Ian (QiYang) Tan

My research focuses on the distributed system and Byzantine consensus protocols. The PBFT protocols tolerates f Byzantine faults with n>=3f+1 replicas and guarantees that the consensus is achieved in every iteration. Some systems achieve an improved fault tolerance by restricting n>=2f+1 but does not guarantee the consensus in each iteration, which means there can be a fork before the consensus is formed. However, the improved fault tolerance system is more vulnerable to attacks and its efficiency degraded since there are more byzantine nodes in the distributed system. My goal is to understand fundamental performance limits of the consensus protocols and to learn how to code and test the performance of the consensus protocols in the real world.

In the 2f+1 distributed system, we focus on the average time to achieve consensus in the synchronous authenticated Byzantine agreement. There is a great improvement on the average time to form a consensus in the last two decades, from 24 rounds (Katz and Koo, 2006) to 8 rounds (Abraham 2019). The method is presented theoretically and some details of implementations have not been discussed, for examples, the way to elect a leader, to create safe authenticated signatures, etc. Some potential directions are to code the protocol practically and to analyze the performance of the consensus protocol. Also, we can further reduce the time complexity to achieve the consensus by modifying the Byzantine consensus protocols.