Yaohua Wang
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Yaohua Wang has a total of 85 co-authors. Here are the top co-authors by number of shared publications:
- Yang Guo
National University of Defense Technology
6 shared publications - Jianjun Chen
National University of Defense Technology
3 shared publications Sheng Liu
3 shared publications- Gaoyang Zhao
National University of Defense Technology, Changsha, Hunan, China
2 shared publications - Jinjin Shao
College of Computer, National University of Defense Technology, Changsha 410073, China
2 shared publications - Rongzhen Lin
College of Computer Science and Technology, National University of Defense Technology
2 shared publications
Latest Publications
- 1 November 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Computers
- Vol. 75 (11) , 3984-3996
LIGHTeX: A Lightweight Framework for Extending Applications Into Trusted Execution
- 1 November 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Computers
- Vol. 75 (11) , 3984-3996
LIGHTeX: A Lightweight Framework for Extending Applications Into Trusted Execution
- Open Access
- 26 July 2026
- Published by Association for Computing Machinery (ACM)
- p. 1-7
Accelerating AI+SAR Applications through Enhanced Graph Optimization
- 26 July 2026
- Published by Association for Computing Machinery (ACM)
- p. 1-7
Open AccessAccelerating AI+SAR Applications through Enhanced Graph Optimization
- Open Access
- 15 June 2026
- Published by Association for Computing Machinery (ACM)
- p. 70-74
CVS: A Metric for Security-Aware Compilation against Side-Channel Attacks in Edge SoCs (WIP)
- 15 June 2026
- Published by Association for Computing Machinery (ACM)
- p. 70-74
Open AccessCVS: A Metric for Security-Aware Compilation against Side-Channel Attacks in Edge SoCs (WIP)
- 10 February 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Nuclear Science
- Vol. PP (00189499) , 1
DAR-SPUF: A Dynamically Adjustable Radiation-hardening Method for SRAM-Based Physical Unclonable Functions
With the increasing demand for encrypted communication in spacecraft, aerospace-grade chips are required to integrate encryption circuits. The lightweight characteristics of static random access memory (SRAM) physical unclonable function (PUF) make it suitable for aerospace applications. However, the space radiation effects can degrade SRAM PUF key reliability and even cause stuck-bit errors. Therefore, it is necessary to harden the SRAM PUF circuit to obtain a stable encryption key. This paper presents a dynamically adjustable radiation-hardening method for SRAM PUF (DAR-SPUF). The method generates stable keys through dynamic key adjustment circuit, stuck-bit error detection circuit, and other peripheral circuits. The proposed method does not modify the structure of traditional SRAM PUF cell. Compared to existing hardening methods, the proposed method exhibits smaller area overhead and power consumption. A FinFET-based test chip was designed and fabricated to verify the effectiveness of the DAR-SPUF. The test chips were irradiated with total ionizing dose (TID) and heavy ions. The experimental results show that radiation effects increase the bit error rate (BER) of unhardened test chip by 1.34x – 2.57x. However, it reduces the BER of the hardened DAR-SPUF chip by 19.73x – 24.52x. The experimental results also demonstrate that the DAR-SPUF can mitigate the stuck-bit errors to 0.
- 10 February 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Nuclear Science
- Vol. PP (00189499) , 1
DAR-SPUF: A Dynamically Adjustable Radiation-hardening Method for SRAM-Based Physical Unclonable Functions
With the increasing demand for encrypted communication in spacecraft, aerospace-grade chips are required to integrate encryption circuits. The lightweight characteristics of static random access memory (SRAM) physical unclonable function (PUF) make it suitable for aerospace applications. However, the space radiation effects can degrade SRAM PUF key reliability and even cause stuck-bit errors. Therefore, it is necessary to harden the SRAM PUF circuit to obtain a stable encryption key. This paper presents a dynamically adjustable radiation-hardening method for SRAM PUF (DAR-SPUF). The method generates stable keys through dynamic key adjustment circuit, stuck-bit error detection circuit, and other peripheral circuits. The proposed method does not modify the structure of traditional SRAM PUF cell. Compared to existing hardening methods, the proposed method exhibits smaller area overhead and power consumption. A FinFET-based test chip was designed and fabricated to verify the effectiveness of the DAR-SPUF. The test chips were irradiated with total ionizing dose (TID) and heavy ions. The experimental results show that radiation effects increase the bit error rate (BER) of unhardened test chip by 1.34x – 2.57x. However, it reduces the BER of the hardened DAR-SPUF chip by 19.73x – 24.52x. The experimental results also demonstrate that the DAR-SPUF can mitigate the stuck-bit errors to 0.
