Selected work

Representative Papers

Enhanced Dynamic Beamwidth Selection-based THz MAC Protocol for Wireless Data Center Networks

Jul 20, 2026

This study addresses the performance trade-off between throughput and link distance in terahertz wireless data center networks caused by fixed beamwidths. To overcome this limitation, the authors propose a distance-aware dynamic beamwidth adaptation mechanism, termed DBS-ADAPT, which—unlike prior approaches—enables real-time adjustment of beamwidth based on node separation within the terahertz MAC protocol, thereby maximizing throughput while maintaining adequate coverage. An enhanced variant, EDBS-ADAPT, is further introduced to substantially reduce beam switching and control overhead. Extensive simulations using the NS-3 terahertz module demonstrate that DBS-ADAPT improves average throughput by 22% and reduces latency by 10% compared to baseline protocols, while EDBS-ADAPT achieves a remarkable 95% reduction in beam switching overhead.

0 citationsRead paper

A Reproducible Software Workflow for Unanchored Approximate MUB Optimization: A Case Study in Dimension Six

Jul 12, 2026

This study investigates the existence and optimization of anchor-free approximate mutually unbiased bases (AMUBs) in high-dimensional Hilbert spaces, with a particular focus on whether more near-exact bases can be constructed in dimension six. We present the first reproducible optimization framework supporting multiple hardware backends—including CPU, Apple MPS, CUDA GPUs, and HPC systems—leveraging Lie algebra-based unitary parameterization and Taylor-series matrix exponential layers to efficiently explore AMUB configurations in arbitrary dimensions. Our key contributions include reproducing the exact three-basis solution, identifying a recurring “spoke–triangle” partially exact structure in four-basis settings, and experimentally validating the optimized unitaries on an IBM Heron quantum processor. Results indicate no near-exact pairs emerge for five or six bases, and quantum processing unit (QPU) experiments reveal that noise-induced losses (0.02–0.08) dominate, obscuring distinctions between classically near-exact and defective pairs.

0 citationsRead paper

Bridging Semantics and Kinematics: A Modular Framework for Zero-Shot Robotic Manipulation

Jun 22, 2026

This work addresses the gap between high-level semantic instructions and low-level motion control in zero-shot, language-guided robotic manipulation, particularly tackling the spatial ambiguity and semantic hallucination challenges of vision-language models in semi-structured environments. The authors propose a training-free, modular framework that decouples the visuomotor pipeline into three stages: visual perception, semantic interpretation, and task execution. It innovatively combines FastSAM with Set-of-Mark prompting to generate verifiable visual anchors and repurposes a unified foundation model as a pure language model for semantic routing—eliminating the need for fine-tuning or coordinate-based programming. Built upon a large language model and MoveIt Task Constructor, the end-to-end reconfigurable pipeline achieves a 62% zero-shot success rate on open-world sequential manipulation and dense relational reasoning tasks, demonstrating its effectiveness without domain-specific training.

0 citationsRead paper

Eliminating Premature Termination in Multihop Rendezvous for Cognitive Radio-based Emergency Response Network

May 21, 2026

This work addresses the premature termination problem in multihop rendezvous protocols caused by the N−1 termination condition, which compromises reliability in post-disaster emergency communications. To resolve this issue, the authors propose a robust rendezvous mechanism incorporating coordinate-assisted neighbor verification and an autonomous termination strategy, ensuring complete neighbor discovery and accurate topology construction prior to protocol termination. As the first study to identify and mitigate this specific limitation, the proposed approach is generic and readily adaptable to various multihop rendezvous protocols. Experimental results demonstrate that, under a challenging scenario with 20 nodes, 20 channels, and high primary user activity, the method achieves 100% accurate topology discovery and reduces rendezvous time by up to 76% compared to baseline protocols.

0 citationsRead paper

See, Learn, Assist: Safe and Self-Paced Robotic Rehabilitation via Video-Based Learning from Demonstration

Mar 14, 2026

This study addresses the challenges of personalizing rehabilitation robots to individual anatomical structures and ensuring safe human–robot interaction by proposing a rehabilitation training framework based on RGB-D video demonstration. The approach encodes therapist demonstrations into body-centered 6-degree-of-freedom trajectories and leverages Cartesian dynamic movement primitives (DMPs) combined with Gaussian mixture regression (GMR) to achieve anatomy-agnostic, precise motion reproduction. A decoupled hybrid control architecture integrates a virtual compliant tunnel with a tangential-force-based temporal scaling mechanism, enabling seamless transitions among passive, active-assistive, and resistive training modes while supporting adaptive rhythm modulation and real-time anomalous force detection. Experimental results demonstrate a mean trajectory reproduction error of 3.7 cm and a joint range-of-motion error of 5.5°, with the system maintaining path accuracy and dynamically adjusting training intensity even under deliberate external disturbances.

