New Approach to Improve the Generalized Byzantine Agreement Problem

Abstract

To achieve reliability, it is very important to study the agreement and fault-tolerance topic in distributed systems. This kind of problem is known as a Byzantine Agreement (BA) problem. It requires a set of processors to agree on a common value, even if some processors and transmission media are faulty. Basically, the traditional BA protocols require (n-1)/3 + 2 rounds of message exchange to make each processor reach an agreement. In other words, those protocols are inefficient and unreasonable, due to the number of message results in a large protocol overhead. In this study, we propose a novel and efficient protocol to reduce the amount of messages. Our protocol can compare and replace the messages received from other processors to find out the reliable processors and replace the value sent by the un-reliable processors through four rounds of message exchange even when the total number of faulty components or the total number of processors in the system is large. Finally, the agreement can be reached by using the minimum number of messages in the distributed system, while tolerating the maximum number of faulty components.