- 29 December 2025
- Published by Elsevier in Computers & Security
- Vol. 163
GLRA: Graph-based leakage risk assessment via minimal transmission cost path analysis
- 29 December 2025
- Published by Elsevier in Computers & Security
- Vol. 163
GLRA: Graph-based leakage risk assessment via minimal transmission cost path analysis
- 14 October 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Dependable and Secure Computing
- Vol. 23 (2) , 1812-1829
Sub-Pixel Hardware Trojan Detection With Enhanced Accuracy, Robustness, and Efficiency via Noise-Informed Thermal Radiation Analysis
Hardware Trojans (HTs) pose a significant and growing threat to the field of hardware security. Several side-channel techniques, including power and electromagnetic radiation (EMR), have been proposed for HT detection, constrained by reliance on the golden chip or test vectors. In response, researchers advocate for the use of thermal radiation (TR) to identify HTs. However, existing TR-based methods are designed for the ideal HT that can fully occupy at least one pixel on the thermal radiation map (TRM). In reality, HTs may occupy multiple pixels, substantially diminishing occupancy in each pixel, thereby reducing the detection accuracy of existing methods reliant on the traditional statistical analysis. This challenge is exacerbated by the noise caused by the thermal camera. To this end, this paper introduces a countermeasure named noise based pixel occupation enhancement (NICE), aiming to improve the ability of TR-based HT detection. The key insight of NICE is that noise can vary the pixel occupation of HTs while disrupting HT detection. Consequently, the noise can be exploited to statistically find out the largest pixel occupation among the variations, thereby enhancing HT detection accuracy. Subsequently, the Gaussian Mixture Model (GMM) and Isolation Forest (iForest) algorithms are integrated into the NICE framework to identify potential HT pixels based on the distinctive distribution characteristics of TR data, enabling efficient and robust HT detection. Experimental results on a 0.13$\boldsymbol{\mu }$m Digital Signal Processing (DSP) show that the detection rate of NICE exceeds the existing TR-based method by 50%, reaching 95%, while maintaining a false alarm rate of 5%. Both metrics of NICE are comparable to the existing power-based and EMR-based methods, eliminating the need for the golden chip and test vectors.
- 14 October 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Dependable and Secure Computing
- Vol. 23 (2) , 1812-1829
Sub-Pixel Hardware Trojan Detection With Enhanced Accuracy, Robustness, and Efficiency via Noise-Informed Thermal Radiation Analysis
Hardware Trojans (HTs) pose a significant and growing threat to the field of hardware security. Several side-channel techniques, including power and electromagnetic radiation (EMR), have been proposed for HT detection, constrained by reliance on the golden chip or test vectors. In response, researchers advocate for the use of thermal radiation (TR) to identify HTs. However, existing TR-based methods are designed for the ideal HT that can fully occupy at least one pixel on the thermal radiation map (TRM). In reality, HTs may occupy multiple pixels, substantially diminishing occupancy in each pixel, thereby reducing the detection accuracy of existing methods reliant on the traditional statistical analysis. This challenge is exacerbated by the noise caused by the thermal camera. To this end, this paper introduces a countermeasure named noise based pixel occupation enhancement (NICE), aiming to improve the ability of TR-based HT detection. The key insight of NICE is that noise can vary the pixel occupation of HTs while disrupting HT detection. Consequently, the noise can be exploited to statistically find out the largest pixel occupation among the variations, thereby enhancing HT detection accuracy. Subsequently, the Gaussian Mixture Model (GMM) and Isolation Forest (iForest) algorithms are integrated into the NICE framework to identify potential HT pixels based on the distinctive distribution characteristics of TR data, enabling efficient and robust HT detection. Experimental results on a 0.13$\boldsymbol{\mu }$m Digital Signal Processing (DSP) show that the detection rate of NICE exceeds the existing TR-based method by 50%, reaching 95%, while maintaining a false alarm rate of 5%. Both metrics of NICE are comparable to the existing power-based and EMR-based methods, eliminating the need for the golden chip and test vectors.
