Linh Tran, Sanjay Chari, Md. Saikat Islam Khan, Aaron Zachariah · 6 authors
We present the Differentially Private Blockchain-Based Vertical Federal Learning (DP-BBVFL) algorithm that provides verifiability and privacy guarantees for decentralized applications. DP-BBVFL uses a smart contract to aggregate the feature representations, i.e., the embeddings, from clients transparently. We apply local differential privacy to provide privacy for embeddings stored on a blockchain, hence protecting the original data. We provide the first prototype application of differential privacy with blockchain for vertical federated learning. Our experiments with medical data show that DP-BBVFL achieves high accuracy with a tradeoff in training time due to on-chain aggregation. This innovative fusion of differential privacy and blockchain technology in DP-BBVFL could herald a new era of collaborative and trustworthy machine learning applications across several decentralized application domains.
AI data sharing platforms must reconcile two pressures that often clash: the need to exchange high‐value datasets for model development and evaluation, and the obligation to guarantee privacy, integrity, and verifiability of computations on that data. This manuscript surveys and synthesizes cryptographic building blocks—differential privacy, homomorphic encryption, multiparty computation with secure aggregation, zero‐knowledge proofs, attribute-based encryption and proxy re-encryption, trusted execution environments, and domain standards such as Crypt4GH—into a pragmatic, layered architecture for AI data sharing. We outline a methodology that integrates policy-aware access control with threshold key management, private training and inference, verifiable analytics, and auditability. A compact statistical analysis (with an illustrative table) demonstrates how such a stack can bound leakage (ε), preserve utility (accuracy), and manage computational overhead (latency). Results show that a hybrid PETs (privacy-enhancing technologies) approach—combining local differential privacy and secure aggregation for ingestion, homomorphic encryption or TEEs for computation, and zk-proofs for verifiability—achieves strong privacy with modest accuracy loss and acceptable latency for many enterprise scenarios. We conclude with design guidelines and research directions for standards-aligned, future-ready AI data sharing platforms.
Lorenzo Cassano, Jacopo D’Abramo, Siraj Munir, Stefano Ferretti
In this paper, we present a study of a Federated Learning (FL) system, based on the use of decentralized architectures to ensure trust and increase reliability. The system is based on the idea that the FL collaborators upload the (ciphered) model parameters on the Inter-Planetary File System (IPFS) and interact with a dedicated smart contract to track their behavior. Thank to this smart contract, the phases of parameter updates are managed efficiently, thereby strengthening data security. We have carried out an experimental study that exploits two different methods of weight aggregation, i.e., a classic averaging scheme and a federated proximal aggregation. The results confirm the feasibility of the proposal.
With the extensive deployment of vehicular ad-hoc networks (VANETs), it becomes an inevitable choice to provide enhanced in-vehicle services for uploading a vast amount of shared vehicular data to cloud storage. However, there is still a lack of effective deduplication and audit methods for cloud-stored data in VANET scenarios. To address the securities of cloud-stored data in VANETs, we propose a blockchain-based data deduplication and distributed audit scheme for shared data under cloud-fog computing-based VANETs in this paper. In our scheme, we construct a distributed audit model for VANETs, where road side units (RSUs) are partitioned as multiple management areas. Each management area can solely make their consensus for data integrity verification to audit the cloud storage provider without depending on any third-party auditors (TPAs). Also, we establish a blockchain-based monitoring mechanism maintained by the fog servers to ensure the integrity of the uploading and auditing records and enable related entities within the system to verify corresponding audit results (or records). Furthermore, we propose a lightweight dual-verifier structure to adapt to resource-constrained VANET scenarios. Through our dual-verifier mechanism, our scheme can effectively resist proof-replay attacks. Related theoretical analysis and experimental results show our data deduplication and distributed audit scheme is efficient and effective for VANET scenarios.
