Selected work

Representative Papers

WiFi-Based People Counting Using Beam-Steerable Antennas: A Test-bed Study

Jun 15, 2026

This study addresses the challenge of high-accuracy, non-intrusive indoor occupancy counting by proposing a novel approach that integrates Wi-Fi 7 with beam-steerable antennas. Leveraging the fine-grained channel state information (CSI) extracted from 160/320 MHz wideband OFDM signals, the method combines MIMO antenna arrays and beamforming techniques to precisely estimate the channel impulse response and resolve multipath characteristics, thereby detecting changes in the number of occupants. This work is the first to synergize beam-steerable antennas with Wi-Fi 7, enabling robust multi-person counting in real-world environments and revealing the distinctiveness of individual-induced wireless signal perturbations. Experimental results demonstrate that the system significantly enhances counting accuracy and stability under complex indoor conditions.

0 citationsRead paper

SPEC CPU: The Next Generation

May 02, 2026

This work addresses the limitations of existing CPU benchmarks in accurately evaluating the performance of modern heterogeneous, multithreaded processors under diverse workloads. To this end, the authors present the SPEC CPU 2026 benchmark suite, developed through community collaboration and principled methodology, which introduces the Rolling-Round-Robin Rate approach to standardize the execution of heterogeneous multiprogrammed workloads. The suite incorporates newly designed multithreaded benchmarks exhibiting varied microarchitectural characteristics, selected and hardened through an open-source application curation process. Emphasizing workload diversity, portability, and long-term viability, SPEC CPU 2026 establishes a robust, representative, and authoritative standard for performance evaluation, thereby supporting next-generation computer architecture research.

0 citationsRead paper

ActionFlow: A Pipelined Action Acceleration for Vision Language Models on Edge

Dec 23, 2025

VLA models suffer from high inference latency (only 3–5 Hz) on edge devices due to memory bottlenecks induced by autoregressive decoding—far below the 20–30 Hz required for real-time robotic control. To address this, we propose the first system-level acceleration framework tailored for real-time VLA inference on edge platforms. Our approach introduces cross-request pipelined scheduling, reformulating VLA decoding as a macro-pipeline; pioneers cross-request state-packed forward operators and a unified KV circular cache to overcome GPU memory constraints; and synergistically optimizes the heterogeneous prefill and decode phases via micro-batching. Evaluated on OpenVLA-7B, our framework achieves a 2.55× FPS improvement with zero retraining, enabling—for the first time on edge hardware—sustained >20 Hz dynamic operation for real-time robotic control.

0 citationsRead paper

SimpleVSF: VLM-Scoring Fusion for Trajectory Prediction of End-to-End Autonomous Driving

Oct 20, 2025

To address the insufficient robustness of end-to-end autonomous driving in complex scenarios—particularly in trajectory decision-making—this paper proposes SimpleVSF, a novel framework that introduces visual-language models (VLMs) into the trajectory scoring stage for the first time, enabling cognitive-level scene understanding. We design a two-tier fusion mechanism that jointly integrates quantitative weighted fusion with qualitative, context-aware reasoning to holistically optimize safety, comfort, and efficiency. By eliminating redundant components, SimpleVSF achieves high performance with a lightweight architecture, significantly improving planning reliability under long-horizon and highly dynamic conditions. Evaluated on the ICCV 2025 NAVSIM v2 Challenge, SimpleVSF attains state-of-the-art (SOTA) overall performance, demonstrating superior balance across multiple metrics compared to existing approaches.

0 citationsRead paper

Pharmacist: Safety Alignment Data Curation for Large Language Models against Harmful Fine-tuning

Oct 11, 2025

To address the limited robustness of harmful fine-tuning defenses in “fine-tuning-as-a-service” for large language models—stemming from low-quality safety-aligned training data—this paper proposes Pharmacist. We first identify and empirically demonstrate that alignment data quality critically constrains defense robustness. Pharmacist introduces a learnable data selector that dynamically identifies high-quality, safety-critical samples while suppressing low-quality and non-safety-relevant ones. Compatible with state-of-the-art defenses such as RepNoise and T-Vaccine, Pharmacist achieves consistent improvements across multiple benchmarks: +2.60–3.30% defense accuracy, +1.10–3.50% inference speedup, and −56.8–57.6% reduction in training time—substantially outperforming existing data selection approaches. Our core contribution lies in integrating explicit data quality modeling into the alignment-based defense framework, enabling efficient, robust, and plug-and-play enhancement of secure fine-tuning.

