Privacy Protection in Average Consensus

Abstract

Consensus into an average is an essential algorithm that is extensively used in decentralized and distributed computing network. To achieve an agreement across all the nodes in the network, the repeated process of communicating with each other is the beginning of everything. However, during such process, it is very likely that some participants undesirably disclose sensitive and private information and thus are unable to be guaranteed with privacy. There are two noticeable approaches to make sure the privacy is not compromised while exchanging state information. One is to add and subtract random noises during the consensus process. The other is to integrate partial homomorphic cryptography so that the consensus protocol is kept confidential. In this project, the two renowned methods are thoroughly studied and fully implemented as a software to simulate the convergence to an average. Moreover, a website is provided for users to simulate the algorithms and observe the performances and results with ease.