Jitao Wang, Nong Tang, Yuzhou Wang, Kai Wang · 5 authors
Cross-chain technology, as a key driver for enhancing interoperability of blockchains, enables asset transfer and exchange between different blockchains. At present, cross-chain models based on light clients are widely adopted due to their fully decentralized nature and applicability to diverse scenarios. However, the rapid advancement of on-chain analysis techniques, such as address linkage and fund flow tracking, has significantly increased risks of de-anonymization in cross-chain transactions, posing serious privacy challenges. In this paper, we propose SharedRXC, a privacy-preserving asset cross-chain scheme for the light-client cross-chain model, which guarantees address unlinkability without extra privacy trust assumptions. First, to hide cross-chain addresses during interchain transmission, we propose the Ring Account (RA) to replace a single address for sending or receiving funds. In addition, we propose a zero-knowledge proof-based method to verify virtual identity ownership, allowing the virtual identity to track fund balances without exposing the addresses. Second, to prevent the exposure of the link between an address and its virtual identity caused by fund amount differences during deposits or withdrawals, which would compromise address unlinkability, we propose the Shared Burn/Mint method to obscure on chain fund change differences. Based on the liability equalization mechanism, we design two privacy-preserving cross-chain protocols: the cross-chain asset transfer (SharedRXC.T) and exchange (SharedRXC.E) protocols. Finally, we evaluateSharedRXC.TandSharedRXC.E, which reduce gas costs by 30% to 40% compared to zkCross and achieve execution times in the millisecond range. Therefore, SharedRXC provides a practical privacy-preserving solution for cross-chain financial applications in the multi-chain ecosystem.
Dhanush K, Tamilvelan S, Hari L, R. Roopa Chandrika
The growing adoption of lightweight, scalable, and resistant to tampering security mechanisms in the face of broad use of Internet-of-Things (IoT) and edge computing devices requires such mechanisms to be not based on centralized trust or bulky cryptography. The promising answer to this is the concept of Physical Unclonable Functions (PUFs) which relies on naturally existing manufacturing differences to produce device-specific, unclonable responses. Nevertheless, existing PUF-based authentication systems have significant flaws, such as centralized Challenge-Response Pair (CRP) storage vulnerable to attacks, vulnerability to machine learning, and no support of secure data recovery and sharing in distributed settings. This paper suggests a distributed authentication and recoverable data sharing framework that reduces these drawbacks, PUF-as-a-Service (PUFaaS). PUFaaS presents a multi-dimensional space of CRP, spreading the domain of challenges with respect to various operational parameters to maximize security against modeling attacks. Helper data of fuzzy extractors are secured with Shamir's secret sharing over distributed nodes and without having single points of weakness. A mechanism based on fuzzy vaults provides recoverable data binding, whereby encryption keys or sensitive data can be re-assembled successfully on successful verification of stable PUF responses. Authentication is carried out by way of lightweight commitment protocol, message authentication protocol and optional zero-knowledge proof guarantees privacy. Experimental analysis shows that PUFaaS can be evaluated as having low false acceptance and rejection, high modeling attack resistance, and scale efficiently (appropriate to large-scale IoT and cloud-edge). PUFaaS will offer an effective, privacy-resilient, and scalable solution to distributed authentication and secure information exchange in a non-trusted environment by converting PUFs into a service-oriented architecture.
