Skip navigation

Improved blockchain - Proof-of-work consensus protocol - performance using machine learning

Improved blockchain - Proof-of-work consensus protocol - performance using machine learning

Ahmed, Mujistapha Safana (2021) Improved blockchain - Proof-of-work consensus protocol - performance using machine learning. PhD thesis, University of Greenwich.

Mujistapha Ahmed Safana 2021.pdf - Published Version
Available under License Creative Commons Attribution.

Download (5MB) | Preview


Blockchain technology has proven to be a secured and reliable technology by bringing security, trust and data integrity to a distributed system. It is a new paradigm that helps in the existence of cryptocurrency and eliminates the third party in a financial transaction. It has the potential to optimise, enhance and streamline many processes outside the cryptocurrency and financial sector but the adoption of the technology is limited by the hindering performance issue. Unfortunately, the current blockchain suffers a performance degrade with the increasing size because of the complexity of its consensus protocol known as Proof-of-Work (PoW). Many industries, researchers and organisation have been working on providing a solution to the performance issues of the technology but most of the proposed solutions has so far ended in proposing a newly designed protocol which ends up facing another issue referred to as the scalability issue; having to trade off one of security or decentralisation to get speed. To address the performance issue, the research has carried out experiments to clear pathways in identifying the specific problem and the outcome has identified the mining process, block size and scalability as the main factors affecting the performance of the technology. The research further investigated these factors and identified the time taken to generate a block as the most time-consuming task within the consensus process, regardless of the traffic, size or number of connected nodes. The research has also explored alternative ways of speeding the nonce finding process and identified machine learning as the best technique because of its ability to learn and predict. Using the quantitative approach of the research, different machine learning models were analysed and compared, and linear regression was identified as the best fit model for the research problem. The research used linear regression model Machine Learning technology to reduce the block generation time without sacrificing security or decentralisation of the proof-of-work consensus protocol. The model has achieved a 58 percent accuracy improvement on the traditional mining process. The model reduces the block generation time when tested on the blockchain simulation by an average of 4 seconds on the Ethereum network and a more significant reduction for the Bitcoin network depending on the computer hardware. In this thesis, blockchain is referred to as the blockchain that uses the PoW consensus protocol.

Item Type: Thesis (PhD)
Uncontrolled Keywords: Blockchain technology, Proof-of-Work (PoW), computer modelling,
Subjects: Q Science > QA Mathematics
Faculty / School / Research Centre / Research Group: Faculty of Engineering & Science
Faculty of Engineering & Science > School of Computing & Mathematical Sciences (CMS)
Last Modified: 11 Sep 2023 15:51

Actions (login required)

View Item View Item


Downloads per month over past year

View more statistics