Enhancing Fault-Tolerance in a Distributed Mutual Exclusion Algorithm

In this paper, we present an efficient fault-tolerant token-based algorithm for achieving mutual exclusion (ME) in distributed systems. Nishio et al's fault-tolerant mutual exclusion algorithm requires feedbacks from every other site to recover from token loss. This results in considerable amount of waiting time and false token loss detection. Though Manivannan et al's algorithm solves the problems in Nishio et al's, their algorithm cannot work if a failed site is not repaired within a finite time. This paper proposes an approach to remove the drawback of Manivannan et al's method for achieving fault-tolerance. Our algorithm gives better performance in terms of message complexity (MC), Synchronization delay (SD), response time (RT) and degree of fault- tolerance in comparison with Manivannan et al's algorithm.