Physical Unclonable Functions (PUFs) and Hardware Security
Background: Mental health accounts for an estimated 14% of the global disease burden yet receives less than 2% of health budgets in most countries, with even lower investment in low- and middle-income settings. This study examines federal mental health financing trends from 2021-2025 to assess whether legislative reform translated into fiscal prioritization. Methods: A mixed-methods policy analysis was conducted, combining quantitative analysis of federal budget appropriation documents (2021-2025) with qualitative documentary review and comparative case studies. Mental health allocations were assessed by recurrent and capital expenditure, institutional distribution, and proportional share of total federal health spending. WHO reports, national policy documents from Ghana and Kenya, and peer-reviewed literature informed comparative analysis. Results: Federal mental health allocations increased from ₦23.33 billion in 2021 to ₦88.24 billion in 2025, a 278% nominal rise. However, the sector’s share of the total health budget declined from 3.67% to 3.12%, indicating relative marginalization. Over 90% of funding supported recurrent expenditures in ten federal neuropsychiatric hospitals, with minimal investment in community-based services or primary care integration. In contrast, Ghana and Kenya more effectively leveraged legislation, fiscal decentralization, and insurance mechanisms to expand access. Conclusion: Despite legislative reform, Nigeria’s mental health financing remains centralized, hospital-focused, and misaligned with population needs. Institutional inertia, weak coordination, and delayed implementation of the Act have constrained equitable scale-up. Activating the Mental Health Fund and integrating mental health into national financing mechanisms are urgently required to prevent deepening inequities.
Amid intensifying challenges of global climate change, China—as the world’s largest carbon emitter and a major manufacturing hub—occupies a pivotal position in the global industrial green transformation. Drawing on environmental federalism theory and China’s decentralized governance model, this study develops a framework of “green finance–local government competition–industrial green transformation.” Using panel data from 283 cities in China, we employ spatial econometrics and mediation effect models to test the dual mechanisms by which green finance promotes industrial green transformation. The findings indicate that (1) green finance promotes industrial green transformation; (2) green finance advances industrial green transformation by dismantling China’s traditional local government competition–based development model and removing the institutional suppression arising from “race-to-the-bottom competition”; (3) the effect of green finance exhibits long-run characteristics and a “benchmark–imitation” pattern; (4) baseline environmental conditions strengthen the influence of green finance on industrial green transformation; (5) incorporating ecological civilization development into officials’ performance evaluations can effectively reshape policy incentives and amplify the positive role of green finance. Thus, we propose differentiated green finance policies, the construction of a governance mechanism that integrates fiscal–financial–ecological compensation, and the optimization of ecological civilization assessment indicators to curb campaign-style governance.
Alzheimer’s disease (AD) is a chronic neurodegenerative disorder profoundly affecting memory and cognitive functions for which an early and precise diagnosis is essential to achieve timely intervention and disease management. Magnetic Resonance Imaging (MRI) is an important tool for detecting structural changes in the brain such as hippocampal shrinkage and ventricular enlargement, which can be correlated with Alzheimer’s disease’s stages of progression. In this work, we present a framework that couples deep learning-based Alzheimer’s MRI classification with blockchain-supported image authenticity verification. Our experimental setup compares five classification approaches, Xception, Long Short-Term Memory (LSTM) networks, ResNet50, Random Forest, and Gradient Boosting across different training durations. The best performing model is integrated to the local IPFS node and Ethereum smart contract through Ganache. This comparative investigation examines the balance of accuracy, efficiency, and training time in a variety of model designs. It also illustrates the viability of a secure, decentralized framework for both diagnostic accuracy and data integrity through blockchain.
The withdrawal of the USA from the World Health Organization and the freeze on USAID are among the major events in the realm of U.S. foreign policy under the U.S. president. Within his broader “America First” policy, aimed at reducing the U.S.‘s international commitments and rethinking its role in global organizations and foreign aid, this review attempts to make a case for Africa by examining the implications of recent reductions in U.S. funding. We conducted a comparative case study of Nigeria, Ghana, Zambia, and Rwanda, selected for their aid volume, exposure to disruption events, and availability of outcome data. Using process tracing and critical narrative synthesis, we analyzed policy documents, expenditure reports and peer-reviewed studies to assess how each country responded to aid disruptions and what structural factors shaped their resilience or fragility. Three dominant patterns emerged: acute service interruptions (Nigeria, Zambia), structural fragmentation (Ghana), and resilient adaptation (Rwanda). Key drivers of vulnerability included overreliance on tied aid, SAP-era health system legacies, and underdeveloped domestic financing mechanisms. Rwanda’s ability to maintain high ART coverage and reduce malaria deaths by 88% during funding cuts reflects a deliberate break from aid dependency through community-based insurance, decentralized governance, and regional procurement strategies. Donor transitions are not neutral events; they expose and exacerbate pre-existing structural weaknesses. Current models that frame aid withdrawal as empowerment risk, replicating past harm unless coupled with institutional reform and reciprocal accountability. This study suggests assessing transition readiness and reorienting global health partnerships toward equitable, resilient, and sovereign systems.