0 citationsRead paper

Recent publications

Latest Papers

Enhanced Dynamic Beamwidth Selection-based THz MAC Protocol for Wireless Data Center Networks

Jul 20, 2026

This study addresses the performance trade-off between throughput and link distance in terahertz wireless data center networks caused by fixed beamwidths. To overcome this limitation, the authors propose a distance-aware dynamic beamwidth adaptation mechanism, termed DBS-ADAPT, which—unlike prior approaches—enables real-time adjustment of beamwidth based on node separation within the terahertz MAC protocol, thereby maximizing throughput while maintaining adequate coverage. An enhanced variant, EDBS-ADAPT, is further introduced to substantially reduce beam switching and control overhead. Extensive simulations using the NS-3 terahertz module demonstrate that DBS-ADAPT improves average throughput by 22% and reduces latency by 10% compared to baseline protocols, while EDBS-ADAPT achieves a remarkable 95% reduction in beam switching overhead.

0 citationsRead paper

A Reproducible Software Workflow for Unanchored Approximate MUB Optimization: A Case Study in Dimension Six

Jul 12, 2026

This study investigates the existence and optimization of anchor-free approximate mutually unbiased bases (AMUBs) in high-dimensional Hilbert spaces, with a particular focus on whether more near-exact bases can be constructed in dimension six. We present the first reproducible optimization framework supporting multiple hardware backends—including CPU, Apple MPS, CUDA GPUs, and HPC systems—leveraging Lie algebra-based unitary parameterization and Taylor-series matrix exponential layers to efficiently explore AMUB configurations in arbitrary dimensions. Our key contributions include reproducing the exact three-basis solution, identifying a recurring “spoke–triangle” partially exact structure in four-basis settings, and experimentally validating the optimized unitaries on an IBM Heron quantum processor. Results indicate no near-exact pairs emerge for five or six bases, and quantum processing unit (QPU) experiments reveal that noise-induced losses (0.02–0.08) dominate, obscuring distinctions between classically near-exact and defective pairs.

0 citationsRead paper

Bridging Semantics and Kinematics: A Modular Framework for Zero-Shot Robotic Manipulation

Jun 22, 2026

This work addresses the gap between high-level semantic instructions and low-level motion control in zero-shot, language-guided robotic manipulation, particularly tackling the spatial ambiguity and semantic hallucination challenges of vision-language models in semi-structured environments. The authors propose a training-free, modular framework that decouples the visuomotor pipeline into three stages: visual perception, semantic interpretation, and task execution. It innovatively combines FastSAM with Set-of-Mark prompting to generate verifiable visual anchors and repurposes a unified foundation model as a pure language model for semantic routing—eliminating the need for fine-tuning or coordinate-based programming. Built upon a large language model and MoveIt Task Constructor, the end-to-end reconfigurable pipeline achieves a 62% zero-shot success rate on open-world sequential manipulation and dense relational reasoning tasks, demonstrating its effectiveness without domain-specific training.

0 citationsRead paper

Eliminating Premature Termination in Multihop Rendezvous for Cognitive Radio-based Emergency Response Network

May 21, 2026

This work addresses the premature termination problem in multihop rendezvous protocols caused by the N−1 termination condition, which compromises reliability in post-disaster emergency communications. To resolve this issue, the authors propose a robust rendezvous mechanism incorporating coordinate-assisted neighbor verification and an autonomous termination strategy, ensuring complete neighbor discovery and accurate topology construction prior to protocol termination. As the first study to identify and mitigate this specific limitation, the proposed approach is generic and readily adaptable to various multihop rendezvous protocols. Experimental results demonstrate that, under a challenging scenario with 20 nodes, 20 channels, and high primary user activity, the method achieves 100% accurate topology discovery and reduces rendezvous time by up to 76% compared to baseline protocols.

0 citationsRead paper

See, Learn, Assist: Safe and Self-Paced Robotic Rehabilitation via Video-Based Learning from Demonstration

Mar 14, 2026

This study addresses the challenges of personalizing rehabilitation robots to individual anatomical structures and ensuring safe human–robot interaction by proposing a rehabilitation training framework based on RGB-D video demonstration. The approach encodes therapist demonstrations into body-centered 6-degree-of-freedom trajectories and leverages Cartesian dynamic movement primitives (DMPs) combined with Gaussian mixture regression (GMR) to achieve anatomy-agnostic, precise motion reproduction. A decoupled hybrid control architecture integrates a virtual compliant tunnel with a tangential-force-based temporal scaling mechanism, enabling seamless transitions among passive, active-assistive, and resistive training modes while supporting adaptive rhythm modulation and real-time anomalous force detection. Experimental results demonstrate a mean trajectory reproduction error of 3.7 cm and a joint range-of-motion error of 5.5°, with the system maintaining path accuracy and dynamically adjusting training intensity even under deliberate external disturbances.

0 citationsRead paper