Event Driven Fault Diagnosis and Partition Detection (ED-FDPD) Algorithm

Abstract

Fault detection in a distributed computing system is a challenge that needs to be adaptive and efficient. Here, a new algorithm, namely "Event Driven Fault Diagnosis and Partition Detection (ED-FDPD) Algorithm for distributed systems" is proposed. ED-FDPD is applicable to any arbitrary topology. This is a sequential algorithm which gets executed periodically on each system to detect faulty nodes with minimum number of tests. The t-diagnosibility of this algorithm is (N-1) where, ‘N’ denotes the number of nodes in the network. ED-FDPD allows new nodes to be recognized in a diagnostic cycle. Also, repaired faulty nodes can reenter during a diagnostic cycle. There may be vulnerable nodes in the system which when becoming faulty can create a partition in the network. Detection of such vulnerable nodes is also incorporated in every diagnostic cycle. The diagnostic cycle does not stop even after the partitioning of the network. Administrator is informed about faulty, fault-free and vulnerable nodes in the entire network after the end of every diagnostic cycle.