- 2 September 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Robotics and Automation Letters
- Vol. 10 (10) , 10642-10649
High-Accuracy and Energy-Efficient Learning of Inverse Dynamics for High-DoF Robotic Manipulators Using Spiking Neural Networks
Inverse dynamics maps the joint space of a robotic system to the torque space, but traditional methods for solving it rely on direct physical modeling of robots, which is difficult or even impossible due to high nonlinearity and external disturbance. Data-driven methods address these issues by learning the complex and unknown inverse dynamics. One of the key features of inverse dynamics is temporal dependency because robot movements are continuous. While recent Artificial Neural Network (ANN) methods leverage this to improve prediction accuracy, their high computational cost limits use in edge computing scenarios. Neuromorphic computing based on Spiking Neural Network (SNN) features extremely low power consumption. However, existing neuromorphic approaches remain limited to 2 degrees-of-freedom (DoFs) due to insufficient temporal modeling and low scalability. This paper presents the first neuromorphic approach capable of accurately learning the inverse dynamics of up to 7 DoF robotic manipulators. The method utilizes a liquid state machine (LSM) with temporally structured inputs and a novel decoding strategy to better model the temporal dependency in robotic motion. To account for the distinct characteristics of each joint while enhancing the model's scalability, we adopt a multi-LSM structure, enabling individualized modeling for each joint. Our neuromorphic method outperforms state-of-the-art approaches by at least 14.6% while using only 40% of their power consumption. It achieves the highest prediction accuracy across four different robotic platforms and is further validated on a UR5 robotic manipulator in a trajectory tracking task.
- 2 September 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Robotics and Automation Letters
- Vol. 10 (10) , 10642-10649
High-Accuracy and Energy-Efficient Learning of Inverse Dynamics for High-DoF Robotic Manipulators Using Spiking Neural Networks
Inverse dynamics maps the joint space of a robotic system to the torque space, but traditional methods for solving it rely on direct physical modeling of robots, which is difficult or even impossible due to high nonlinearity and external disturbance. Data-driven methods address these issues by learning the complex and unknown inverse dynamics. One of the key features of inverse dynamics is temporal dependency because robot movements are continuous. While recent Artificial Neural Network (ANN) methods leverage this to improve prediction accuracy, their high computational cost limits use in edge computing scenarios. Neuromorphic computing based on Spiking Neural Network (SNN) features extremely low power consumption. However, existing neuromorphic approaches remain limited to 2 degrees-of-freedom (DoFs) due to insufficient temporal modeling and low scalability. This paper presents the first neuromorphic approach capable of accurately learning the inverse dynamics of up to 7 DoF robotic manipulators. The method utilizes a liquid state machine (LSM) with temporally structured inputs and a novel decoding strategy to better model the temporal dependency in robotic motion. To account for the distinct characteristics of each joint while enhancing the model's scalability, we adopt a multi-LSM structure, enabling individualized modeling for each joint. Our neuromorphic method outperforms state-of-the-art approaches by at least 14.6% while using only 40% of their power consumption. It achieves the highest prediction accuracy across four different robotic platforms and is further validated on a UR5 robotic manipulator in a trajectory tracking task.
- 15 July 2025
- Published by Springer Nature
- p. 370-382
Anomaly-Driven Defense: Mitigating Bit-Flip Attacks in Deep Neural Networks via Runtime Parameter Clipping
- 15 July 2025
- Published by Springer Nature
- p. 370-382
Anomaly-Driven Defense: Mitigating Bit-Flip Attacks in Deep Neural Networks via Runtime Parameter Clipping
- Open Access
- 28 June 2025
- Published by Association for Computing Machinery (ACM) in ACM Transactions on Architecture and Code Optimization
- Vol. 22 (2) , 1-25
Shift-CIM: In-SRAM Alignment To Support General-Purpose Bit-level Sparsity Exploration in SRAM Multiplication
Multiplication plays a critical role in SRAM-based Computing-in-Memory (CIM) architectures. However, current SRAM-based CIMs face three major limitations. First, they do not fully exploit bit-level sparsity, resulting in unnecessary overhead in both latency and energy consumption. Second, the generation of numerous zero-dot products is superfluous. Third, the irregular organization of SRAM complicates the implementation. To address these issues, we propose Shift-CIM, a general-purpose approach that fully leverages bit-level sparsity within SRAM-based multiplications. Shift-CIM aligns the multipliers within the SRAM array, accumulating only the required dot products based on the non-zero bits of the multipliers. Shift-CIM achieves a regular SRAM organization by assembling two irregular SRAM arrays in a transposed manner. Our evaluations show that Shift-CIM is highly efficient, operating at a supply voltage of 0.9 V and a frequency of 833 MHz, while incurring only a 4.8% area overhead. Despite these modest requirements, Shift-CIM significantly accelerates multiplication operations, achieving up to 3.08× the performance improvement and a 60% reduction in energy consumption compared to state-of-the-art designs.
