|A new Byzantine faults tracing protocol
Our tracing protocol will exploit the broadcast channel assumption, ping (heart beat) channels and an authentication mechanism:
The broadcast channel assumption: All nodes within one broadcast-hop range to listen to each other. In particular, if nodes A and C are one hop neighbors of node B, then node A can listen to all communication from B to C.
The ping channel: This is established by having each node broadcast at regular intervals a ping when there are network activities.
The authentication mechanism: This makes it possible for all the nodes to authenticate their messages. Depending on the tracing requirements we will either use a keyed MAC or a digital signature (such as the RSA cryptosystem ).
It is easy to see that by combining these primitives, it is possible to trace all ad hoc link faults which involve at least one non-faulty node, provided that there are fault-free paths linking any two nodes. Indeed, if a link is broken then the adjacent nodes will become aware of this since they will not receive each other’s pings. They can then send an authenticated broken link report to the sender and receiver on any route, which uses this link. Alternatively, the broken link may be fixed as in AODV  and a report sent to the sender node providing the updated path length. A broken link between malicious nodes is of little relevant since it is, in any case, under control of the adversary.
Share with your friends: