A process q sends to processes an “I am alive” message and in addition, sends a request for a lease for some duration d . Some time before d expires, q must send a request for lease renewal to all the processes. Otherwise, if a process does not receive any message from q after d , it will put q in its suspected list.
Published in Chapter:
Unreliable Failure Detectors for Mobile Ad-hoc Networks
Luciana Arantes (University Paris 6, France), Fabíola Greve (Federal University of Bahia, Brazil), and Pierre Sens (University Paris 6, France)
Copyright: © 2011
|Pages: 17
DOI: 10.4018/978-1-60960-042-6.ch063
Abstract
Failure detection is an important abstraction for the development of fault-tolerant middleware, such as group communication toolkits, replication and transaction services. Unreliable failure detector (namely, FD) can be seen as an oracle which provides information about process failures. The dynamics and self-organization of Mobile Ad-hoc Networks (MANETs) introduce new restrictions and challenges for the implementation of FDs with which static traditional networks do not have to cope. It is worth mentioning that in some way, fault tolerance is more critical for MANETs than for the latter, since wireless network can present high error rates and mobile nodes are more prone to failures, physical damages or transient disconnections. The aim of this chapter is thus to discuss the impact of all these characteristics, intrinsic to MANET, in the implementation of FDs. It presents a survey of the few works about FD implementations for wireless networks, including the different possible assumptions to overcome the dynamics and lack of both global view and synchrony of MANETs.