To be useful and widely accepted, automated contact tracing schemes (also called exposure notification) need to solve two seemingly contradictory problems at the same time: they need to protect the anonymity of honest users while also preventing malicious users from creating false alarms. In this paper, we provide, for the first time, an exposure notification construction that guarantees the same levels of privacy and integrity as existing schemes but with a fully malicious database (notably similar to Auerbach et al. CT-RSA 2021) without special restrictions on the adversary. We construct a new definition so that we can formally prove our construction secure. Our definition ensures the following integrity guarantees: no malicious user can cause exposure warnings in two locations at the same time and that any uploaded exposure notifications must be recent and not previously uploaded. Our construction is efficient, requiring only a single message to be broadcast at contact time no matter how many recipients are nearby. To notify contacts of potential infection, an infected user uploads data with size linear in the number of notifications, similar to other schemes. Linear upload complexity is not trivial with our assumptions and guarantees (a naive scheme would be quadratic). This linear complexity is achieved with a new primitive: zero knowledge subset proofs over commitments which is used by our "no cloning" proof protocol. We also introduce another new primitive: set commitments on equivalence classes, which makes each step of our construction more efficient. Both of these new primitives are of independent interest.
Farhana Javed, Josep Mangues‐Bafalluy, Engin Zeydan, Luis Blanco
In the domain of Collaborative Artificial Intelligence, Federated Learning ($\mathbf{F L}$) is a technique that enables multiple entities to collaboratively refine AI models while adhering to stringent data privacy standards, without the need for direct data sharing. This paper explores the integration of blockchain technology with FL to establish reliable trust mechanisms within this collaborative framework. We highlight and review current blockchain-enabled reputation mechanisms that evaluate the reliability and quality of contributions from participants, which are crucial for maintaining trust and operational integrity in distributed settings. Through our review, we address the concept and implementation challenges. Additionally, we discuss recent technological advances and explore the emerging opportunities that blockchain presents to address trust-related challenges in FL, emphasizing significant prospects for future research directions, such as decentralized identities, zero trust, and zero-knowledge proofs to enhance trust in these environments.
Blockchain’s decentralization, transparency, and tamper‐resistance are celebrated properties for auditability and trust, yet they collide with core data protection duties under the EU General Data Protection Regulation (GDPR). This manuscript analyzes the principal compliance challenges that arise when blockchain processes personal data and proposes a practical, design-oriented framework to address them. First, we synthesize legal and regulatory positions on what counts as “personal data,” the difference between anonymization and pseudonymization, and the implications of the right to erasure, data protection by design and by default, allocation of controller/processor roles, and international data transfers. We then map these requirements to blockchain architectures (public permissionless, public permissioned, and private permissioned) and data patterns (on-chain, off-chain, hybrid). Building on recent guidance from the European Data Protection Board (EDPB) and national authorities, we outline concrete technical and governance controls—off-chain storage and on-chain commitments, keyed hashing, encryption/key-revocation strategies, chameleon-hash/redactable-ledger designs, selective-disclosure credentials/zero-knowledge proofs, and robust consortium governance—to reduce risk and improve demonstrable compliance. Applying a six-step assessment methodology to three realistic use cases (NFT profile registry, supply-chain provenance, and consortium KYC), we show that while no single pattern fully reconciles immutability with erasure, practicable combinations can align processing with GDPR’s principles of minimization, purpose limitation, storage limitation, and accountability. The paper concludes with a prioritized checklist for engineering “compliance-by-design” blockchains, and delineates scope and limitations for practitioners and researchers.
Smart contracts are essential applications for blockchains, which have been used in a wide variety of fields and are handling large amounts of valuable assets. Once deployed on the blockchain network, smart contracts cannot be altered, thus making the pre-deployment testing of them extremely critical. Nevertheless, the features of smart contracts, especially their transaction-driven nature, pose huge challenges to testing. In particular, since their inputs are not static data but dynamic sequences of user behavior, it is very difficult to obtain a feasible oracle for testing, which refers to the systematic mechanism to verify the correctness of test results given any test input. This is the notorious oracle problem in the context of software testing, for which the metamorphic testing technique has been widely recognized as a simple yet effective solution. In this paper, we develop a comprehensive framework, namely MT4SC, for implementing metamorphic testing on smart contracts. Specifically, we propose a systematic way to construct metamorphic relations, the core component of metamorphic testing, based on the user behavior sequences. A series of experiments have been conducted to evaluate the performance of MT4SC on eight different smart contract scenarios. The experimental results demonstrate MT4SC’s high effectiveness in detecting potential faults in smart contracts, even without the need for test oracles. This study bolsters the research on the testing of smart contracts, thereby improving their quality and ultimately advancing the reliability of blockchains.
