Thomas Lange
Research Subjects
Thomas Lange has a total of 27 co-authors. Here are the top co-authors by number of shared publications:
Dan Alexandrescu
RoC Technologies, Grenoble, France
3 shared publicationsMaximilien Glorieux
RoC Technologies, Grenoble, France
3 shared publications- Aneesh Balakrishnan
System Architectures, IHP-Leibniz-Institut fr Innovative Mikroelektronik, Frankfurt, Germany
2 shared publications - Dan Alexandrescu
IROC Technologies, Grenoble, France
2 shared publications - Luca Sterpone
Dipartimento di Automatica e Informatica, Politecnico di Torino, Torino, Italy
2 shared publications A-Duong In
IROC Technologies, 38000 Grenoble, France
1 shared publication
Latest Publications
- 4 October 2021
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 1-6
On Antagonism Between Side-Channel Security and Soft-Error Reliability in BNN Inference Engines
Recently, several research works have emphasized the problem of stealing the intellectual property of trained Machine Learning (ML) models from hardware neural network inference engines spotlighting Binarized Neural Networks (BNNs). The binary operations in BNNs can be executed bitwise, which notably saves storage memory, reduces the execution time and power and, therefore, makes them convenient for implementation in hardware. Unfortunately, these advantages may also enable a vulnerability to Differential Power Analysis (DPA) side-channel attacks, which, in turn, necessitates dedicated masking techniques to protect the models. Notably, the recent BNN hardware inference engines are being increasingly adopted for critical applications and demand, along with security, also high levels of in-filed reliability throughout their lifetime. The state-of-the-art power side-channel masking in BNNs implies glitch-resistant structures, such as Trichina AND gates and sequences of flip-flops, and may create soft-error reliability issues that are currently overlooked in the literature. This paper presents an analysis for the soft-error reliability jeopardy by the security countermeasures in hardware implementations of BNN inference engines. Our work reveals a steep increase (hundreds of times) of vulnerability to single-event effects, introduced by the state-of-the-art security enhancement techniques, and emphasizes the interdependency of the design’s reliability and security aspects.
- 4 October 2021
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 1-6
On Antagonism Between Side-Channel Security and Soft-Error Reliability in BNN Inference Engines
Recently, several research works have emphasized the problem of stealing the intellectual property of trained Machine Learning (ML) models from hardware neural network inference engines spotlighting Binarized Neural Networks (BNNs). The binary operations in BNNs can be executed bitwise, which notably saves storage memory, reduces the execution time and power and, therefore, makes them convenient for implementation in hardware. Unfortunately, these advantages may also enable a vulnerability to Differential Power Analysis (DPA) side-channel attacks, which, in turn, necessitates dedicated masking techniques to protect the models. Notably, the recent BNN hardware inference engines are being increasingly adopted for critical applications and demand, along with security, also high levels of in-filed reliability throughout their lifetime. The state-of-the-art power side-channel masking in BNNs implies glitch-resistant structures, such as Trichina AND gates and sequences of flip-flops, and may create soft-error reliability issues that are currently overlooked in the literature. This paper presents an analysis for the soft-error reliability jeopardy by the security countermeasures in hardware implementations of BNN inference engines. Our work reveals a steep increase (hundreds of times) of vulnerability to single-event effects, introduced by the state-of-the-art security enhancement techniques, and emphasizes the interdependency of the design’s reliability and security aspects.
- 19 October 2020
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 1-6
Hardware Accelerator Design with Supervised Machine Learning for Solar Particle Event Prediction
The intensity of cosmic radiation can differ over five orders of magnitude within a few hours or days during Solar Particle Events (SPEs), thus increasing the probability of Single-Event Upsets (SEUs) in space applications for several orders of magnitude. Therefore, it is vital to employ the early detection of the SEU rate changes in order to ensure timely activation of the radiation hardening measures. In this paper, a hardware accelerator for forecasting the SPEs by the prediction of in-flight SEU variation is proposed. An embedded on-chip SRAM is used as the real-time particle detector. The dedicated hardware accelerator implements a supervised machine learning model to forecast the SRAM SEUs one hour in advance with fine-grained hourly tracking of SEU variations during SPEs as well as under normal conditions. The whole design is intended for a highly dependable and self-adaptive multiprocessing system employed in space applications. Therefore, the target system can drive the appropriate radiation hardening mechanisms before the onset of high radiation levels.
