stefan buschmann tillsteinbach franzkorf · pdf filehochschule f r angewandte wissenschaften...

Simulation based Timing Analysis ofFlexRay Communication at System Level

Stefan Buschmann Till SteinbachFranz Korf Thomas C. Schmidt

Hamburg University of Applied [email protected]

{till.steinbach, korf, schmidt}@informatik.haw-hamburg.de

6th International Workshop on OMNeT++March 5th, 2013

RE NET

Hochschule für Angewandte Wissenschaften Hamburg Hamburg University of Applied Sciences

FlexRay-Simulation inOMNeT++

S. Buschmann

Introduction

Background &Requirements

Concept

Results & Evaluation

Conclusion & Outlook

REAgenda

1 Introduction

2 Background & Requirements

3 Concept

4 Results & Evaluation

5 Conclusion & Outlook

2 / 21



S. Buschmann

Introduction


Concept



REMotivation

Why simulate FlexRay on system level?State-of-the-art automotive fieldbusSimulation of complex networksImportant in automotive development

3 / 21



S. Buschmann

Introduction


Concept



REGoals

Configurable FlexRay simulationCompliant to FlexRay specificationCompatible with other simulation models

CAN, Real-time Ethernet, Ethernet-AVB

4 / 21



S. Buschmann

Introduction


Concept



REAgenda

1 Introduction


3 Concept



4 / 21



S. Buschmann

Introduction


Concept



REFlexRayTechnical features

Layer 1 and 2 in the OSI modelCommunication over two channels

Redundant transmissionDifferent data per channel

10 MBit/s per channelSynchronised time baseEvent- and time-triggered communication

5 / 21



S. Buschmann

Introduction


Concept



REFlexRayCommunication cycle

Node 1Node 2Node 3Node 4

Slot 1 2 3 4 5 61 2 3 4 5 6Slot 1 2 3 4 5 6 8 9 101 2 3 4 5 6 7 8

static segment

static segment

dynamic segment

dynamic segment

symbol window

network idle timeSegment

cycle 0 cycle 1

Time-triggered communicationEvent-triggered communication

6 / 21



S. Buschmann

Introduction


Concept



REFlexRaySynchronisation

Time measurement with synchronisation messages inthe static segmentCombination of two synchronisation methods

Offset correctionnode 1

node 2devi

atio

n

t

Rate correctionnode 1

node 2devi

atio

n

t

7 / 21



S. Buschmann

Introduction


Concept



RERequirementsFor the FlexRay model

Layer 2 in the OSI modelFlexRay functions

CommunicationSynchronisation

Implementation of a model of an oscillatorConfiguration of the network structure and theparametersSystem level error detection

8 / 21



S. Buschmann

Introduction


Concept



REAgenda

1 Introduction


3 Concept



8 / 21



S. Buschmann

Introduction


Concept



REConceptArchitecture

FlexRay nodes as modulesBus topology as module

9 / 21



S. Buschmann

Introduction


Concept



REConceptNode

Several submodulesConnection to the bus moduleIndependent configuration

10 / 21



S. Buschmann

Introduction


Concept



REConceptTopology

OMNeT++ only provides point to pointcommunicationBus topologyRealised as moduleProvides a maximum of two connections for eachnodeDistribution of incoming messages

busbus

unit1unit1

unit5unit5unit3unit3

unit2unit2

unit4unit4

channel A

channel B

11 / 21



S. Buschmann

Introduction


Concept



REConceptModel of the oscillator

Very accurate model would simulate every tickHuge amount of events

Our approach for the clock driftOnly one drift value per cycleReducing the number of events

t

Cycle Cycle

Constant Ticklength variable DriftfactorEvent Event

Cycle

Event

12 / 21



S. Buschmann

Introduction


Concept



REAgenda

1 Introduction


3 Concept



12 / 21



S. Buschmann

Introduction


Concept



REProtocol conformance & errordetection

Protocol conformance:Requirements are fulfilled

Typical error detection:Configuration problems

Too many sync nodesFrames in the same slot

Timing errorsFrames in wrong slot

13 / 21



S. Buschmann

Introduction


Concept



RELatency Analysis for the dyn. segmentSimulation parameter

4 nodes10 minislotsTransmission points distributed over dynamicsegmentDynamic frames require 1 to 3 minislots

1 2 3 411 12 13 14 15 16 17 18 19 20

dynamic segment

minislot ID

14 / 21



S. Buschmann

Introduction


Concept



RELatency Analysis for the dyn. segmentLatency of frames with different IDs

0 , 0 0 , 3 1 2 3 4 50

1 02 03 04 05 06 0

02468

1 0

3 , 2

3 , 3

3 , 2

3 , 3

Time f

or Bu

s Acce

ss [m

s]

S i m u l a t i o n T i m e [ s ]

I D 1 9

I D 1 7

I D 1 5

I D 1 1

15 / 21



S. Buschmann

Introduction


Concept



RELatency Analysis for the dyn. segmentLatency distribution

0 1 0 2 0 3 0 4 0 5 0 6 0 7 00123456789

9 51 0 0

Re

lative

Numb

er of

Packe

ts [%

]

T i m e f o r B u s A c c e s s [ m s ]

I D 1 1 I D 1 5 I D 1 7 I D 1 9

16 / 21



S. Buschmann

Introduction


Concept



REPerformance test

Several networks of various sizeOnly messages in the static segmentFurther parameters identical

number of nodes channels t_sim/t_real [s]10 single ∼0.9620 single ∼0.5830 single ∼0.4510 dual ∼0.6220 dual ∼0.32

Nearly worst case scenarioTiming parameter and configuration have a largeinfluence

17 / 21



S. Buschmann

Introduction


Concept



REEvaluation against CANoeComparison of latency results

CANoeCommercial network simulatorVariety of automobile communication protocols

Two equal networksThree nodesSame parameters

Same behaviour in both networksAmount and timing of messagesRepressed dynamic frames

Difference of approximatly 100 ns

18 / 21



S. Buschmann

Introduction


Concept



REAgenda

1 Introduction


3 Concept



18 / 21



S. Buschmann

Introduction


Concept



REConclusion

System level simulationSupport of different applicationsEvaluation against CANoeCan be used for simulation of completecommunication-matrices

19 / 21



S. Buschmann

Introduction


Concept



REOutlook

Extension of the simulationActive star topologyStartup procedure and node integration duringoperation

Gateway between FlexRay and other communicationmodelsSimulation of complex real communication-matrices

20 / 21



S. Buschmann

Introduction


Concept



REThank you!

Thank you for your attention!

Website of CoRE research group:http://www.haw-hamburg.de/core

21 / 21

stefan buschmann tillsteinbach franzkorf · pdf filehochschule f r angewandte wissenschaften...

Documents