A Unified Approach to Assessing the Structural Resilience of Blockchain Overlay Networks

Abstract

Blockchains have gained significant attention for their unique properties, such as immutability, public verifiability, and decentralization. These attributes have generated interest in both the software industry and the research academia, sparking enthusiasm for blockchain-based applications. The success of Bitcoin has showcased the potential of blockchain technology, and its distinct features are expected to revolutionize various industries and disrupt sectors that rely on centralized third parties.

Blockchain systems operate through peer-to-peer (P2P) networks, which play a crucial role in ensuring consensus and data propagation. The security and correctness of blockchain applications hinge on the resilience of these P2P networks. In this thesis, we dive into the structural properties of seven distinct blockchain networks, focusing on the implications of their characteristics for network resilience. Our study reveals vulnerabilities to targeted attacks and uncovers hidden interconnections among networks, emphasizing the importance of strengthening network defenses.

To overcome the challenge of accurate topology inference, we introduce a simple yet effective approach. Using the node advertisements shared by the network nodes, we construct connectivity graphs that include any potential connections between peers. Our methodology captures both actual and potential connections, improving our understanding of blockchain P2P networks and their structural characteristics. Furthermore, we avoid classifying nodes and links according to their role or position in the network. Instead, we adopt a unified network model, streamlining the analysis and enabling the identification of shared structural vulnerabilities present in diverse network configurations across various blockchain systems. Addressing these vulnerabilities can lead to improved network resilience.

Our findings shed light on the dynamic nature of blockchain overlay networks and their susceptibility to targeted attacks. We observe variations in the distribution of session lengths over time and a strong correlation between a node’s uptime and its degree. We highlight the importance of considering network implementation in blockchain systems and the need for tailored solutions to enhance security and resilience.