Mining for Better: An Energy-Recycling Consensus Algorithm to Enhance Stability with Deep Learning

As the most popular consensus algorithm for blockchain, the Proof-of-Work (PoW) is suffering from the inability of handling computing power fluctuations. Meanwhile, PoW consumes a significant amount of energy without producing actual value. To address these issues, this paper proposes a deep learning-based consensus framework called Proofof-Improvement (PoI), which recycles the energy from mining blocks to improve the blockchain itself. In PoI, a new reward mechanism is used to encourage miners to include the high-accuracy model in their blocks. Then, based on PoI, a difficulty adjustment algorithm is designed. Experiments are done on real-world data and the result shows the proposed algorithm's proficiency in preserving block time stability with fluctuating hash rates. To the best of the authors' knowledge, PoI is the first to handle both energy recycling and difficulty adjustment concurrently.