- 28 June 2025
- Published by Association for Computing Machinery (ACM) in ACM Transactions on Architecture and Code Optimization
- Vol. 22 (2) , 1-25
Open AccessShift-CIM: In-SRAM Alignment To Support General-Purpose Bit-level Sparsity Exploration in SRAM Multiplication
Multiplication plays a critical role in SRAM-based Computing-in-Memory (CIM) architectures. However, current SRAM-based CIMs face three major limitations. First, they do not fully exploit bit-level sparsity, resulting in unnecessary overhead in both latency and energy consumption. Second, the generation of numerous zero-dot products is superfluous. Third, the irregular organization of SRAM complicates the implementation. To address these issues, we propose Shift-CIM, a general-purpose approach that fully leverages bit-level sparsity within SRAM-based multiplications. Shift-CIM aligns the multipliers within the SRAM array, accumulating only the required dot products based on the non-zero bits of the multipliers. Shift-CIM achieves a regular SRAM organization by assembling two irregular SRAM arrays in a transposed manner. Our evaluations show that Shift-CIM is highly efficient, operating at a supply voltage of 0.9 V and a frequency of 833 MHz, while incurring only a 4.8% area overhead. Despite these modest requirements, Shift-CIM significantly accelerates multiplication operations, achieving up to 3.08× the performance improvement and a 60% reduction in energy consumption compared to state-of-the-art designs.
- Open Access
- 10 January 2025
- Published by Association for Computing Machinery (ACM) in ACM Transactions on Design Automation of Electronic Systems
- Vol. 30 (2) , 1-24
ISOAcc: In-situ Shift Operation-based Accelerator For Efficient in-SRAM Multiplication
Digital SRAM-based CIM architectures must balance three critical factors: quantized neural network bitwidth, accuracy loss, and computational efficiency, each crucial to optimizing performance and efficiency. In Domain Specific Accelerators (DSAs), flexible and specific hardware design, when incorporated with tailored Power-of-2 (P-2) quantization schemes, addresses this issue. However, in CIMs, the absence of flexible and specific hardware to support dynamic switching between general and tailored quantization schemes hinders the adoption of efficient quantization methods. In this article, we propose the I n-situ S hift O peration based Acc elerator ( ISOAcc ) for efficient SRAM-based multiplication. The key idea is to introduce transmission gates near the SRAM array to enable the selection of bits from either the same or the neighbor line when data flows from one row to another. This functionally equals a shift operation. By configuring the transmission gates array in a cascade manner, ISOAcc can support 0 to 15-bit shift with a negligible overhead. The ISOAcc can directly leverage P-2 quantization schemes in hardware, thereby greatly reducing multiplication cycles. We have chosen five well-known neural networks to evaluate ISOAcc. The evaluations show that ISOAcc achieves an average performance improvement of 3.24× and an energy reduction of 75%, compared with the state-of-the-art (SOTA) SRAM-based CIM design, Bit-Parallel.
- 10 January 2025
- Published by Association for Computing Machinery (ACM) in ACM Transactions on Design Automation of Electronic Systems
- Vol. 30 (2) , 1-24
Open AccessISOAcc: In-situ Shift Operation-based Accelerator For Efficient in-SRAM Multiplication
Digital SRAM-based CIM architectures must balance three critical factors: quantized neural network bitwidth, accuracy loss, and computational efficiency, each crucial to optimizing performance and efficiency. In Domain Specific Accelerators (DSAs), flexible and specific hardware design, when incorporated with tailored Power-of-2 (P-2) quantization schemes, addresses this issue. However, in CIMs, the absence of flexible and specific hardware to support dynamic switching between general and tailored quantization schemes hinders the adoption of efficient quantization methods. In this article, we propose the I n-situ S hift O peration based Acc elerator ( ISOAcc ) for efficient SRAM-based multiplication. The key idea is to introduce transmission gates near the SRAM array to enable the selection of bits from either the same or the neighbor line when data flows from one row to another. This functionally equals a shift operation. By configuring the transmission gates array in a cascade manner, ISOAcc can support 0 to 15-bit shift with a negligible overhead. The ISOAcc can directly leverage P-2 quantization schemes in hardware, thereby greatly reducing multiplication cycles. We have chosen five well-known neural networks to evaluate ISOAcc. The evaluations show that ISOAcc achieves an average performance improvement of 3.24× and an energy reduction of 75%, compared with the state-of-the-art (SOTA) SRAM-based CIM design, Bit-Parallel.