If you do not use JBossCache or SessionReplication the traffic between the notes of a cluster not high.
You might watch the multicast traffic with the configured multicast address for JGroups or analyse the JGroups logging.
Lot of traffic might cause by an application if you want to have a failover, in this case statefull sessions must be replicated between the cluster parts.
Look at the attributes "BytesSent, BytesReceived, MessagesSent, MessagesReceived" for the mbeans matching the following filter in for example the jmx-console:
(or "jboss.jgroups:protocol=TCP,type=protocol,*" if you changed the transport to TCP)
If you want to get even more detailed usage stats, you could add SIZE or STATS to the JGroups stack. Both protocols expose their information via JMX, so jconsole or jmx-console would allow you to see them.
If you use UDP as transport, and have IP multicasting enabled, you could also use probe.sh, e.g. probe.sh query=jmx (or probe.sh jmx=STATS in 2.10+) would ping nodes and return JMX info