Frequent digitalization of power systems has increased the areas of cyber attack on smart grids that require more developed security frameworks capable of responding to cyber and physical threats. The paper is a detailed discussion on the implementation of Artificial Intelligence (AI) and Block chain in improving the cyber security of smart grids. To prevent new cyber threats, we suggest a new hybrid security model, which uses AI-based anomaly detection and decentralized integrity checks, based on block chains. The architecture is a fusion of machine learning based real time threat detection and immutable block chain ledgers to keep transactions safe and under control. We prove with the help of the performance evaluation and case studies that the model is efficient in the detection of false data injection, distributed denial of service (DDOS) attacks, and other advanced threats. The paper also cites major issues such as the scalability, interoperability, and computational overhead, as well as outlining opportunities of future research in resilient smart grid infrastructure. We have found out that the AI-block chain implementation has the potential to enhance the accuracy of threat detection up to 30 percent relative to traditional approaches as well as integrity and transparency of data management in smart grids.
This paper presents zero knowledge proofs, their cryptographic significance and applications. It presents a basic classification: interactive and noninteractive zero knowledge proofs. It presents and compares three protocols of non-interactive zero knowledge proofs: ZK-SNARK, ZK-STARK and Bulletproofs. It presents the quadratic residue problem and proofs it with both interactive and non-interactive zero knowledge proofs. The non-interactive protocol used to prove the quadratic residue problem is ZK-SNARK. The proof is implemented in the Python programming language, using python-snark library.
The digital transformation of finance and accounting is accelerating with AI, blockchain, and automation, reshaping financial operations, auditing, and compliance. This study conducts a thematic analysis of academic literature (2018–2025) and industry reports from PwC, Deloitte, EY, HSBC, and central banks to examine key trends. Six themes emerged: automation and efficiency, security and fraud prevention, decentralization, financial inclusion, regulatory challenges, and adoption barriers. Findings show that AI and RPA enhance financial reporting and fraud detection, while blockchain improves transparency and security but poses scalability and regulatory challenges. Decentralized finance (DeFi) and digital currencies like JPM Coin and the Digital Yuan are transforming transactions but raise concerns over compliance and illicit activity risks. Mobile banking and blockchain-based solutions improve financial inclusion, yet digital literacy and security risks remain barriers. Using NVivo-based thematic analysis, the study identifies key trends shaping the future of financial digitalization. While AI and blockchain drive efficiency, regulatory complexities and adoption barriers must be addressed for sustainable transformation. Future research should explore scalability, AI-enhanced compliance, and blockchain’s role in financial security.
Large Language Models (LLMs) have demonstrated significant potential in smart contract auditing. However, they are still susceptible to hallucinations and limited context-aware reasoning. In this article, we propose SmartAuditFlow, a dynamic Plan-Execute framework that customizes audit strategies based on the unique characteristics of each smart contract. Unlike static, rule-based workflows, our approach iteratively generates and refines audit plans in response to intermediate outputs and newly detected vulnerabilities. To improve reliability, the framework incorporates structured reasoning, prompt optimization, and external tools such as static analyzers and Retrieval-Augmented Generation (RAG). This multi-layered design reduces false positives and enhances the accuracy of vulnerability detection. Experimental results show that SmartAuditFlow achieves 100% accuracy on common vulnerability benchmarks and successfully identifies 13 additional CVEs missed by existing methods. These findings underscore the framework’s adaptability, precision, and practical utility as a robust solution for automated smart contract security auditing. The source code is available at: https://github.com/JimmyLin-afk/SmartAuditFlow .
