FireChain: An Efficient Blockchain Protocol using Secure Gossip

Abstract

Blockchains have become an integral part of many distributed applications, providing a new platform for interaction between system components. Blockchains are perhaps most known for their use in crypto-currency systems, such as Bitcoin and Ethereum, where pseudo-anonymous parties engage in transactions without a trusted third party. Blockchain systems often struggle to meet performance demands of real-word applications, rendering them inappropriate for performance sensitive applications. There is also concerns regarding the immense amount of electrical energy required to securely run existing public blockchain systems. Bitcoin alone consumes more than small countries. Private systems have higher throughput and avoid excessive energy consumption, but have closed membership and do not scale to the same extent.

Both public and private blockchains rely on some form of membership mechanism providing peers with a view of other participants. Existing systems often employ partial view protocols due to their natural scalability. However, recent work have shown that full view protocols are feasible in practice, and can scale to thousands of participants. With full membership, applications can send messages directly to their destination without any intermediate hops.

This thesis presents FireChain, which combines a Byzantine fault-tolerant gossip service and full membership, with a proposal for blockchain systems that does not consume excessive energy. We evaluate FireChain's performance through experiments on PlanetLab, and show that it scales beyond hundreds of members.