Haibo Wang, Hongwei Gao, Teng Ma, Chong Li · 5 authors
Distributed Federated Learning (DFL) technology enables participants to cooperatively train a shared model while preserving the privacy of their local data sets, making it a desirable solution for decentralized and privacy-preserving Web3 scenarios. However, DFL faces incentive and security challenges in the decentralized framework. To address these issues, this paper presents a Hierarchical Blockchain-enabled DFL (HBDFL) system, which provides a generic solution framework for the DFL-related applications. The proposed system consists of four major components, including a model contribution-based reward mechanism, a Proof of Elapsed Time and Accuracy (PoETA) consensus algorithm, a Distributed Reputation-based Verification Mechanism (DRTM) and an Accuracy-Dependent Throughput Management (ADTM) mechanism. The model contribution-based rewarding mechanism incentivizes network nodes to train models with their local datasets, while the PoETA consensus algorithm optimizes the tradeoff between the shared model accuracy and system throughput. The DRTM improves the system efficiency in consensus, and the ADTM mechanism guarantees that the throughput performance remains within a predefined range while improving the shared model accuracy. The performance of the proposed HBDFL system is evaluated by numerical simulations, which show that the system improves the accuracy of the shared model while maintaining high throughput and ensuring security.
Bei Chen, Gaolei Li, Xi Lin, Zheng Wang · 5 authors
Recent advancements in multi-agent systems based on large language models (LLM) have shown potential for problem-solving and planning tasks. However, most existing LLM-based multi-agent approaches show vulnerability against byzantine attacks. First, agents instantiated on diverse LLMs may inherit biases present in the LLMs and thus exhibit deception behavior. Second, as the number of agents grows, collusive behavior among multiple malicious agents poses a potential threat. In this paper, we propose BlockAgents, an innovative framework that integrates blockchain into LLM-based cooperative multi-agent systems to mitigate byzantine behaviors. BlockAgents completes multi-agent collaboration through a unified workflow including role assignment, proposal statement, evaluation, and decision-making. To help the agent who contributes the most to the group thinking process acquire accounting rights, we propose a proof-of-thought (PoT) consensus mechanism combined with stake-based miner designation and multi-round debate-style voting. To effectively distinguish valid and abnormal answers, we design a multi-metric prompt-based evaluation method for each evaluator to score each proposal by carefully and comprehensively considering multiple dimensions. Experiments on three datasets show that BlockAgents reduces the interference of poisoning attacks on accuracy to less than 3% and reduces the success rate of backdoor attacks to less than 5%, demonstrating the resistance ability against Byzantine attacks.
Artificial intelligence (AI) and distributed ledger technologies are increasingly integrated into public and private surveillance infrastructures—from city-wide camera networks to critical-infrastructure monitoring and access control. This integration promises higher integrity and accountability through immutable logs, faster incident response via on-device inference, and interoperable audit trails across organizations. Yet it also amplifies ethical risks: mass data collection, opacity in model decisions, function creep, demographic harms, cross-border data governance conflicts, and accountability gaps when immutable records meet “right to erasure” regimes. This manuscript proposes an ethics-by-design reference architecture for blockchain-powered surveillance that embeds privacy, proportionality, and fairness controls into each lifecycle stage (purpose definition → data capture → model training → inference → access → audit → decommissioning). Technically, it composes privacy-enhancing technologies (PETs)—including differential privacy, federated learning, zero-knowledge proofs, verifiable credentials (VCs), and content-provenance standards (C2PA)—with permissioned blockchain ledgers, model cards, and risk management aligned to the NIST AI RMF, ISO/IEC 23894, ISO/IEC 42001, UNESCO, and ACM guidance. A simulated evaluation illustrates how the architecture can reduce false-positive disparities and unauthorized access, while preserving evidentiary integrity. We discuss tensions with GDPR (e.g., Article 17 erasure; DPIA obligations), constraints introduced by the EU AI Act (e.g., prohibitions and high-risk biometric uses), and strategies to reconcile immutability with privacy (e.g., off-chain storage with revocation, redaction-friendly commitments). The paper closes with limitations and a future research agenda for measurable, auditable ethical guarantees in real-time surveillance.
