Thorsten Herfet
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Thorsten Herfet has a total of 16 co-authors. Here are the top co-authors by number of shared publications:
- Ishwar Mudraje
Saarland University, Saarbrücken, Germany
3 shared publications - Kai Vogelgesang
Saarland University, Saarbrücken, Germany
2 shared publications - Luis Gerhorst
Friedrich-Alexander-Universität Erlangen-Nürnberg, Erlangen, Germany
2 shared publications - Peter Wägemann
Friedrich-Alexander-Universität Erlangen-Nürnberg, Erlangen, Germany
2 shared publications Carsten Dennis Quint
Nanostructure research and Nanotechnology, Saarland University, Saarbrcken, Germany
1 shared publication- Chuanbeibei Shi
Saarland University, Saarbrücken, Germany
1 shared publication
Latest Publications
- Open Access
- 7 December 2026
- Published by Association for Computing Machinery (ACM)
- p. 1-9
Froxels for Production: Ray-Level Inspection and Consensus-Guided Light-Field Rendering
- 7 December 2026
- Published by Association for Computing Machinery (ACM)
- p. 1-9
Open AccessFroxels for Production: Ray-Level Inspection and Consensus-Guided Light-Field Rendering
- Open Access
- 13 June 2025
- Published by Association for Computing Machinery (ACM)
- p. 97-108
vNV-Heap: An Ownership-Based Virtually Non-Volatile Heap for Embedded Systems
- 13 June 2025
- Published by Association for Computing Machinery (ACM)
- p. 97-108
Open AccessvNV-Heap: An Ownership-Based Virtually Non-Volatile Heap for Embedded Systems
- Open Access
- 6 May 2025
- Published by Association for Computing Machinery (ACM)
- p. 1-7
Reverse Engineering the ESP32-C3 Wi-Fi Drivers for Static Worst-Case Analysis of Intermittently-Powered Systems
- 6 May 2025
- Published by Association for Computing Machinery (ACM)
- p. 1-7
Open AccessReverse Engineering the ESP32-C3 Wi-Fi Drivers for Static Worst-Case Analysis of Intermittently-Powered Systems
- Open Access
- 19 June 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 12, 86728-86738
Design of an Autonomous Intrusion Classification Device for FIDS Robustness
Fence intrusion detection system (FIDS) must ideally detect all malicious intrusions without producing false alarms arising from environmental sources. Classification of intrusions can provide security personnel with additional information regarding the level of threat. Modern FIDS are equipped with several sensing channels capable of recording fence vibrations either optically or mechanically. Introducing autonomy to each channel can improve the robustness of the FIDS as well as introduce resilience to changing conditions such as visibility/weather conditions. In this work, an autonomous accelerometer-based FIDS edge device capable of detecting and classifying intrusions is presented. The FIDS consists of three stages. First, threshold detection is used to flag potential intrusions allowing the microcontroller (MCU) to save power during idle state of fence. In the second stage, the threshold exceedence probability is evaluated to discriminate between background noise and human intrusions. An oscillator model was fitted to derive the parameters of the first two stages based on physical properties of the fence. Third, a convolutional neural network (CNN) was trained to classify the detected disturbances into two types namely rattling and climbing. The intrusion detection stages generated only a single false alarm from 17 hours of storm data while the classification stage produced a 5-fold cross validation accuracy of ≈90.5%. The intrusion detection and classification was implemented by rounding weights and using a custom CNN inference engine on an 8-bit MCU. The implementation showed no degradation in classification accuracy and no drift in sampling frequency during real-time operation.
- 19 June 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 12, 86728-86738
Open AccessDesign of an Autonomous Intrusion Classification Device for FIDS Robustness
Fence intrusion detection system (FIDS) must ideally detect all malicious intrusions without producing false alarms arising from environmental sources. Classification of intrusions can provide security personnel with additional information regarding the level of threat. Modern FIDS are equipped with several sensing channels capable of recording fence vibrations either optically or mechanically. Introducing autonomy to each channel can improve the robustness of the FIDS as well as introduce resilience to changing conditions such as visibility/weather conditions. In this work, an autonomous accelerometer-based FIDS edge device capable of detecting and classifying intrusions is presented. The FIDS consists of three stages. First, threshold detection is used to flag potential intrusions allowing the microcontroller (MCU) to save power during idle state of fence. In the second stage, the threshold exceedence probability is evaluated to discriminate between background noise and human intrusions. An oscillator model was fitted to derive the parameters of the first two stages based on physical properties of the fence. Third, a convolutional neural network (CNN) was trained to classify the detected disturbances into two types namely rattling and climbing. The intrusion detection stages generated only a single false alarm from 17 hours of storm data while the classification stage produced a 5-fold cross validation accuracy of ≈90.5%. The intrusion detection and classification was implemented by rounding weights and using a custom CNN inference engine on an 8-bit MCU. The implementation showed no degradation in classification accuracy and no drift in sampling frequency during real-time operation.
