Selected work

Representative Papers

TroPUF: Evaluating Hardware Trojan Insertion in Delay-Based Physical Unclonable Functions

Jul 25, 2026

This work addresses the vulnerability of delay-based physically unclonable functions (PUFs) to stealthy hardware Trojan insertion, which exploits process-induced timing uncertainties—a threat inadequately mitigated by existing security verification methods. For the first time, the study integrates PUF security and hardware Trojan risks into a unified circuit-level simulation framework to systematically evaluate multiple delay-based PUF architectures in terms of functional reliability, hardware overhead, and resistance to machine learning modeling, both before and after Trojan implantation. Experimental results demonstrate that dormant Trojans can preserve normal PUF behavior and modeling resilience, revealing critical blind spots in current PUF validation approaches that fail to detect such threats prior to Trojan activation.

0 citationsRead paper

PATCH-FFT: Unmasking Dormant Hardware Trojans with Patch-Based Frequency-Domain Transformers

Jul 25, 2026

This work addresses the challenge of detecting dormant hardware Trojans, which are typically undetectable by conventional time-domain power analysis. To overcome this limitation, the authors propose a novel frequency-domain Patch Transformer architecture that integrates real-valued Fast Fourier Transform (rFFT) with a patch-based Transformer. This approach converts power traces into spectral representations and leverages the Transformer’s capacity to effectively extract and identify subtle spectral signatures indicative of hidden Trojans. Experimental results demonstrate that the proposed method achieves an average detection accuracy of 90.94% across both dormant and active hardware Trojan scenarios, substantially outperforming existing techniques. Notably, it delivers breakthrough performance in detecting dormant-state Trojans, where prior methods have largely failed.

0 citationsRead paper

Security-Induced Braess Paradoxes in Service Function Chain Orchestration

Jun 16, 2026

This work addresses the security-induced Braess paradox in service function chain (SFC) orchestration, where indiscriminate insertion of security functions can concentrate traffic, degrade performance, and amplify risk. The paper formally defines “Braessian security management actions,” derives sufficient conditions for the paradox to arise, and proposes a pre-deployment filtering mechanism that leverages an affine load-dependent delay model and game-theoretic equilibrium analysis to identify harmful orchestration choices. Extensive experiments on multi-tenant SFCs and real-world topologies—including fat-tree and NSFNET—demonstrate that conventional scaling strategies increase service costs by 27.2–30.8% and elevate risk concentration by 6.1–9.7×. In contrast, the proposed paradox-aware orchestration strategy limits performance penalty to under 1.9%, reduces service costs by 20.0–22.1%, and decreases proxy metrics for attack-induced losses by 93.5% on average.

0 citationsRead paper

Optimal FALQON for Quantum Approximate Optimization via Layer-wise Parameter Tuning

May 08, 2026

This work addresses the slow convergence of standard FALQON, which typically requires hundreds to thousands of layers to yield feasible solutions due to fixed hyperparameters. The authors propose the first approach that treats the time step and scaling factor at each FALQON layer as optimizable variables, adaptively tuning them via classical optimization algorithms. This adaptive strategy is leveraged to provide an efficient warm-start initialization for QAOA. By integrating feedback-based adaptive quantum optimization, classical optimization, and the QAOA framework, the method significantly enhances solution efficiency for combinatorial optimization problems on NISQ devices. Experiments on 94 twelve-vertex 3-regular graphs demonstrate clear advantages over standard FALQON and several QAOA variants in terms of success probability, evaluation efficiency, and depth-normalized cost.

0 citationsRead paper

Multirate Stein Variational Gradient Descent for Efficient Bayesian Sampling

Apr 05, 2026

This work addresses the limitations of standard Stein Variational Gradient Descent (SVGD), which employs a single global step size and struggles to balance the disparate evolution rates of attraction and repulsion effects in high-dimensional, anisotropic, or hierarchical posteriors, often leading to inefficiency or instability. To overcome this, the paper introduces a multi-timescale perspective into SVGD for the first time, proposing a multi-rate SVGD framework that decouples the updates of attractive and repulsive dynamics. It develops three strategies: a symmetric splitting scheme, a fixed multi-rate variant (MR-SVGD), and an adaptive multi-rate version (Adapt-MR-SVGD). Empirical evaluations across six benchmark tasks demonstrate substantial improvements over standard SVGD, particularly in rigid hierarchical models, strongly anisotropic distributions, and multimodal settings, with the adaptive variant achieving the best performance and the fixed variant offering a computationally efficient and robust alternative.