Yonghua Zhan, Yang Yang, Hongju Cheng, Xiangyang Luo · 6 authors
More vehicles are connecting to the Internet of Things (IoT), transforming Vehicle Ad hoc Networks (VANETs) into the Internet of Vehicles (IoV), providing a more environmentally friendly and safer driving experience. Vehicular announcement networks show promise in vehicular communication applications. However, two major issues arise when establishing such a system. First, user privacy cannot be guaranteed when messages are forwarded anonymously, thus the reliability of these messages is in question. Second, users often lack interest in responding to announcements. To address these problems, we introduce a Blockchain-based incentive announcement system called PIAS. This system enables anonymous message commitment in a semi-trusted environment and encourages witnesses to respond to requests for traffic information. Additionally, PIAS uses blockchain accounts as identities to participate in the system with incentives, ensuring privacy in anonymous announcements. PIAS successfully protects the privacy of participants and motivates witnesses to respond to requests. Furthermore, our assessment of security and compatibility shows that PIAS can maintain privacy and incentivization while being compatible with both the Bitcoin and Ethereum blockchains. Further evaluation has confirmed the system's efficiency in terms of performance.
S. Gopalakrishnan, E. D. Kanmani Ruby, D. Hemanand, R. Anitha · 6 authors
The incorporation or combination of Artificial Intelligence (AI) and blockchain technology into Mobile Ad Hoc Networks (MANETs) shows important factor for modern and advance smart city infrastructure and autonomous vehicular networks. This paper describes the complementary potential of the technologies to help the built-in difficulties of MANETs includes flexibility, protection, and data integrity. AI techniques such as machine learning and reinforcement learning, are emphasized to improve routing protocols to optimize data transmission rates, and decrease latency. Blockchain technology using Practical Byzantine Fault Tolerance (PBFT) and other consensus mechanisms, gives a tight and decentralized architecture for data handling assuring trust and integrity amidst network nodes. The appeal of these incorpoarted technologies is especially related for smart cities which depand on collection of data and evaluation for effective handling of urban operations such as flow of traffic, environmental observing, and consumption of energy. Autonomous vehicular networks needing rigd and strong communication and data transfer between vehicles and infrastructure, also help from the enhanced network functions and security provided by AI and blockchain incorpoaration. Experimental evaluation denotes improvements in crucial performance metrics. Sensor 2 persists the highest data transmission rate of 12 Mbps. Sensor 4 had the decreased at 9 Mbps. Latency measurements observed that Sensor 2 recorded the lowest latency at 45 ms, with Sensor 3 having the highest at 55 ms.
In this article, we propose zero-knowledge named proof, a stateless replay attack prevention strategy that ensures the user’s anonymity against malicious administrators. We begin with adopting the zero-knowledge set-membership proof into an authentication setting in which users would delegate their requests to an agent that obstructs the user’s identity from the administrator. This anonymous agent carries the guarantee of authenticity, which the administrator through the set-membership proof can confirm. Next, we prevent replay attacks from other parties by binding the agent’s identity to the authentication proof verifiable by the administrators. By leveraging these properties, a scalable blockchain-based authentication scheme is then built. We quantitatively evaluate the security and measure the time and monetary cost of our scheme under both ideal and realistic environments. On top of it, we provide a third-party authorization scheme derived from our authentication framework to demonstrate its real-world applicability.