- 19 October 2020
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 1-6
Hardware Accelerator Design with Supervised Machine Learning for Solar Particle Event Prediction
The intensity of cosmic radiation can differ over five orders of magnitude within a few hours or days during Solar Particle Events (SPEs), thus increasing the probability of Single-Event Upsets (SEUs) in space applications for several orders of magnitude. Therefore, it is vital to employ the early detection of the SEU rate changes in order to ensure timely activation of the radiation hardening measures. In this paper, a hardware accelerator for forecasting the SPEs by the prediction of in-flight SEU variation is proposed. An embedded on-chip SRAM is used as the real-time particle detector. The dedicated hardware accelerator implements a supervised machine learning model to forecast the SRAM SEUs one hour in advance with fine-grained hourly tracking of SEU variations during SPEs as well as under normal conditions. The whole design is intended for a highly dependable and self-adaptive multiprocessing system employed in space applications. Therefore, the target system can drive the appropriate radiation hardening mechanisms before the onset of high radiation levels.
- Open Access
- 1 July 2019
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- Vol. 12, 7-14
Machine Learning to Tackle the Challenges of Transient and Soft Errors in Complex Circuits
The Functional Failure Rate analysis of today’s complex circuits is a difficult task and requires a significant investment in terms of human efforts, processing resources and tool licenses. Thereby, de-rating or vulnerability factors are a major instrument of failure analysis efforts. Usually computationally intensive fault-injection simulation campaigns are required to obtain a fine-grained reliability metrics for the functional level. Therefore, the use of machine learning algorithms to assist this procedure and thus, optimising and enhancing fault injection efforts, is investigated in this paper. Specifically, machine learning models are used to predict accurate per-instance Functional De-Rating data for the full list of circuit instances, an objective that is difficult to reach using classical methods. The described methodology uses a set of per-instance features, extracted through an analysis approach, combining static elements (cell properties, circuit structure, synthesis attributes) and dynamic elements (signal activity). Reference data is obtained through first-principles fault simulation approaches. One part of this reference dataset is used to train the machine learning model and the remaining is used to validate and benchmark the accuracy of the trained tool. The presented methodology is applied on a practical example and various machine learning models are evaluated and compared.
- 1 July 2019
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- Vol. 12, 7-14
Open AccessMachine Learning to Tackle the Challenges of Transient and Soft Errors in Complex Circuits
The Functional Failure Rate analysis of today’s complex circuits is a difficult task and requires a significant investment in terms of human efforts, processing resources and tool licenses. Thereby, de-rating or vulnerability factors are a major instrument of failure analysis efforts. Usually computationally intensive fault-injection simulation campaigns are required to obtain a fine-grained reliability metrics for the functional level. Therefore, the use of machine learning algorithms to assist this procedure and thus, optimising and enhancing fault injection efforts, is investigated in this paper. Specifically, machine learning models are used to predict accurate per-instance Functional De-Rating data for the full list of circuit instances, an objective that is difficult to reach using classical methods. The described methodology uses a set of per-instance features, extracted through an analysis approach, combining static elements (cell properties, circuit structure, synthesis attributes) and dynamic elements (signal activity). Reference data is obtained through first-principles fault simulation approaches. One part of this reference dataset is used to train the machine learning model and the remaining is used to validate and benchmark the accuracy of the trained tool. The presented methodology is applied on a practical example and various machine learning models are evaluated and compared.
