Download presentation
Presentation is loading. Please wait.
Published byDenzel Thurmond Modified over 9 years ago
2
Performance Analysis of FlexRay-based ECU Networks Andrei Hagiescu, Unmesh D. Bordoloi, Samarjit Chakraborty Department of Computer Science, National University of Singapore Prahladavaradan Sampath, P. Vignesh V. Ganesan, S. Ramesh General Motors R&D – India Science Laboratory, Bangalore
3
In a high-end car there are upto 70 ECUs exchanging upto 2500 signals. Commonly used protocols include CAN, LIN, J1850 etc. These can be broadly divided into event triggered and time triggered protocols – each has certain advantages and dis-advantages. Lot of emphasis on hybrid protocols – FlexRay. FlexRay is also backed by major automotive companies and hence, there has been a lot of interest in performance analysis of FlexRay-based designs. Performance Analysis of FlexRay-based ECU Networks
4
FlexRay-based ECU Networks Tasks have different activation rates and execution demands Each computation/communication element has a different scheduling/arbitration policy ECU FlexRay Bus ECU Comm. Controller Round Robin Fixed Priority EDF Input Events Output Events Timing Properties? End-to-end delay? Buffer requirements?
5
Abstract Models for Performance Analysis Processor Task Input Data/Events Service Model Event Model Concrete Instance Abstract Representation Processing Model
6
t [ms] events Arrival Pattern maximum/minimum number of events in any interval of length 2.5 ms 2.5 Arrival Curves [ l, u ] events uu [ms] ll 2.5 number of events in t=[0.. 2.5] ms slide window and record max and min l ( ) <= R(t+ ) – R(t) <= u ( ) Event Model – Modeling Execution Requirements t
7
Service Model – Modeling Resource Availability t [ms] availability Resource Availability maximum/minimum available service in any interval of length 2.5 ms available service in t=[0.. 2.5] ms 2.5 Service Curves [ l, u ] service uu [ms] ll 2.5 l ( ) <= S(t+ ) – S(t) <= u ( ) t
8
remaining supply processed events Processing Model Service Model Event Model Process ing Model
9
PE 1 PE 2 Modeling dependency Compositional Schedulability/Timing Analysis Compositional Analysis
10
Brake Control TDMA Object Detection Object Detection Radar 1Radar 2 ECU2 Data Fusion Object Selection Adaptive Cruise Control Throttle and Brake Torque Arbitration Actuators Path Estimator Throttle Control Wheel Sensor ECU 4 Anti-lock Braking System Sensor (to crash control subsystem) 11 22 33 TfTf BfBf ’’ FlexRay Bus Task dependencies DYN message ST message ECU1 Fixed Priority ECU3 Fixed Priority m1m1 m2m2 m3m3 m4m4 m5m5 m6m6 m7m7 An adaptive cruise control application Case Study – An ACC Application
11
Performance Analysis Minimum end-to-end delay when DYN segment length = 9ms and ST segment length = 5 ms Variations in the end-to-end delay with different sampling periods (ST= 8 ms and DYN = 6 ms) Delay (ms) ST Length (ms) DYN Length (ms) 9 8 7 6 5 5 7 9 115 116 117 118 119 120 121 122 123 124 125 Delay (ms) Radar Period (ms) Wheel Sensor Period (ms) 25 30 35 40 60 55 50 45 120 125 130 135 140 145 150 bus cycle length = 14 ms
12
Thank You!
Similar presentations
© 2025 SlidePlayer.com. Inc.
All rights reserved.