Resource-efficient Blockchains for Reliable Data Management in Large-scale Distributed Systems
Loading...
Authors
Al-Mamun, Abdullah
Issue Date
2022
Type
Dissertation
Language
Keywords
blockchain , edge computing , HPC , privacy , provenance , supply chain
Alternative Title
Abstract
Nowadays, reliable data management is an essential task due to the explosion of data. The volume of data is increasing rapidly in each paradigm such as scientific computing and edge computing, etc. As a consequence, reliable data management such as dependable distributed resiliency, provenance tracking, and data fidelity become paramount in high-performance computing (HPC) systems. Besides, trustworthy data management in edge computing poses new technical security challenges. Although blockchain-based decentralized protocols exhibit a lot of potential for reliable data management in such paradigms, adopting the blockchain in its current model both in high-performance computing (HPC) and in edge computing ecosystems comes with a vast array of challenges. The shortcomings of the current model of blockchain lead to many open questions in terms of adaptability, scalability, privacy, and usability. To overcome the challenges lying with the state-of-the-art solutions, this dissertation aims to investigate numerous techniques to adopt blockchain in HPC and edge computing ecosystems. This dissertation takes a holistic view of the various infrastructure gaps and presents scalable approaches to provide reliable data management through resource-efficient blockchain protocols.This dissertation mainly consists of two parts. The first part investigates techniques on how to adopt blockchain-like decentralized protocols to ensure reliable distributed data management in the HPC system. More specifically, first, we introduce our work on blockchain that incorporates lightweight protocol to adopt blockchain in the HPC infrastructure to guarantee data fidelity for scientific provenance. Second, we introduce a series of protocols on blockchain that bridge the HPC-blockchain gap with two key components: (i) Lightweight consensus protocols for the HPC's shared-storage architecture, and (ii) A new fault-tolerant mechanism compensating for the MPI to guarantee consistency.The second part studies trustworthy data management in edge computing infrastructure. We propose a resource-efficient mechanism to adopt blockchain seamlessly in an edge computing infrastructure to prevent data manipulation and allow fair data sharing with quick recovery under resource constraints of limited storage, computing, and network capacity. We also focus on an edge computing use case: secure data management in electronic supply chain ecosystem. We propose a framework with the following novelties: (i) it leverages uniquely hashed physical unclonable function (PUF) challenge-response-pairs (CRPs) to manage the security of sensitive PUF data; (ii) it allows most transactions to be processed in parallel by distributing the workload among the participating blockchain nodes to maximize system efficiency; and (iii) it employs PUF-based keyless encryption to protect users’ privacy from the blockchain system.