Publications

Protects maritime RTK correction streams with post-quantum key establishment and lightweight authentication, detecting tampering with less than one percent bandwidth overhead.

Compares Ghana's unified virtual-asset statute with the fragmented US anti-money-laundering framework, identifying implementation and supervisory-capacity tradeoffs.

Combines physical-unclonable-function device fingerprints with zero-knowledge proofs for scalable, privacy-preserving authentication in blockchain-IoT.

Benchmarks explainability methods (attention, saliency maps, integrated gradients, SHAP, LIME) on a BiLSTM for blood-glucose forecasting, finding attention and gradient-based methods most faithful for clinical time series.

Federated temporal graph neural networks for lightweight, adversarially robust malware detection across distributed edge networks.

Uses post-quantum authentication and blockchain verification to protect model updates in federated learning.

Models the MEV supply chain (searchers, builders, validators) as a three-stage game of incomplete information, analyzes commit-reveal and threshold-encryption mitigations, and validates the theory against Ethereum on-chain data.

Combines ML threat scoring and game-theoretic incentives to turn idle IoT devices into honeypots as threats change.

A post-quantum, blockchain-verified federated learning framework tailored to privacy-sensitive healthcare analytics.

Quantifies how ageing of Two-Line Element orbital data causes duty-cycled LEO IoT terminals to miss satellite passes, and finds that keeping TLEs fresher than about six hours is needed for over 99% pass reliability.

A scheme for securing IoT firmware updates against supply-chain attacks on the update pipeline.

An enhanced federated learning approach for secure, privacy-preserving collaboration on medical data across institutions.

A comprehensive review of security mechanisms, threat models, and open challenges for blockchain-based IoT systems.

Compares 3DES, AES, Blowfish, and RSA for encrypting and classifying datasets in cloud storage, weighing security against performance.

Fuses vessel behavior with weather context to attribute, in real time, whether maritime AIS signal dropouts are benign or suspicious.

A 24-month measurement study of congestion dynamics and reliability in production blockchain Layer-2 networks.

Federated, clustered, quantization-aware models for lightweight DDoS detection in the Internet of Robotic Things.

Uses lightweight digital twins to predict low-value client updates and skip them, cutting federated-learning communication without destabilizing convergence.

A unified, lightweight, reproducible benchmark for privacy-preserving federated learning in cyber-physical systems, demonstrated on Fashion-MNIST.

Resource-aware clustered federated learning for industrial digital twins, with a reproducible Fashion-MNIST benchmark.

Uses blockchain service contracts to manage and incentivize adaptive behavior in robotic systems.

The conference foundation for PUFZIN: device authentication for blockchain-IoT built on physical unclonable functions and zero-knowledge proofs.

Applies federated learning to secure the Internet of Robotic Things without centralizing sensitive device data.

Studies where to place honeypots in blockchain-based IoT systems for the most efficient deception-based defense.

Combines an evolution strategy with a GAN to generate ransomware variants that stay functional while evading a majority of real-world VirusTotal detection engines.

Shows that adding an attractive security-service option can worsen adapted service-chain equilibria, then screens or caps harmful options before deployment.

Uses post-quantum-protected boundary-embedding exchange for federated fraud detection. Results depend on graph connectivity, and shared embeddings retain some information about input features.

Verifies that a client’s local loss is below a threshold without revealing the loss value, using zero-knowledge proofs in federated evaluation.

Studies incentives for discouraging data poisoning by rational clients under a Bayesian model of participation.

Benchmarks post-quantum cryptography (for example, ML-KEM) on consumer and edge hardware such as the Raspberry Pi, showing it is practical and offering concrete deployment guidance.

Combines differential privacy with blockchain-verified federated learning to protect health data during collaborative training.