Multi-attribute based self-stabilizing algorithm for leader election in distributed systems

Abstract

Cloud computing, edge computing, replicated databases, and various modern technologies are adopting distributed computing concepts for a reliable, high-performance, large-scale computational platform. In distributedcomputing, leader election is a fundamental problem because the elected leader helps coordinate andutilize the resources efficiently. Several state-of-the-art works mentioned that a good quality leader is essentialas it improves system performance, simplifies management procedures, reduces coordinational complexity, andmakes the system more fault-tolerant. However, designing a self-stabilizing leader election algorithm with weakassumptions in a failure-prone distributed environment is challenging. This paper proposes a multi-attributebased,self-stabilizing, network partition-tolerant leader election method for failure-prone distributed systems.Here, based on the system requirements, the pertinent attributes of a good quality leader are identified and assignedweights according to their importance. Then the identified attributes and their weight are used to elect asuitable node as the leader.We show that the algorithm is self-stabilizing and can tolerate multiple nodes and linkfailures. Further, we analyze the proposed algorithm’s time, communication, and bit complexities. We consider adistributed database system scenario to simulate the proposed method and compare it with existing approachesto evaluate and validate the proposed method’s performance and the elected leader’s quality.