De Angelis, Stefano (2022) Assessing security and performance of blockchain systems and consensus protocols: taxonomies, methodologies and benchmarking procedures. University of Southampton, Doctoral Thesis, 173pp.
Abstract
Blockchain promises to improve systems security and trust by decentralising computer infrastructures. However, decentralisation also requires higher complexity that may lead to performance issues. With the rapid growth of blockchain adoption, such properties are paramount, and it becomes crucial to assess them in different application scenarios. In this PhD thesis, we study performance and security of modern blockchain systems. We first refine the standard concepts of security and dependability, defining a set of properties for blockchain systems. We provide a taxonomy of platforms, consensus protocols, and smart contacts vulnerabilities, and we assess their security according to the proposed properties. We show that consensus strictly impacts system’s security. We also argue that it introduces trade-offs with performance that must be understood for building secure and efficient systems. So we design METHUS, a systematic methodology to assess blockchain consensus protocols applying qualitative and quantitative methods. Hence we evaluate two families of consensus protocols used in permissioned blockchains, and we show that a traditional Byzantine Fault Tolerant approach is preferable in this context. Extending the study to permissionless blockchains, we propose PETHARD, a framework to measure performance of consensus employed in two famous blockchains, namely Ethereum and Algorand. Despite promising results, PETHARD only simulates testing setups and cannot be used to evaluate realistic deployments. To this extent, we design PERSECUS which defines the standards for blockchain benchmarking. PERSECUS fosters efficient and precise measurements simulating various setups and real-world scenarios. We benchmark two blockchains, namely Parity and GoQuorum, evaluating their security, performance, and scalability properties. We illustrate that, besides consensus, other blockchain components, such as configuration of nodes parameters and transactions serialisation, strictly affect performance and security. To conclude this thesis, we discuss the possibility of using elasticity, broadly adopted in Cloud Computing to automatise the provisioning of a system, to enhance performance and security in blockchain systems.
More information
Identifiers
Catalogue record
Export record
Contributors
Download statistics
Downloads from ePrints over the past year. Other digital versions may also be available to download e.g. from the publisher's website.