Please use this identifier to cite or link to this item:
|Title:||Blockchain and federated learning for data sharing in vehicular network||Authors:||Qiu, Stanley||Keywords:||Engineering::Electrical and electronic engineering||Issue Date:||2022||Publisher:||Nanyang Technological University||Source:||Qiu, S. (2022). Blockchain and federated learning for data sharing in vehicular network. Final Year Project (FYP), Nanyang Technological University, Singapore. https://hdl.handle.net/10356/158055||Project:||A3303-211||Abstract:||A relatively new development in the field of machine learning, is the federated learning framework. It was mainly developed to preserve the privacy of data contributors in the process of collective machine learning training and data mining. Federated learning, however, have a weakness in that the global aggregator of the local gradients is centralized in the server whereupon data contributors will upload their local training gradient updates. This is a type of single point weakness that is able to be rectified using the forefront of decentralization technology: the blockchain. The blockchain primarily utilizes the principle of distributed consensus in its core, making it robustly reliable even in network systems with multi-millions of users with competing self-interests like in Ethereum or Bitcoin. This report endeavors to simulate a federated machine learning framework implemented in an internet of vehicles (IoV), with 2 classes of participating nodes: vehicles, and road side units (RSU). Taking into account the aforementioned single-point failure of the central aggregator in the federated learning framework, a blockchain using Delegated Proof of Stake (DPoS) consensus protocol will also be integrated into the overarching framework, resulting in a blockchain-federated learning (BFL) hybrid framework. This report will then evaluate the soundness of the framework in the end by using the federated learning’s global model accuracy and blockchain currency distribution as conclusive metrics.||URI:||https://hdl.handle.net/10356/158055||Fulltext Permission:||restricted||Fulltext Availability:||With Fulltext|
|Appears in Collections:||EEE Student Reports (FYP/IA/PA/PI)|
Updated on Jun 27, 2022
Updated on Jun 27, 2022
Items in DR-NTU are protected by copyright, with all rights reserved, unless otherwise indicated.