Guy Zyskind, Yonatan Erez, Tom Langer, Itzik Grossman · 5 authors
Blockchains ensure that all transactions, including those that execute deterministic programs known as smart contracts, are processed correctly and without interruption. However, blockchains inherently provide no confidentiality - all transaction data, including inputs sent to smart contracts, are public. This has led to a rise of confidential smart contract blockchains. These blockchains utilize privacy-preserving techniques to add privacy to smart contracts, but they usually rely on Trusted Execution Environments (TEEs) (e.g., [14, 24]) that are susceptible to side-channel attacks and other security concerns ([7, 13, 33] to name a few).
In the domain of spatial crowdsourcing drone services, which includes tasks like delivery, surveillance, and data collection, secure communication is paramount. The Public Key Infrastructure (PKI) ensures this by providing a system for digital certificates that authenticate the identities of entities involved, securing data and command transmissions between drones and their operators. However, the centralized trust model of traditional PKI, dependent on Certificate Authorities (CAs), presents a vulnerability due to its single point of failure, risking security breaches. To counteract this, the paper presents D2XChain, a blockchain-based PKI framework designed for the Internet of Drone Things (IoDT). By decentralizing the CA infrastructure, D2XChain eliminates this single point of failure, thereby enhancing the security and reliability of drone communications. Fully compatible with the X.509 standard, it integrates seamlessly with existing PKI systems, supporting all key operations such as certificate registration, validation, verification, and revocation in a distributed manner. This innovative approach not only strengthens the defense of drone services against various security threats but also showcases its practical application through deployment on a private Ethereum testbed, representing a significant advancement in addressing the unique security challenges of drone-based services and ensuring their trustworthy operation in critical tasks.
Blockchain technology is a key technology in the current information field and has been widely used in various industries. Blockchain technology faces significant challenges in privacy protection while ensuring data immutability and transparency, so it is crucial to implement private computing in blockchain. To target the privacy issues in blockchain, we design a secure multi-party computation (SMPC) protocol DHSMPC based on homomorphic encryption in this paper. On the one hand, homomorphic encryption technology can directly operate on ciphertext, solving the privacy problem in the blockchain. On the other hand, this paper designs the directed decryption function of DHSMPC to resist malicious opponents in the CRS model, so that authorized users who do not participate in the calculation can also access the decryption results of secure multi-party computation. Analytical and experimental results show that DHSMPC has smaller ciphertext size and stronger performance than existing SMPC protocols. The protocol makes it possible to implement complex calculations in multi-party scenarios and is proven to be resistant to various semi-malicious attacks, ensuring data security and privacy. Finally, this article combines the designed DHSMPC protocol with blockchain and cloud computing, showing how to use this solution to achieve trusted data management in specific scenarios.
With the development of Internet of Things technology, intelligent door lock devices are widely used in the field of house leasing. In the traditional housing leasing scenario, problems of door lock information disclosure, tenant privacy disclosure and rental contract disputes frequently occur, and the security, fairness and auditability of the housing leasing transaction cannot be guaranteed. To solve the above problems, a blockchain-based proxy re-encryption scheme with conditional privacy protection and auditability is proposed. The scheme implements fine-grained access control of door lock data based on attribute encryption technology with policy hiding, and uses proxy re-encryption technology to achieve auditable supervision of door lock information transactions. Homomorphic encryption technology and zero-knowledge proof technology are introduced to ensure the confidentiality of housing rent information and the fairness of rent payment. To construct a decentralized housing lease transaction architecture, the scheme realizes the efficient collaboration between the door lock data ciphertext stored under the chain and the key information ciphertext on the chain based on the blockchain and InterPlanetary File System. Finally, the security proof and computing performance analysis of the proposed scheme are carried out. The results show that the scheme can resist the chosen plaintext attack and has low computational cost.