The rise in Bitcoin and Ethereum is well-known. Research shows that public sentiment greatly affects their price changes. Thus, public sentiment analysis is a key factor in making investment decisions. This study analyzes public sentiment towards Bitcoin and Ethereum on platform X using Graph Neural Networks (GNN), specifically the GCN and AGN-TSA models. GCN is utilized for its ability to capture syntactic relationships between words, while AGN-TSA integrates textual content with user-level social interactions through an attention mechanism. The dataset is collected from$X$using the keywords Bitcoin, BTC, Ethereum, and ETH, followed by preprocessing and labeling based on FinBERT and Vader as a benchmark labeling technique to construct the training and test sets. Evaluation employs a confusion matrix to compare model performance. The results show that GCN without addressing class imbalance achieves 83.89 % accuracy, whereas AGN-TSA achieves 86.21%. However, confusion matrix analysis revealed severe bias toward the majority class, so we needed to use extreme class weighting (ratio 17:1:10), which improved minority-class recall. However, it caused training instability and reduced accuracy: GCN dropped to 80.58 %, and AGN-TSA dropped to 86.15% (a decrease of$0.06 \%)$. Despite the decrease, AGN-TSA still achieves the best accuracy compared to GCN, this result reinforcing our initial hypothesis that attention-based graph modeling which leverages social ties yields superior performance for sentiment classification in crypto related discourse. Furthermore, the accuracy of the data labeling technique and the imbalance in the label distribution also affect the final accuracy results.
The article presents an empirical comparison of three contemporary Layer-2 scaling solutions for the Ethereum blockchain: Scroll, Linea, and Base, representing zk-rollup and optimistic rollup architectures. The study aims to evaluate the transaction processing speed and stability of selected Layer-2 networks using real-time data collected from blockchain explorers (Blockscout, Lineascan, Basescan). The dataset comprises 45,000 transactions processed in October 2025 and aggregated at one-second resolution (1 Hz). Statistical analyses include ANOVA, Kruskal–Wallis, Levene, and Brown–Forsythe tests, as well as ADF and KPSS stationarity diagnostics, used to assess diferences in throughput and operational stability across the examined networks. The results indicate that the Base network achieves the highest mean throughput (≈ 102 TPS) and the lowest temporal volatility, whereas Linea and Scroll exhibit non-stationary, highly variable transaction dynamics driven by periodic batching. The fndings confrm the persistence of the scalability trilemma—where improvements in performance may come at the cost of higher centralization and operational dependency. This research contributes to the quantitative assessment of rollup efciency and provides a reference point for further empirical studies on blockchain scalability.
Saiful Ruchiyat Cosahan, Ahmad Yunani, Asrid Juniar, Muzdalifah Muzdalifah
This Systematic Literature Review (SLR) analyzes 38 empirical studies published between 2015 and 2025 (sourced from Scopus and Sci-ScienceDirect) to map blockchain-based funding mechanisms in the context of venture capital (VC) and entrepreneurial finance. The review addresses four research questions concerning the evolution of these mechanisms, their impact on startup performance, and associated risks and regulatory challenges. The findings establish a robust taxonomy of mechanisms, including Initial Coin Offerings (ICOs), Security Token Offerings (STOs), and Decentralized Autonomous Organizations (DAOs), each presenting unique features and regulatory profiles. Crucially, the review highlights significant gaps in long-term performance data, revealing challenges related to investor protection, fraud risk, and regulatory uncertainty. By integrating Signaling Theory and Governance Theory, the study discusses how tokenomics and team credibility function as signals instead of traditional VC due diligence, presenting a critical comparison between token-based funding and traditional-al venture capital financing. This paper offers valuable insights for academics, policymakers, and industry practitioners by providing a com-comprehensive map of the field, suggesting avenues for future empirical research, and offering focused policy implications regarding regulation and investor safety in emerging markets.
