A research paper from the Decentralized and Distributed Systems lab at the Swiss Federal Institute of Technology in Lausanne proposes a new protocol, Byzcoin, that can be incrementally incorporated into Bitcoin in a backward-compatible way, allowing the network to handle more than one hundred transactions per second at a 1MB blocksize and near PayPal levels with just 4MB blocks.
The protocol’s consensus mechanism is highly scalable, says Philipp Jovanovic, a cryptographer and post-doctoral researcher in the Decentralized and Distributed Systems lab, who told CCN in an interview that:
[It] uses tree-structures for communication which gives you logarithmic communication overhead. Moreover, blocks that contain transactions do not require Proof-of-Work, but instead are formed by the current leader and sent to the consensus group for signing.
ByzCoin is divided into keyblocks and microblocks. A miner who fids a keyblock becomes the leader, then forms microblocks made of transactions. The microblocks are sent to a consensus group formed of miners that have recently found a keyblock to co-sign the microblocks, thus verifying authenticity and compliance with protocol rules. Any double spending, therefore, is prevented as each block is approved by a supermajority of all miners.
The overall structure is somewhat similar to Bitcoin-NG, but, in Bitcoin-NG the leader can misbehave while he is in power, Jovanovic says, which is prevented by the new protocol as a misbehaving leader can be replaced by a supermajority of 67%, with the removed leader losing “the rewards of his leadership epoch,” according to Jovanovic.
In combination with the rest of the protocol, this division into keyblocks and microblocks as well as the supermajority’s verification and approval of the leader’s actions, makes possible a number of features, such as instant, non-reversible, transactions, while solving quite a few problems, such as selfish mining.
Moreover, it proposes a solution to numerous attacks that become possible when miners primarily rely on transaction fees as recently revealed in a paper by Arvind Narayanan, a researcher at Princeton university, which states that “miner incentives will start to go haywire as Bitcoin rewards shift from block rewards to transaction fees.”
The primary reason seems to be because of bitcoin’s instant distribution of mining reward which can be solved with delayed gratification according to Bryan Ford, a former researcher at Yale University who now leads the Decentralized and Distributed Systems Laboratory. The new protocol therefore proposes that mining reward is distributed daily or weekly, instead of instantly, thus preventing a number of attacks illustrated by Narayana.
The paper, first published at USENIX Security’16, was presented at the Scaling Bitcoin workshop held last month in Milan where “it was quite well received,” says Jovanovic, but “for a real deployment, e.g. in Bitcoin, there are still several engineering challenges that need to be addressed first.”
Jovanovic says that the researchers plan to present it as a Bitcoin Improvement Proposal (BIP) to start a more concrete discussion without stating whether there will be a ByzCoin-based cryptocurrency or not, but the techniques can be applied to any public blockchain, according to Jovanovic.
The team’s proposal is the latest on scalability, a hot topic in the public blockchain space with a number of discussions and research papers suggesting ways to scientifically address the problem. One solution by the Bitcoin Core client is segwit, which is awaiting activation, likely to be followed by the Lightning Network. Other teams, such as Bitcoin Unlimited, are focused on on-chain scalability with Peter Rizun proposing subchains, a similar proposal to Bitcoin-NG or ByzCoin.
However, Lightning itself will need much greater capacity than two or three transactions per second. Therefore, the ultimate solution will probably be a combination of on-chain and off-chain scaling, with the recently published paper further contributing to the debate as it uniquely allows for vastly more transactions while using just 1MB or 4MBs of space, solving the problem therefore, at least for another decade.
But, whether it finds approval among the vast majority of bitcoin developers, or whether bitcoin developers find something new that scales bitcoin while using the same resources, remains to be seen.
Images from Shutterstock and Hacking Distributed