Data owners seeking to boost processing power, storage, or bandwidth can take advantage of cloud computing services. However, this shift poses new challenges related to privacy and data security. Searchable Encryption (SE), which combines encryption and search techniques, addresses these issues (violation of data users' privacy) by allowing user data to be encrypted, transmitted to a cloud server, and searched using keywords. Despite its benefits, several recent real-world attacks have raised concerns about the security of searchable encryption. Ensuring forward and backward privacy is likely to become a standard requirement in the development of new SE systems. To address these issues, we propose a scheme that exclusively uses symmetric cryptographic primitives, achieving high communication efficiency and forward and backward privacy. In addition, we emphasize improved I/O efficiency because only the results of subsequent updates are loaded when searching. The time required to retrieve results is so significantly reduced compared to existing SE methods that we have shown that our scheme achieves superior efficiency. Moreover, by integrating blockchain network services with cloud services, we have developed a searchable intelligent cryptosystem suitable for lightweight smart devices. In our study conducted on the Ethereum network, we found our method to be both efficient and secure, especially when compared to methods such as PPSE and Jiang. The results indicate that our system delivers results in terms of performance and privacy within dynamic cloud environments making it a solution for protecting confidential information.
Na Wang, Yao Zhao, Youyang Qu, Lei Cui · 6 authors
Decentralized Federated Learning (DFL) is extensively applied in various areas, e.g., healthcare, finance, and Internet of Things (loT), offering practical solutions for distributed intelligent applications and data collaboration. In DFL systems, participants, e.g., edge devices, organizations, or nodes, collaborate in the training of a shared global model by aggregating local models from various participants. During this process, participants need to communicate frequently with a central authority/node/server to share model parameters. Such communication is vulnerable to malicious attacks or tampering, posing a significant threat to the integrity of model training. The integrity verification method can provide an integrity guarantee for the global model of DFL. However, most of the existing integrity verification schemes are centralized and not suitable for resource-constrained DFL scenarios. Therefore, how to verify the integrity of the global model becomes an important issue in DFL. To address it, we devise a global model integrity verification method for DFL. Specifically, we generate a digital signature for each global model parameter as proof of integrity, while improving the efficiency of integrity verification by electing delegates to conduct the verification process. A series of experiments is conducted to validate the performance of the proposed method. The experimental results demonstrate that our approach not only effectively ensures the integrity of the global model but also functions well under limited resources.
In recent trends of growth in technologies, data management, maintenance of medical records, sharing of data, diagnosis of disease, and medication are the key areas where digital healthcare plays a vital role. Despite enormous improvement, handling huge amounts of data, privacy, secure sharing, accuracy, and computational speed remains challenging. Federated learning is a machine learning technology that allows distributed model training using users’ own data to train a model. The model update is done through a central server that aggregates individual users and sends a global model. This ensures privacy protection and is suitable for handling large data. Blockchain technology is a publicly distributed ledger that collects the information of nodes as blocks and sends a copy to all nodes in the network so that data transparency is maintained and secure. However, blockchain has a limitation in handling large volumes of data. In such cases, federated learning can be used with a blockchain for better performance. By integrating federated learning with blockchain, accurate prediction, computational speed, data security, privacy, and accuracy can be achieved. A comprehensive review of how various federated learning technologies can integrate with blockchain networks to achieve accuracy and efficiency is presented.
The use of digital certificates is crucial for verifying the authenticity of various credentials in our digital age.However, traditional digital certificate systems suffer from centralization, vulnerability to tampering, and the risk of loss.This paper presents an approach to issuing and managing certificates as Non-Fungible Tokens (NFTs) on the Ethereum blockchain, ensuring their immutability and perpetual existence.The proposed system aims to overcome the limitations of traditional methods by utilizing decentralized storage through the InterPlanetary File System (IPFS) and implementing a robust incentive mechanism for voting and Proof of Stake; the goal is to achieve true permanence and enhanced security for digital certificates.A first phase of a smart contract and front-end interface was achieved, to preserve diplomas and provide a reliable method for verifying their authenticity for employers and other stakeholders.This paper is part of an ongoing effort to develop a robust system, create a prototype to validate the perpetuity assumption, and propose improvements for a global, enduring repository of certificates.