Identity management is a critical component in egovernance, ensuring secure, reliable, and efficient verification of citizens' identities. With increasing digitization, protecting personal data while enabling seamless access to government services has become essential. Existing identity management systems often rely on centralized databases, which are prone to data breaches, unauthorized access, and lack of transparency, raising concerns over privacy and trust. To address these challenges, this research proposes a Blockchain Identity Framework that integrates Zero-Knowledge Proof (BIF-ZKP) authentication with blockchain consensus mechanisms. In this framework, ZKP enables users to prove their identity without revealing sensitive information, while blockchain ensures that identity records are decentralized, tamperproof, and auditable. The consensus mechanism guarantees that all identity transactions are verified by multiple nodes, reducing the risk of fraud and unauthorized modifications. The proposed method is applied in an e-governance context to securely manage citizens' digital identities, enabling authentication for services such as online voting, tax filings, and social welfare schemes while maintaining privacy. Experimental evaluation demonstrates that the BIF significantly enhances data security, privacy preservation, and trustworthiness compared to traditional centralized identity systems. It reduces the risk of identity fraud and ensures the verifiable and transparent management of citizens' information. The proposed method improves data security by 96.2 % and reduces fraud by 89 %.
In recent years, blockchain technologies such as Ethereum have secured widespread adoption, yet they have also become increasingly targeted by fraudulent activities. Discovering these fraudulent patterns is challenging due to the complexity and size of transaction data. This study investigates the application of the XGBoost algorithm, a gradient boosting technique optimized for performance plus scalability, in discovering fraudulent transactions on the Ethereum network. The model is instructed to distinguish between valid as well as suspicious behaviour based on transactional and behavioural characteristics by examining a dataset of Ethereum transaction records. XGBoost is a popular machine learning algorithm used for a range of tasks, including fraud detection on Ethereum and other blockchain networks. It is a highly effective model due to its performance, flexibility, and ability to handle complex, imbalanced, as well as large datasets. This study emphasises accuracy and precision, also recalling key metrics, demonstrating that XGBoost not only enhances prediction performance but also minimizes false positives over other classification algorithms. Our findings: XGBoost is the optimal model for real-time fraudulent detection in a blockchain context. The accuracy of the XGBOOST algorithm achieves a high level of accuracy.
The thesis deals with the development of a decentralized Ethereum-based application for purchasing, selling and playing music. The goal of the application is to demonstrate the use of a blockchain-based platform that can replace corporate intermediaries.
Savings and loan cooperatives play an essential role in promoting financial inclusion and supporting Indonesia's local economy by providing affordable credit and encouraging community-based savings. However, many cooperatives still depend on manual or semi-digital procedures for credit approval, resulting in inefficiencies, delayed loan processing, human errors, and limited transparency. These weaknesses often lead to mismatched capital-to-loan ratios, data inconsistencies, and reduced member trust in cooperative governance. To address these challenges, this study proposes a blockchain-based smart contract framework that automates the credit approval process through secure, rule-based decisionmaking. The research employs the Design Science Research Methodology (DSRM) to design, implement, and evaluate a prototype system developed using Solidity on the Ethereum blockchain, integrated with Web3.js and Metamask for decentralized interactions. The smart contract encodes cooperative business rules, automatically verifies member eligibility, and records transactions immutably on the blockchain ledger. A case study conducted in an Indonesian cooperative demonstrates that the proposed system reduces credit approval time from several days to a few seconds, eliminates manual verification errors, and achieves 100 % transaction success and data consistency. The findings highlight the potential of blockchain and smart contracts to enhance operational efficiency, transparency, and trust in cooperative finance, contributing to Indonesia's digital transformation and offering a scalable model for other community-based financial institutions.
