The FlexRay Protocol Peter Bhm 27.9.05 Overview 1. Introduction - - PowerPoint PPT Presentation

The FlexRay Protocol

Peter Böhm 27.9.05

Peter Böhm 27.09.05

Overview

• 1. Introduction
• 2. Network Topology
• 3. Nodes
• 4. Communication Controller
• 5. Schedule
• 6. Message Processing
• 7. Clock Synchronization
• 8. Wake-up/Start-up
• 9. Summary

2

Peter Böhm 27.09.05

• 1. Introduction
• FlexRay: Communication in distributed systems within

automotive context

• developed by the FlexRay consortium (BMW, DaimlerChrysler,

Motorola, Philips) founded in 1999

• since 1999 many well-known companies joined (e.g. Bosch, GM,

VW, Mazda, etc.)

• aim: flexible, fault-tolerant communication protocol

3

Peter Böhm 27.09.05

• 2. Network Topology
• star, bus topology or

combination

• max. 2 channels
• optional bus guardians

➡ various, flexible network topologies

4 Node 1 Node 5 Node 4 Node 3 Node 2 Channel A Channel B Node 1 Node 5 Node 4 Node 3 Node 2

Star A Star B

2 typical network topologies:

Peter Böhm 27.09.05

• 3. Nodes
• main interest:

communication controller (CC)

• CC’s task:
• interface to host
• message processing
• transmission
• reception
• clock synchronization

5

HOST COMMUNICATION CONTROLLER Bus Guardian Bus Guardian

Peter Böhm 27.09.05

• 4. Communication Controller

6

controller host interface host protocol operation control media access control (1 per channel) frame and symbol processing (1 per channel) coding/decoding processes (1 per channel) from channel interface to channel interface clock synchronization

Peter Böhm 27.09.05

• 4. Communication Controller
• Controller Host Interface:
• interface between host and controller
• control command interface
• message interface
• handles configuration and status data
• message buffers for reception and transmission
• Protocol Operation Control
• purpose: react to host commands and protocol conditions
• change operation modes of core processes
• Clock Synchronization
• 3 parts: macrotick generation, clock synchronization and clock synchronization

startup

• macrotick: smallest synchronized time unit

7

Peter Böhm 27.09.05

• 4. Communication Controller
• Media Access Control (Transmission)
• schedules the bus write accesses
• assembles message header
• Frame and Symbol Processing (Reception)
• handles received messages
• performs timing and error checks; e.g. syntax tests, etc.
• Coding/Decoding Processes (Read/Write)
• encodes frames for transmission, i.e. each bit 8 times on bus
• decodes received frames
• appends CRC for transmission
• CRC check on received frames

8

Peter Böhm 27.09.05

• 5. Schedule
• time-triggered
• time-devision multiple access (TDMA)
• fixed time intervals for bus writing
• fixed assignment: node → intervals

➡ static, deterministic schedule

• nodes: only list with own transmission times
• different approach: event-triggered
• fundamental element: communication cycle
• periodically, recurring time unit
• whole schedule executed once

9

Peter Böhm 27.09.05

• 5. Schedule: Communication Cycle
• static slot:
• 1 message per static slot
• all same length, i.e. same amount of macroticks
• TDMA part of schedule
• unique, fixed assignment to a node
• symbol window:
• special messages, called symbols
• wake-up symbol
• network idle time:
• needed for clock synchronization

10

communication cycle

static segment symbol window network idle time

t

static slot static slot

Peter Böhm 27.09.05

• 6. Message Processing

11 Controller Host Interface Payload Media Access Control Payload Header, Payload CRC Append Coding Frame physical bus Bits Decoding Bits CRC Check Frame Frame & Symbol Processing Header, Payload Controller Host Interface Payload Host B (Receiver) Payload Host A (Sender)

Peter Böhm 27.09.05

• 7. Clock Synchronization
• problem:

➡ physical clocks deviate ➡ TDMA-schedule: consistent view of time required to ensure communication

• synchronization of local clock against a fictive global clock
• fictive global clock derived from some node’s view of time
• FlexRay clock synchronization provides:
• ability to use the most accurate clocks for synchronization
• fault-tolerance

12

Peter Böhm 27.09.05

• 8. Wake-up/Start-up: Error Model
• 3 level error model
• active
• normal operation
• no error state
• passive
• an error occurred (e.g. clock synchronization failed)
• node does not transmit and just listens the bus
• trying to reintegrate
• halt
• entered on host request or a fatal error detection
• node completely stops operation

13

Peter Böhm 27.09.05

• 8. Wake-up/Start-up
• wake-up/start-up strategy needed after:
• 1. power-on
• 2. entering passive mode
• power-on: nodes start with non-synchronized clocks
• some nodes serve at masters
• others adopt their view of time
• passive mode: (e.g. due to clock synchronization failure)
• node need to reintegrate itself
• performs clock synchronization until its view of time is

corrected

14

Peter Böhm 27.09.05

• 9. Summary
• very flexible network topology

➡ scalable fault-tolerance

• time-triggered schedule with no common knowledge
• fault-tolerant message transmission with error checks
• fault-tolerant clock synchronization
• passive mode

➡ self-diagnostic error mechanism with possible reintegration

15

➡ flexible as well as fault-tolerant communication protocol