FlexRay 44-5
44.3.2 Node architecture
A.typical.node.architecture.consists.of.a.host.CPU.subsystem,.a.network.controller,.and.bus.drivers.
for.the.separate.channels;.see.Figure.44.5..Interfacing.between.the.host.and.the.network.subsystem.is.
facilitated.via.a.controller.host.interface.employing.some.kind.of.dual-ported.memory..is.choice.of.an.
interface.allows.separating.control.from.data.ow,.creating.some.kind.of.temporal.rewalls.restricting.
error.propagation.to.dened.areas..Optional.bus-guardians.may.be.installed.either.at.the.nodes.or.at.
central,.active.star.couplers..e.bus-guardians.oversee.and.inhibit.bus.access.whenever.a.node.is.not.
allowed.to.transmit.in.order.to.avoid.so-called.babbling.idiot.failures;.situations.where.a.node.playing.
havoc.disrupts.the.entire.communication.
44.3.3 Star Couplers
Both.passive.and.active.star.couplers.are.supported.by.the.FlexRay.protocol..Whereas.passive.star.cou-
plers
.are.simply.circuits.that.relay.the.signal.received.on.one.branch.to.all.other.branches;.active.star.
couplers.provide.logic.and.circuitry.for.power-down.and.wakeup,.error.detection.and.isolation,.as.well.
as.signal.reshaping..Typically,.an.active.star.coupler.listens.on.all.branches.for.trac.and.when.a.signal.
is.detected.it.is.relayed—in.an.error.free.case—to.all.other.branches..Whenever.an.error.occurs.or.before.
a.cluster.is.synchronized,.collisions.of.data.frames.or.symbols.may.occur..In.this.case,.an.active.star-
coupler
.relays.the.superimposed.signal.to.the.other.branches.
44.4 System Design Considerations
Guaranteed.bandwidth.and.reliable.operation.are.some.of.the.benets.of.time-triggered.communica-
tion
.systems..On.the.other.hand,.such.systems.may.become.less.exible.(especially.when.frequent.sys-
tem
.changes.occur).and.require.higher.eorts.at.design.time.
44.4.1 Conguration
Next.to.topology.considerations,.one.has.to.congure.which.nodes.act.as.coldstarters.and.participate.
at.the.clock.synchronization.(sending.of.sync.frames)..At.least.two.or.more.nodes.should.be.assigned.
therefore;.in.fact,.more.of.these.nodes.enhance.the.availability.of.the.cluster.but.increase,.however,.the.
likeliness.for.the.establishment.of.cliques.(cliques.are.groups.of.nodes.that.are.able.to.communicate.with.
each.other,.however,.not.with.other.nodes),.cf..[MHS08].
Furthermore,
.the.physical.distribution.of.these.nodes.as.well.as.the.distribution.of.their.slots.impairs.
the.overall.robustness.of.the.cluster..In.fact,.having.these.nodes.and.slots.evenly.dispersed.over.the.
cluster.and.schedule,.respectively,.will.likely.improve.the.overall.availability..Furthermore,.conguring.
more.nodes.as.the.minimum.to.transmit.the.specied.wakeup.patterns.and.perform.coldstart.attempts.
will.improve.the.startup.mechanism..When.one.of.these.so.congured.nodes,.however,.is.faulty,.a.higher.
Controller
host
interface
Protocol
engine
Bus guardian
Host
subsystem
Bus
driver
Communication channels
FIGURE 44.5 Node.architecture.
© 2011 by Taylor and Francis Group, LLC