Historical Context and Problem Statement The digital revolution has created two parallel challenges that have resisted comprehensive solutions: Internet Data Transfer Limitations: Despite decades of progress, internet download speeds remain constrained by inefficient protocols that don't adapt to network topology dynamics. Traditional download managers like IDM operate with static segmentation strategies that ignore the quantum-inspired probabilistic nature of network paths. Web3 Liquidity Fragmentation: Decentralized finance (DeFi) suffers from fragmented liquidity across multiple venues, resulting in significant MEV exploitation. As documented by Qin et al. (2021), MEV extraction has cost users over $680 million in 2021 alone, with no comprehensive solution addressing the root cause. These seemingly disconnected problems share a common underlying structure: both involve the transfer of "value" (data or financial assets) across complex networks where efficiency is hampered by non-resonant transmission strategies.
Hau Vasio Sarmento Soares, Sundaru Guntur Wibowo, Syahrul Anwar
This study analyzes the legal framework of Non-Fungible Token (NFT)-based digital vaccine certificates in the context of digital free trade, focusing on security, privacy, and international recognition. Using normative and comparative legal research methods with a multidisciplinary approach, the study integrates perspectives from law, digital technology, and international policy. The study examines three main aspects: first, security, evaluating how NFTs ensure authenticity, data integrity, and protection against manipulation through encryption, blockchain, and smart contracts; second, privacy, analyzing how personal data and the privacy rights of certificate holders are protected under national and international regulations, emphasizing data minimization, user consent, and secure access; and third, international recognition, assessing the extent to which NFT-based certificates can be recognized globally, highlighting regulatory harmonization and legal barriers. The findings indicate that NFT-based vaccine certificates provide strong technical security and privacy protection, but legal recognition across jurisdictions remains inconsistent. The study concludes that while NFTs have significant potential to facilitate secure and verifiable digital health credentials in global trade, harmonization of national and international regulations and the implementation of legal standards are crucial to ensure their effectiveness and legal validity worldwide.
The development of digital technology has enabled the emergence of Non-Fungible Tokens (NFTs) as digitally authenticated assets recorded on blockchain networks, including their use in representing ownership of digital land within metaverse ecosystems. However, Indonesia has not yet formulated explicit regulatory provisions governing the legal classification of NFTs, the scope of supervisory authority, or the standards for consumer protection. This regulatory absence results in legal ambiguity regarding the placement of NFTs within the framework of Financial Sector Technology Innovation (ITSK) under Law No. 4 of 2023, and simultaneously presents risks to consumers, including fraud, data misuse, loss of access to digital assets, and a lack of clear accountability mechanisms on NFT platforms. This research examines the legal status of NFTs in relation to ITSK and analyzes the adequacy of current consumer protection measures in NFT-based digital land transactions. Through a normative juridical method, the study finds that NFTs have not been formally classified within ITSK, nor assigned to a definitive supervisory authority, whether OJK or Bappebti. As such, consumer protection remains reliant on general norms under Law No. 8 of 1999, which are insufficient to address the specific risks inherent in NFT transactions. This research recommends the issuance of derivative regulations by OJK and/or Bappebti to clarify NFT classification, establish platform obligations, and strengthen consumer protection.
ABSTRACT Phishing attacks in decentralized Web3 systems continue to evolve beyond the detection capabilities of traditional Web2 security models. Existing decentralized authentication systems typically lack either mutual verification or dynamic threat awareness. We present PhishGuard++, a cross‐chain, privacy‐preserving authentication framework that introduces two core innovations: (1) a novel mutual Zero‐Knowledge Proof (ZKP) protocol that validates both users and services using Decentralized Identifiers (DIDs), and (2) a real‐time, on‐chain Graph Neural Network (GNN) threat oracle that assigns phishing risk scores integrated directly into smart contract‐based access control logic. A stake‐based validator reputation system with anti‐collusion incentives further reinforces trust without sacrificing decentralization or privacy. Experimental results on a simulated Arbitrum testnet show a statistically significant 40.4% reduction in phishing success rate across five attack vectors, 98.6% authentication accuracy, and sub‐second latency with gas‐efficient operations. Unlike prior works that independently apply ZKPs, DIDs, or GNNs, this framework offers the first privacy‐preserving, mutual authentication system that combines these technologies with stake‐based economic enforcement and real‐time smart contract enforcement. The novelty lies in the architecture's real‐time threat‐aware access decisions, validator‐linked risk accountability, and practical cross‐chain deployment—an integration not previously achieved.
