Back to search

Aristotle University Of Thessaloniki (AUTH)

Decentralized systems frameworks for transactional data analytics

Abstract

dc:description

The emergence of decentralized systems, blockchain technologies, and cryptocurrency applications marks a new era of digital transactions and financial interactions. These advancements hold immense potential for transforming traditional systems; however, their adoption and practical implementation are hindered by multifaceted challenges. These challenges encompass technical, regulatory, and usability issues, including transaction pattern analysis in pseudo-anonymous environments, privacy and efficiency in lightweight Bitcoin clients, the influence of social media on cryptocurrency markets, the design of secure decentralized applications, reliable recovery methods for cryptocurrency wallets, and performance optimization in fog-based database management. This thesis systematically addresses these critical challenges, presenting a series of targeted contributions that explore diverse yet complementary aspects of decentralized technologies. The contributions span a broad spectrum of topics, and while each contribution tackles a distinct problem, they collectively underscore the shared challenges of enhancing security, scalability, and usability in decentralized systems. The first contribution focuses on the classification of Bitcoin transactions within a pseudoanonymous ecosystem. The thesis introduces an unsupervised clustering framework that relies solely on intrinsic blockchain data, bypassing the need for external labeling. This framework uncovers behavioral patterns in Bitcoin transactions, such as common user activities, exchanges, mining operations, and potential illicit behavior, providing a foundation for deeper analysis of decentralized ecosystems. The second contribution examines the trade-offs between privacy and efficiency in lightweight Bitcoin clients using Bloom filters. The work evaluates how varying false positive rates impact privacy, resource consumption, and security risks. This analysis sheds light on optimization strategies to balance privacy with resource constraints, addressing challenges critical to the scalability and security of lightweight Bitcoin solutions.The third contribution explores the dynamic relationship between social media activity and Bitcoin market trends. By analyzing data from the Reddit platform, the thesis identifies significant correlations between online discourse and key market metrics such as price fluctuations and trading volume. This research bridges the gap between digital communication and financial market analysis, providing a framework for understanding and leveraging investor sentiment in cryptocurrency ecosystems. The fourth contribution focuses on designing decentralized frameworks that ensure privacy, security, and trustworthiness. The thesis introduces DeTRACT, a blockchain-based decentralized public key infrastructure (PKI), and DeCStor, a secure file-sharing architecture that integrates strong encryption with decentralized storage. These frameworks address key challenges in trust, usability, and scalability, demonstrating how decentralized systems can replace traditional centralized solutions effectively. The fifth contribution addresses the issue of secure and reliable recovery phrase retention in cryptocurrency wallets. A gamified solution, MnemonicMaker, incorporates the Method of Loci, a proven spatial memory technique, into an engaging framework that enhances users' ability to recall recovery phrases over extended periods. This contribution not only addresses a critical security concern but also improves the overall user experience for cryptocurrency adoption. The sixth and final contribution provides a comparative performance analysis of distributed in-memory database systems—Apache Ignite and Redis—within fog computing environments. By benchmarking these systems under various workloads and network conditions, the thesis evaluates performance trade-offs between advanced features like ACID compliance and the high-speed efficiency of Redis. This analysis offers practical guidance for selecting database systems that balance feature requirements and performance in dynamic, resource-constrained fog ecosystems. By addressing these six challenges through targeted contributions, this thesis advances the state of knowledge and practice in decentralized and the relevant issues in distributed systems. The findings provide foundational insights into enhancing the security, privacy, usability, and performance of blockchain-based applications, paving the way for their broader adoption and practical implementation in diverse contexts.

Degree

thesis:*
Grantor dc:publisher
Aristotle University Of Thessaloniki (AUTH)
Year dc:date
2025

Author and committee

dc:creator, dc:contributor.*
Authors dc:creator
  • Vlachavas, Georgios
  • Βλαχάβας, Γεώργιος

Subjects

dc:subject × 37

Rights

Language dc:language
eng

Identifiers

dc:identifier.*
Identifier
10.12681/eadd/58481

Chain of custody

source
Harvested from
Greek National Archive of PhD Theses
Base URL
phdtheses.ekt.gr/eadd_oai/request
Last updated
2026-11-18
Source record
OAI-PMH GetRecord
citation

Vlachavas, Georgios; Βλαχάβας, Γεώργιος. Decentralized systems frameworks for transactional data analytics. Aristotle University Of Thessaloniki (AUTH), 2025. http://hdl.handle.net/10442/hedi/58481