada for 6 months. His research is mostly driven by industrial projects. His research interests include model-based design, formal specification and analysis of embedded real-time systems, and software architectures in particular for automotive and avionic systems. Further details are available on his homepage:  http://www.ida.liu.se/~abdbo42/

Forelæsning / foredrag