0 citationsRead paper

Recent publications

Latest Papers

TroPUF: Evaluating Hardware Trojan Insertion in Delay-Based Physical Unclonable Functions

Jul 25, 2026

This work addresses the vulnerability of delay-based physically unclonable functions (PUFs) to stealthy hardware Trojan insertion, which exploits process-induced timing uncertainties—a threat inadequately mitigated by existing security verification methods. For the first time, the study integrates PUF security and hardware Trojan risks into a unified circuit-level simulation framework to systematically evaluate multiple delay-based PUF architectures in terms of functional reliability, hardware overhead, and resistance to machine learning modeling, both before and after Trojan implantation. Experimental results demonstrate that dormant Trojans can preserve normal PUF behavior and modeling resilience, revealing critical blind spots in current PUF validation approaches that fail to detect such threats prior to Trojan activation.

0 citationsRead paper

PATCH-FFT: Unmasking Dormant Hardware Trojans with Patch-Based Frequency-Domain Transformers

Jul 25, 2026

This work addresses the challenge of detecting dormant hardware Trojans, which are typically undetectable by conventional time-domain power analysis. To overcome this limitation, the authors propose a novel frequency-domain Patch Transformer architecture that integrates real-valued Fast Fourier Transform (rFFT) with a patch-based Transformer. This approach converts power traces into spectral representations and leverages the Transformer’s capacity to effectively extract and identify subtle spectral signatures indicative of hidden Trojans. Experimental results demonstrate that the proposed method achieves an average detection accuracy of 90.94% across both dormant and active hardware Trojan scenarios, substantially outperforming existing techniques. Notably, it delivers breakthrough performance in detecting dormant-state Trojans, where prior methods have largely failed.

0 citationsRead paper

Security-Induced Braess Paradoxes in Service Function Chain Orchestration

Jun 16, 2026

This work addresses the security-induced Braess paradox in service function chain (SFC) orchestration, where indiscriminate insertion of security functions can concentrate traffic, degrade performance, and amplify risk. The paper formally defines “Braessian security management actions,” derives sufficient conditions for the paradox to arise, and proposes a pre-deployment filtering mechanism that leverages an affine load-dependent delay model and game-theoretic equilibrium analysis to identify harmful orchestration choices. Extensive experiments on multi-tenant SFCs and real-world topologies—including fat-tree and NSFNET—demonstrate that conventional scaling strategies increase service costs by 27.2–30.8% and elevate risk concentration by 6.1–9.7×. In contrast, the proposed paradox-aware orchestration strategy limits performance penalty to under 1.9%, reduces service costs by 20.0–22.1%, and decreases proxy metrics for attack-induced losses by 93.5% on average.

0 citationsRead paper

Optimal FALQON for Quantum Approximate Optimization via Layer-wise Parameter Tuning

May 08, 2026

This work addresses the slow convergence of standard FALQON, which typically requires hundreds to thousands of layers to yield feasible solutions due to fixed hyperparameters. The authors propose the first approach that treats the time step and scaling factor at each FALQON layer as optimizable variables, adaptively tuning them via classical optimization algorithms. This adaptive strategy is leveraged to provide an efficient warm-start initialization for QAOA. By integrating feedback-based adaptive quantum optimization, classical optimization, and the QAOA framework, the method significantly enhances solution efficiency for combinatorial optimization problems on NISQ devices. Experiments on 94 twelve-vertex 3-regular graphs demonstrate clear advantages over standard FALQON and several QAOA variants in terms of success probability, evaluation efficiency, and depth-normalized cost.

0 citationsRead paper

Multirate Stein Variational Gradient Descent for Efficient Bayesian Sampling

Apr 05, 2026

This work addresses the limitations of standard Stein Variational Gradient Descent (SVGD), which employs a single global step size and struggles to balance the disparate evolution rates of attraction and repulsion effects in high-dimensional, anisotropic, or hierarchical posteriors, often leading to inefficiency or instability. To overcome this, the paper introduces a multi-timescale perspective into SVGD for the first time, proposing a multi-rate SVGD framework that decouples the updates of attractive and repulsive dynamics. It develops three strategies: a symmetric splitting scheme, a fixed multi-rate variant (MR-SVGD), and an adaptive multi-rate version (Adapt-MR-SVGD). Empirical evaluations across six benchmark tasks demonstrate substantial improvements over standard SVGD, particularly in rigid hierarchical models, strongly anisotropic distributions, and multimodal settings, with the adaptive variant achieving the best performance and the fixed variant offering a computationally efficient and robust alternative.

0 citationsRead paper