Towards more energy-efficient blockchain

Bitcoin has historically used the proof-of-work (PoW) consensus in which the validation of a block in the chain is associated with a random mathematical problem requiring validators to deploy substantial and energy-intensive computing resources.  Validators’ investment in hardware and energy is rewarded with a predetermined amount of the cryptocurrency (or asset) associated with the blockchain.

The problem, in short, is that PoW uses a lot of energy, which raises major  environmental concerns. Proof-of-stake (PoS) protocols are different. Validators pledge their own assets to verify transactions, which uses mush less energy. If a minority participant tries to validate a fraudulent transaction on the blockchain, the currency he or she put up is lost. This effectively penalizes fraud and rewards correct validation.

PoS is more susceptible to manipulation than PoW, however. CEA-List is working with academic research labs and with companies to develop blockchains that are both energy-efficient and reliable by detecting vulnerabilities in existing protocols and developing new secure protocols.

CEA-List has been conducting theoretical research on PoS and investigating ways to detect vulnerabilities in pre-existing blockchain implementations since 2018. This research has led to the identification and correction of critical security flaws.

• Research on Tendermint, used for the cryptocurrency Cosmos, resulted in a precise theoretical description demonstrating that the algorithm did not in fact reward players in proportion to the stake effectively put up. Tendermint (now Ignite) decided to partner with CEA-List to implement the necessary corrections.
• Ethereum, whose market capitalization was estimated at $250 billion in December of 2023, was also potentially affected by theoretical vulnerabilities identified by CEA-List researchers. The research resulted in several publications, including one on “Byzantine Attacks Exploiting Penalties in Ethereum PoS” at the International Conference on Dependable Systems and Networks (DSN) in 2024.

The Fantastyc project, which combines CEA-List’s trusted distributed systems know-how with expertise in artificial intelligence and cryptography, is developing a breakthrough use case for blockchain technology: trusted federated learning.

In federated learning, multiple participants can share data used to train an AI without revealing the contents of the data. By storing multiple participants’ contributions in a distributed yet traceable manner, blockchain can transform federated learning into trusted learning. For companies, for instance, the contributions of all employees could be validated using blockchain. Bad actors could also be identified more easily, such as during an audit.

CEA-List has partnered with academic labs and companies to develop multiple variants of Tendermint to make the protocol more resilient to network instabilities and attempted fraud.

• For example, in 2019, Nomadic Labs asked CEA-List to implement a version of Tendermint that would be resilient to an unstable communication network on the French blockchain Tezos. CEA-List developed an algorithm that could withstand an intermittent internet connection and synchronization errors.
• Later, in 2021, CEA-List helped Clearmatics develop Autonity, a blockchain for financial markets that leveraged new algorithms to detect unusual behavior by participants attempting to form alliances to carry out fraudulent transactions.

CEA-List’s innovations were effectively implemented on both of these blockchains.