P.M. Khilar and S. Mahapatra (India)
Distributed diagnosis, dynamic failures, fault tolerance, synchronous systems
: The problem of identifying faulty mobiles in ad-hoc networks is considered. Current diagnostic models were designed for wired network. For the suitability to multi-hop ad-hoc network this paper considers the problem of achieving distributed diagnosis in arbitrary topology networks in dynamic fault environments. In order to diagnose the state of the wireless adhoc network, diagnostic tasks are assigned to all mobiles and the outcomes of these tasks are compared. We describe a new distributed self-diagnosis protocol for wireless adhoc networks that is able to identify crash and value faults in a finite amount of time. Heartbeat based mechanisms have been used to detect the failure and recovery events. It is assumed that the mobiles can change their states at any time provided the network remains connected and the frequency with which events can occur on a node has an upper limit. This is characterized and quantified by deriving a minimum state holding time. The algorithm DDMAN is compared with a similar algorithm FHB given in paper [5] and is shown to produce shorter average diagnostic latency with comparable message complexity under a more realistic fault model.
Important Links:
Go Back
