Schedule Integration for Time-Triggered Systems
Outline Motivation Automotive software Automotive architectures Integration Challenge Time-triggered automotive systems Sychronization Schedule Integration FlexRay ILP scheduling approach Case study & Scalability Conclusion 2
Example: Adaptive Cruise Control (AUDI AG, ACC) Florian Sagstetter 3
Distributed Control System Speed Adjustment Radar Actuator System Sensor Network Controller Network ACC ECU Florian Sagstetter 4
Delay Requirements Sensor t s Network m 1 Controller t c Network m 2 Actuator t a τ network τ network τ control Florian Sagstetter 5
Increasing Complexity in Automotive Electronics 400 350 300 250 240 200 175 150 100 90 100 100 50 0 1 3 5 13 50 Memory [MB] ECUs Sources: Paul Milbredt, AUDI AG, EFTA 2010 - Switched FlexRay: Increasing the Effective Bandwidth and Safety of FlexRay Networks BMW Group, FTF 2010 Orlando - Energy Saving Strategies in Future Automotive E/E Architectures Florian Sagstetter 6
Integration Challenge Application: Implementation: Mapping, Configuration and Schedule Integration: Defined by: Message Timing Deadline Florian Sagstetter 7
Rising Gap between Required and Actual System Complexity TODAY: INTEGRATION CHALLENGE? Source: The Software Car: Information and Communication Technology (ICT) as an Engine for the Electromobility of the Future ForTISS GmbH Florian Sagstetter 8
Paradigm Shift Event-triggered systems Time-triggered systems utilization flexibility (real-time properties need to be verified) Ethernet (PTP) predictability (real-time properties are guaranteed) Florian Sagstetter 9
Introduction: Asynchronous vs. Synchronous Asynchronous task 1 Jitter, lost/unused data bus x x task 2 Synchronous task 1 bus task 2 time Florian Sagstetter 10
Vision: Holistic Scheduling for Time Triggered Systems Ethernet PTP (Backbone) Gateway Task level Communication controller TT- Florian Sagstetter 11
Related Work: Global Scheduling Schedule Optimization of Time-Triggered Systems Communicating Over the FlexRay Static Segment H. Zeng, M. Di Natale, A. Ghosal, and A. Sangiovanni-Vincentelli. IEEE Transactions on Industrial Informatics, 7(1):1 17, 2011. Modular Scheduling of Distributed Heterogeneous Time-Triggered Automotive Systems. M. Lukasiewycz, R. Schneider, D. Goswami, and S. Chakraborty. In Proceedings of ASPDAC 2012, pages 665 670, 2012. Florian Sagstetter 12
Schedule Integration approach Cluster Schedules: Integration: Florian Sagstetter 13
Degrees of Freedom Schedules: Global Schedule: x x Florian Sagstetter 14
Introduction to FlexRay 3.0 Foundation of FlexRay Consortium (more than 100 members by 2005) Presentation: FlexRay Past Present Future Perfect - Christopher Temple Used in active damping system of BMW X5 Release of version 3.0, Disbandment of consortium 2000 2005 2006 2008 2010 future Release of version 2.1, the current industry standard Fully introduced with BMW 7 series ISO standard in preparation 2.1 Florian Sagstetter 15 BMW 7-series
FlexRay Protocol (Static Segment) CC 0 (Communication Cycle) CC 1 CC C Static Segment Dynamic Segment Symbol Window Network Idle Time Static Slot 1 Static Slot 2 Static Slot n Florian Sagstetter 16
Schedule Integration for FlexRay CC 0 CC 1 CC 2 CC 3 CC C Static Segment Static Segment Static Segment Static Segment x x+1 x+2 x x+1 x+2 x x+1 x+2 x x+1 x+2 Florian Sagstetter 17
ILP-based Schedule Integration Approach Only occupy static FlexRay segment Each static slot is only occupied once Search Space Reduction: Filter Offsets: Cycle Breaking: Florian Sagstetter 18
Case Study cluster ECUs frames repetition variance body 2 9 2,4 3-217 chassis 3 12 2,4 0-85 information 2 11 4,8 9-278 electric 3 10 1,2 0-55 total 10 42 1,2,4,8 0-278 Utilization of FlexRay bus: 29% Feasible schedule for FlexRay 2.1 in 163ms (Xeon 3.2GHz Quadcore,12GB RAM) ILP formulation of 5207 variables and 146926 constraints Feasible schedule for FlexRay 3.0 in 208ms ILP formulation of 14895 variables and 1070730 constraints Florian Sagstetter 19
Scalability Analysis with 5400 Synthetic Test Cases Florian Sagstetter 20
Conclusion Automotive functions benefit from a synchronous schedule Concurrent scheduling of ECUs and buses necessary Concurrent scheduling allows reduction of integration efforts ILP based scheduling Florian Sagstetter 21