- 1 April 2019
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 1-6
Functional Failure Rate Due to Single-Event Transients in Clock Distribution Networks
With technology scaling, lower supply voltages, and higher operating frequencies clock distribution networks become more and more vulnerable to transients faults. These faults can cause circuit-wide effects and thus, significantly contribute to the functional failure rate of the circuit. This paper proposes a methodology to analyse how the functional behaviour is affected by Single-Event Transients in the clock distribution network. The approach is based on logic-level simulation and thus, only uses the register-transfer level description of a design. Therefore, a fault model is proposed which implements the main effects due to radiation-induced transients in the clock network. This fault model enables the computation of the functional failure rate caused by Single-Event Transients for each individual clock buffer, as well as the complete network. Further, it allows the identification of the most vulnerable flip-flops related to Single-Event Transients in the clock network.The proposed methodology is applied in a practical example and a fault injection campaign is performed. In order to evaluate the impact of Single-Event Transients in clock distribution networks, the obtained functional failure rate is compared to the error rate caused by Single-Event Upsets in the sequential logic.
- 1 April 2019
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 1-6
Functional Failure Rate Due to Single-Event Transients in Clock Distribution Networks
With technology scaling, lower supply voltages, and higher operating frequencies clock distribution networks become more and more vulnerable to transients faults. These faults can cause circuit-wide effects and thus, significantly contribute to the functional failure rate of the circuit. This paper proposes a methodology to analyse how the functional behaviour is affected by Single-Event Transients in the clock distribution network. The approach is based on logic-level simulation and thus, only uses the register-transfer level description of a design. Therefore, a fault model is proposed which implements the main effects due to radiation-induced transients in the clock network. This fault model enables the computation of the functional failure rate caused by Single-Event Transients for each individual clock buffer, as well as the complete network. Further, it allows the identification of the most vulnerable flip-flops related to Single-Event Transients in the clock network.The proposed methodology is applied in a practical example and a fault injection campaign is performed. In order to evaluate the impact of Single-Event Transients in clock distribution networks, the obtained functional failure rate is compared to the error rate caused by Single-Event Upsets in the sequential logic.
- Open Access
- 1 August 2018
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 120-126
A Novel Error Rate Estimation Approach forUltraScale+ SRAM-based FPGAs
SRAM-based FPGA devices manufactured in FinFET technologies provide performances and characteristics suitable for avionics and aerospace applications. The estimation of error rate sensitivity to harsh environments is a major concern for enabling their usage on such application fields. In this paper, we propose a new estimation approach able to consider the radiation effects on the configuration memory and logic layer of FPGAs, providing a comprehensive Application Error Rate probability estimation. Experimental results provide a comparison between radiation test campaigns, which demonstrates the feasibility of the proposed solution.
- 1 August 2018
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 120-126
Open AccessA Novel Error Rate Estimation Approach forUltraScale+ SRAM-based FPGAs
SRAM-based FPGA devices manufactured in FinFET technologies provide performances and characteristics suitable for avionics and aerospace applications. The estimation of error rate sensitivity to harsh environments is a major concern for enabling their usage on such application fields. In this paper, we propose a new estimation approach able to consider the radiation effects on the configuration memory and logic layer of FPGAs, providing a comprehensive Application Error Rate probability estimation. Experimental results provide a comparison between radiation test campaigns, which demonstrates the feasibility of the proposed solution.
- Open Access
- 1 July 2018
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 1-5
Single-Event Characterization of Xilinx UltraScale+® MPSOC under Standard and Ultra-High Energy Heavy-Ion Irradiation
Heavy-Ion irradiation of a Xilinx Ultrascale+ MPSOC was performed to measure Single-Event-Latch-up and Single-Event-Upset Cross-Sections. Additionally, irradiation with a ultra high energy xenon beam shows similar upset sensitivity.
- 1 July 2018
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 1-5
Open AccessSingle-Event Characterization of Xilinx UltraScale+® MPSOC under Standard and Ultra-High Energy Heavy-Ion Irradiation
Heavy-Ion irradiation of a Xilinx Ultrascale+ MPSOC was performed to measure Single-Event-Latch-up and Single-Event-Upset Cross-Sections. Additionally, irradiation with a ultra high energy xenon beam shows similar upset sensitivity.