REAL-TIME DIMENSIONING FOR IN VEHICLE EMBEDDED AND DISTRIBUTED SYSTEMS
November 5-6, 2008
Your educator
laurent george is doctor on real time systems. he is director of the research department in the famous graduate school of engineering in paris ece and director of the laboratory for analysis et control of complex systems. he works with car manufacturers and tiers 1, especially with valeo on research programmes dedicated to embedded systems.
You are concerned
• embedded software engineers,
• real time software engineers,
• embedded applications engineers,
• every person concerned by embedded systems engineering.
Training programme
TEMPORAL ROBUSTNESS OF REAL-TIME SYSTEMS BASED ON OSEK/VDX AND AUTOSAR
November 5, 2008
Objectives
this training programme will give you the algorithmic tools available for a correct dimensioning of real-time systems. a focus on osek/vdx will allow you to learn how to apply the results from the state of the art on real-time scheduling to osek/vdx and autosar with a case study and some practical exercises.
Real-time dimensioning - principles
Principles of the temporal robustness of a real-time system
• how to guarantee that a real-time system conforms to its specifications?
• how to identify the worst-case scenario for the respect of the timeliness constraints of a real-time system?
• how to establish feasibility conditions for the conception and the dimensioning of a real-time system?
• some key problems for the conception of a real-time system: how to set the priorities of the tasks? what is the optimal scheduling policies? how to defi ne the cpu load imposed of a real time system?
Real-time dimensioning of in vehicle centralised systems
Real-time system based on the osek/vdx kernel
• general principles and characteristics of osek/vdx
• feasibility conditions applied to osek/vdx
- study of a mixed scheduling system composed of basic preemptive and non preemptive tasks, it isr1 and it isr2
- identifying the different sources of overheads in osek/vdx
- taking into account the overheads of osek/vdx in the feasibility conditions
Case study:
- applying the feasibility conditions on a representative example (without overhead)
- infl uence of the overheads of osek/vdx on the feasibility conditions
- experimental result obtained on a real osek/vdx platform
How new results on the temporal robustness can be used in autosar
• review of a subset of the autosar operating system specification from a schedulability analysis point-of-view
• analysis of its adequacy with real-time scheduling theory
• how the real-time feasibility conditions of the state of the art can be used in autosar
• focus on the temporal robustness and the concept of protected operating system
TEMPORAL ROBUSTNESS OF IN VEHICLE DISTRIBUTED REAL-TIME SYSTEMS
November 6, 2008
Objectives:
The goal of this technical training day is to present the state of the art of existing approach to propose a correct real-time dimensioning of a distributed system in a vehicle. The networks studied are LIN, CAN and FlexRay.
We show for an application submitting periodic messages how it is possible to grant that any message can be sent and received in a bounded end-to-end delay. This delay also called the end-to-end response time of a message depends on the network considered and the scheduling policy used at the MAC level.
We consider the case of a single network and the case of several networks interconnected by gateways (CAN/CAN or CAN/FlexRay)
Real-time dimensioning of in vehicle distributed systems:
State of the art of existing approach to propose a correct real-time dimensioning of a
distributed system in a vehicle
- how to compute the worst case response time of a message on a LIN network.
- characterize on a CAN network the worst case response time of a message, function of its CAN ID with a non-preemptive scheduling analysis.
- how to assign with an optimal method the CAN message IDs.
- how to determine the worst case response time of a message sent in static and in the dynamic segment of a FlexRay network.
Case study:
Some practical exercises and case study are proposed that can help understanding how to implement new tools for the dimensioning of a distributed system.
