Modeling improvement and application of DAG-based blockchain networks

Abstract

Blockchain, a revolutionary distributed ledger technology (DLT), has garnered significant attention in recent years. Leveraging cryptographic techniques, blockchain ensures the immutability, traceability robust security, and privacy of data. It operates as a decentralized system, thereby safeguarding against single-point failure attacks. However, as the volume of data, continues to surge due to the proliferation of Internet of Things (IoT) devices, the blockchain ecosystem faces challenges. A key issue lies in scalability. Blockchain’s transaction processing capacity significantly lags behind centralized systems. In response to this challenge, researchers have devised various strategies to enhance blockchain scalability. One promising approach involves adopting a Directed Acyclic Graph (DAG) data structure, as opposed to the conventional linear chain structure. IOTA, a renowned blockchain built on the DAG model, employs a data structure known as the “tangle”. This innovative structure employs a Markov Chain Monte Carlo (MCMC) random walk algorithm to attach new transactions to the tangle. Theoretically, a higher volume of transactions attached to the tangle should result in improved Transaction per Seconds (TPS). In practice, however, several challenges emerge. There is no universally prescribed transaction selection algorithm, and the true behavior and structure of the tangle remain elusive. The MCMC algorithm introduces an influential random walk weight factor, impacting the security and scalability of the IOTA tangle. A larger weight factor may enhance security but lead to more unconfirmed transactions, while a smaller factor may reduce unconfirmed transactions but compromise security. Achieving high TPS under real-world conditions proves to be a formidable undertaking. As a result, while blockchain technology offers transformative potential, addressing the scalability issue in a practical setting, especially within the context of DAG-based systems like IOTA, presents a complex and multifaceted challenge. In this thesis, we undertake a comprehensive analysis and dynamic model generation of the real IOTA tangle. Additionally, we introduce a secure, and scalable TSA and devise a lightweight authentication mechanism rooted in IOTA. Our overarching objectives encompass: (i) unveiling the topological attributes, performance metrics, and generation model of the actual IOTA tangle, (ii) crafting a tip selection algorithm tailored to the unique characteristics of DAG-based blockchains, and (iii) designing an IoT application underpinned by the IOTA framework. Our exploratory journey begins with the retrieval of the authentic tangle database. By comparing these characteristics with those of a simulated tangle, we discern substantial differences. Furthermore, we embark on a quest for greater precision in mapping the in-degree distribution and charting the evolving tangle topology within the IOTA realm. The application of various long-tail distributions reveals that the double Pareto Lognormal (dPLN) distribution surpasses its peers in terms of fitting accuracy. The in-degree distribution unveils that the majority of transactions garner just one approval, signifying inherent fragility in the topology with a profusion of blowballing structures. In the subsequent segment of our thesis, we shift our focus towards optimizing the Tip Selection Algorithm (TSA) for DAG-based blockchains. A particular emphasis lies in addressing real-world challenges faced by IOTA. We inaugurate with a swift TSA tailored for burst transaction arrivals in DAG-based blockchains. This novel approach avoids the weighted random walk process and precomputes the tip selection probability distribution, dramatically expediting the selection task. We then extend this algorithm into a secure and scalable variant. After calculating tip selection probabilities, we identify and select abnormal tips based on predefined thresholds and subsequently attach new transactions randomly. Our proposed tip selection algorithm tackles two critical issues: (i) fortifying the tangle against the influence of irregular structures and (ii) stabilizing and minimizing the count of unconfirmed transactions. In the end, the thesis culminates with the design of a lightweight proximity-based authentication mechanism tailored for cross-domain IoT devices, underpinned by the IOTA platform. IOTA serves as the repository for certificates employed during authentication. We substantiate the feasibility of our proposal through the implementation of a compact in-house prototype system.