The web3 applications have recently been growing, especially on the Ethereum platform, starting to become the target of scammers. The web3 scams, imitating the services provided by legitimate platforms, mimic regular activity to deceive users. However, previous studies have primarily concentrated on de-anonymization and phishing nodes, neglecting the distinctive features of web3 scams. Moreover, the current phishing account detection tools utilize graph learning or sampling algorithms to obtain graph features. However, large-scale transaction networks with temporal attributes conform to a power-law distribution, posing challenges in detecting web3 scams. To overcome these challenges, we present ScamSweeper, a novel framework that emphasizes the dynamic evolution of transaction graphs, to identify web3 scams on Ethereum. ScamSweeper samples the network with a structure temporal random walk, which is an optimized sample walking method that considers both temporal attributes and structural information. Then, the directed graph encoder generates the features of each subgraph during different temporal intervals, sorting as a sequence. Moreover, a variational Transformer is utilized to extract the dynamic evolution in the subgraph sequence. Furthermore, we collect a large-scale transaction dataset consisting of web3 scams, phishing, and normal accounts, which are from the first 18 million block heights on Ethereum. Subsequently, we comprehensively analyze the distinctions in various attributes, including nodes, edges, and degree distribution. Our experiments indicate that ScamSweeper outperforms SIEGE, Ethident, and PDTGA in detecting web3 scams, achieving a weighted F1-score improvement of at least 17.29% with the base value of 0.59. In addition, ScamSweeper in phishing node detection achieves at least a 17.5% improvement over DGTSG and BERT4ETH in F1-score from 0.80.
We introduce the Overlap-Weighted Hierarchical Normalized Persistence Velocity (OW-HNPV), a novel topological data analysis method for detecting anomalies in time-varying networks. Unlike existing methods that measure cumulative topological presence, we introduce the first velocity-based perspective on persistence diagrams, measuring the rate at which features appear and disappear, automatically downweighting noise through overlap-based weighting. We also prove that OW-HNPV is mathematically stable. It behaves in a controlled, predictable way, even when comparing persistence diagrams from networks with different feature types. Applied to Ethereum transaction networks (May 2017-May 2018), OW-HNPV demonstrates superior performance for cryptocurrency anomaly detection, achieving up to 10.4% AUC gain over baseline models for 7-day price movement predictions. Compared with established methods, including Vector of Averaged Bettis (VAB), persistence landscapes, and persistence images, velocity-based summaries excel at medium- to long-range forecasting (4-7 days), with OW-HNPV providing the most consistent and stable performance across prediction horizons. Our results show that modeling topological velocity is crucial for detecting structural anomalies in dynamic networks.
The rapid expansion of AI-driven applications powered by large language models has led to a surge in AI interaction data, raising urgent challenges in security, accountability, and risk traceability. This paper presents AiAuditTrack (AAT), a blockchain-based framework for AI usage traffic recording and governance. AAT leverages decentralized identity (DID) and verifiable credentials (VC) to establish trusted and identifiable AI entities, and records inter-entity interaction trajectories on-chain to enable cross-system supervision and auditing. AI entities are modeled as nodes in a dynamic interaction graph, where edges represent time-specific behavioral trajectories. Based on this model, a risk diffusion algorithm is proposed to trace the origin of risky behaviors and propagate early warnings across involved entities. System performance is evaluated using blockchain Transactions Per Second (TPS) metrics, demonstrating the feasibility and stability of AAT under large-scale interaction recording. AAT provides a scalable and verifiable solution for AI auditing, risk management, and responsibility attribution in complex multi-agent environments.