0 citationsRead paper

Recent publications

Latest Papers

WiFi-Based People Counting Using Beam-Steerable Antennas: A Test-bed Study

Jun 15, 2026

This study addresses the challenge of high-accuracy, non-intrusive indoor occupancy counting by proposing a novel approach that integrates Wi-Fi 7 with beam-steerable antennas. Leveraging the fine-grained channel state information (CSI) extracted from 160/320 MHz wideband OFDM signals, the method combines MIMO antenna arrays and beamforming techniques to precisely estimate the channel impulse response and resolve multipath characteristics, thereby detecting changes in the number of occupants. This work is the first to synergize beam-steerable antennas with Wi-Fi 7, enabling robust multi-person counting in real-world environments and revealing the distinctiveness of individual-induced wireless signal perturbations. Experimental results demonstrate that the system significantly enhances counting accuracy and stability under complex indoor conditions.

0 citationsRead paper

SPEC CPU: The Next Generation

May 02, 2026

This work addresses the limitations of existing CPU benchmarks in accurately evaluating the performance of modern heterogeneous, multithreaded processors under diverse workloads. To this end, the authors present the SPEC CPU 2026 benchmark suite, developed through community collaboration and principled methodology, which introduces the Rolling-Round-Robin Rate approach to standardize the execution of heterogeneous multiprogrammed workloads. The suite incorporates newly designed multithreaded benchmarks exhibiting varied microarchitectural characteristics, selected and hardened through an open-source application curation process. Emphasizing workload diversity, portability, and long-term viability, SPEC CPU 2026 establishes a robust, representative, and authoritative standard for performance evaluation, thereby supporting next-generation computer architecture research.

0 citationsRead paper

ActionFlow: A Pipelined Action Acceleration for Vision Language Models on Edge

Dec 23, 2025

VLA models suffer from high inference latency (only 3–5 Hz) on edge devices due to memory bottlenecks induced by autoregressive decoding—far below the 20–30 Hz required for real-time robotic control. To address this, we propose the first system-level acceleration framework tailored for real-time VLA inference on edge platforms. Our approach introduces cross-request pipelined scheduling, reformulating VLA decoding as a macro-pipeline; pioneers cross-request state-packed forward operators and a unified KV circular cache to overcome GPU memory constraints; and synergistically optimizes the heterogeneous prefill and decode phases via micro-batching. Evaluated on OpenVLA-7B, our framework achieves a 2.55× FPS improvement with zero retraining, enabling—for the first time on edge hardware—sustained >20 Hz dynamic operation for real-time robotic control.

0 citationsRead paper

SimpleVSF: VLM-Scoring Fusion for Trajectory Prediction of End-to-End Autonomous Driving

Oct 20, 2025

To address the insufficient robustness of end-to-end autonomous driving in complex scenarios—particularly in trajectory decision-making—this paper proposes SimpleVSF, a novel framework that introduces visual-language models (VLMs) into the trajectory scoring stage for the first time, enabling cognitive-level scene understanding. We design a two-tier fusion mechanism that jointly integrates quantitative weighted fusion with qualitative, context-aware reasoning to holistically optimize safety, comfort, and efficiency. By eliminating redundant components, SimpleVSF achieves high performance with a lightweight architecture, significantly improving planning reliability under long-horizon and highly dynamic conditions. Evaluated on the ICCV 2025 NAVSIM v2 Challenge, SimpleVSF attains state-of-the-art (SOTA) overall performance, demonstrating superior balance across multiple metrics compared to existing approaches.

0 citationsRead paper

Pharmacist: Safety Alignment Data Curation for Large Language Models against Harmful Fine-tuning

Oct 11, 2025

To address the limited robustness of harmful fine-tuning defenses in “fine-tuning-as-a-service” for large language models—stemming from low-quality safety-aligned training data—this paper proposes Pharmacist. We first identify and empirically demonstrate that alignment data quality critically constrains defense robustness. Pharmacist introduces a learnable data selector that dynamically identifies high-quality, safety-critical samples while suppressing low-quality and non-safety-relevant ones. Compatible with state-of-the-art defenses such as RepNoise and T-Vaccine, Pharmacist achieves consistent improvements across multiple benchmarks: +2.60–3.30% defense accuracy, +1.10–3.50% inference speedup, and −56.8–57.6% reduction in training time—substantially outperforming existing data selection approaches. Our core contribution lies in integrating explicit data quality modeling into the alignment-based defense framework, enabling efficient, robust, and plug-and-play enhancement of secure fine-tuning.

0 citationsRead paper