- 1 December 2017
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 31-36
Tap coefficient based cognitive framework for estimating a dynamic channel
The dynamic environment of a vehicular communication system poses a difficult task of estimating the channel at minimal complexity. A time-varying multipath channel is estimated by computationally intensive algorithms that are generally not suitable for implementation on resource limited consumer hardware. Compressed Sensing (CS) schemes have been established to provide an accurate estimate by exploiting the inherent sparsity of a wireless communication channel. Correspondingly, the Rake-Matching Pursuit (RMP) and its low complexity variant, the Gradient Rake-Matching Pursuit (GRMP) algorithm, first identify different delay taps in the environment. The Doppler is then implicitly estimated by a tracking stage of respective tap coefficients. Although their performance is encouraging even under high Doppler shifts, its adoption for a static multipath environment is excessive due to the required computational resources. A low complexity scheme, like Least Squares (LS), is sufficient to estimate and compensate such channels. The cognitive framework envisages the switch between a high mobility scheme, like RMP, and a low mobility scheme, like LS, based on the channel conditions. In this paper, an enhanced cognitive framework is proposed to interchange between the channel estimation schemes to provide an adequate Bit Error Rate (BER) performance at optimum complexity. Even though the experimentation is performed for the IEEE 802.11p standard, the proposed metrics are relevant for any Orthogonal Frequency- Division Multiplexing (OFDM) based wireless communication system.
- 1 December 2017
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 31-36
Tap coefficient based cognitive framework for estimating a dynamic channel
The dynamic environment of a vehicular communication system poses a difficult task of estimating the channel at minimal complexity. A time-varying multipath channel is estimated by computationally intensive algorithms that are generally not suitable for implementation on resource limited consumer hardware. Compressed Sensing (CS) schemes have been established to provide an accurate estimate by exploiting the inherent sparsity of a wireless communication channel. Correspondingly, the Rake-Matching Pursuit (RMP) and its low complexity variant, the Gradient Rake-Matching Pursuit (GRMP) algorithm, first identify different delay taps in the environment. The Doppler is then implicitly estimated by a tracking stage of respective tap coefficients. Although their performance is encouraging even under high Doppler shifts, its adoption for a static multipath environment is excessive due to the required computational resources. A low complexity scheme, like Least Squares (LS), is sufficient to estimate and compensate such channels. The cognitive framework envisages the switch between a high mobility scheme, like RMP, and a low mobility scheme, like LS, based on the channel conditions. In this paper, an enhanced cognitive framework is proposed to interchange between the channel estimation schemes to provide an adequate Bit Error Rate (BER) performance at optimum complexity. Even though the experimentation is performed for the IEEE 802.11p standard, the proposed metrics are relevant for any Orthogonal Frequency- Division Multiplexing (OFDM) based wireless communication system.
- 1 March 2010
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- Vol. 15 (21555044) , 1-6
Routing: Why less intelligence sometimes is more clever
Overlay architectures are the information theoretical playgrounds of the Internet. They provide significant gains in throughput by shifting complexity into the network nodes. A consequence is the requirement for smart routers between adjacent network segments. Since meanwhile audiovisual media allocate the major part of the Internet traffic, protocols as well as overlay structures have to serve Predictable Reliability under Predictable Delay (PRPD). The link-level operation mode of the AHEC architecture is an approach to fit individual channel coding parameters to inhomogeneous network segments. All link segments of a chosen network route share an overall time and reliability budget determined by the respective constraints of the application. Whereas some structures require support by all intermediate nodes, just the contribution from a limited number of routers is reasonable in the AHEC overlay: The budget distribution hits a saturation point above which configurations are suboptimal. This paper derives the saturation point with respect to the distribution of the delay constraint of audiovisual applications in the link-level operation mode of the AHEC. The result has major impact on the complexity of coding as well as the protocol parametrization.
- 1 March 2010
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- Vol. 15 (21555044) , 1-6
Routing: Why less intelligence sometimes is more clever
Overlay architectures are the information theoretical playgrounds of the Internet. They provide significant gains in throughput by shifting complexity into the network nodes. A consequence is the requirement for smart routers between adjacent network segments. Since meanwhile audiovisual media allocate the major part of the Internet traffic, protocols as well as overlay structures have to serve Predictable Reliability under Predictable Delay (PRPD). The link-level operation mode of the AHEC architecture is an approach to fit individual channel coding parameters to inhomogeneous network segments. All link segments of a chosen network route share an overall time and reliability budget determined by the respective constraints of the application. Whereas some structures require support by all intermediate nodes, just the contribution from a limited number of routers is reasonable in the AHEC overlay: The budget distribution hits a saturation point above which configurations are suboptimal. This paper derives the saturation point with respect to the distribution of the delay constraint of audiovisual applications in the link-level operation mode of the AHEC. The result has major impact on the complexity of coding as well as the protocol parametrization.