A5044795728- Real-Time Systems Scheduling
- Spacecraft Design and Technology
- Advanced Neural Network Applications
- Radiation Effects in Electronics
- Satellite Communication Systems
- Non-Invasive Vital Sign Monitoring
- Parallel Computing and Optimization Techniques
- Distributed and Parallel Computing Systems
- Neural Networks and Applications
- Advanced Memory and Neural Computing
- CCD and CMOS Imaging Sensors
- Distributed systems and fault tolerance
- Advanced SAR Imaging Techniques
- Optical Imaging and Spectroscopy Techniques
- Fusion materials and technologies
- Ferroelectric and Negative Capacitance Devices
- Embedded Systems Design Techniques
- Risk and Portfolio Optimization
- VLSI and Analog Circuit Testing
- Physical Unclonable Functions (PUFs) and Hardware Security
- Scheduling and Optimization Algorithms
- Corporate Governance and Law
- Superconducting Materials and Applications
- Magnetic confinement fusion research
- Gait Recognition and Analysis
Centre National de la Recherche Scientifique
2018-2024
Université Gustave Eiffel
2022-2024
CY Cergy Paris Université
2022-2024
École Nationale Supérieure de l'Électronique et de ses Applications
2022-2024
Equipes Traitement de l'Information et Systèmes
2022-2024
UniLaSalle Amiens (ESIEE-Amiens)
2022-2023
Institut d'Électronique et des Systèmes
2022
Laboratoire de Recherche en Informatique de Paris 6
2021
Université Paris Cité
2021
Sorbonne Université
2021
Our society will be deeply impacted by neural network inference on embedded devices. Many of them are based the use microcontroller units (MCUs) which extremely resource-scarce. The best modality to solve most computer vision problems artificial intelligence algorithms such as Convolutional Neural Networks (CNNs). Although, a CNN's accuracy implies significant costs within targeted hardware: an important energy consumption, high latency, and memory footprint. This is Tiny Machine Learning...
Neural network inference on embedded devices will have an important industrial impact our society. Embedded are ubiquitous in many fields, like human activity recognition or visual object detection. As a matter of fact, Convolutional Networks (CNNs) now the best modality to solve most computer vision problems. Although, accuracy offered by these algorithms has cost: energy consumption, high execution time, and significant memory footprint. This cost is major challenge implement CNNs within...
The computing continuum's actual trend is facing a growth in terms of devices with any degree computational capability. Those may or not include full-stack, including the Operating System layer and Application layer, just pure bare-metal solutions. In either case, reliability full system stack has to be guaranteed. It crucial provide data regarding impact faults at all levels potential hardening solutions design highly resilient systems. While most work usually concentrates on application...
CubeSats are small satellites operating in harsh space environment. In order to ensure correct functionality on board despite faults, fault tolerant techniques taking into account spatial, time and energy constraints should be considered. This paper presents a software-level solution advantage of several processors available board. Two online scheduling algorithms introduced evaluated. The results show their performances the tradeoff between rejection rate consumption. Last but not least, it...
The rapid aging of the population combined with correlation between age and increase in falls pushes us to create new ways monitor elderly. privacy radar data can respond one weaknesses existing technologies, but huge amount process becomes a challenge process. We therefore introduce first architecture allowing processing its real time. technology used is an off-the-shelf Frequency Modulated Continuous Wave Ancortek (SDR 980AD2). It followed by pre-processing chain composed Fast Fourier...
The correlation between an ageing population glob- ally and the increased risk of falling is a real challenge for health care infrastructures. This calls development new ways to monitor elderly at home. confidentiality radar data coupled with its richness information can address weaknesses existing technologies, namely, privacy acceptance. produce large quantity that needs be processed in real-time ensure timely detection fall/critical events necessary well-being elderly. We introduce...
Falls represent the main risk of injury for elderly people. One-third adults aged over 65 and half people 80 will have at least one fall a year. People should visit clinical service to detect gait difficulties. Solutions detecting daily activities are being studied more more, aiming develop complementary method early this type health as effectively possible. Non-intrusiveness in person's life problem is an important criterion, which why current research focusing on solutions involving...
Embedded systems based on Microcontroller Units (MCUs) often gather significant quantities of data and solve various issues. Convolutional Neural Networks (CNNs) have proven their effectiveness in solving computer vision natural language processing tasks. However, implementing CNNs within MCUs is challenging due to high inference costs, which varies widely depending hardware targets CNN topologies. Despite state-of-the-art advancements, no efficient design space exploration solutions handle...
Small satellites, such as CubeSats, have to respect time, spatial and energy constraints in the harsh space environment. To tackle this issue, paper presents evaluates two fault tolerant online scheduling algorithms: algorithm all tasks aperiodic (called ONEOFF) placing arriving or periodic ONEOFF & CYCLIC). Based on several scenarios, results show that performances of ordering policies are influenced by system load proportions simple double be executed. The "Earliest Deadline" Arrival Time"...
As transistors scale down, systems are more vulnerable to faults. Their reliability consequently becomes the main concern, especially in safety-critical applications such as automotive sector, aeronautics or nuclear plants. Many methods have already been introduced conceive fault-tolerant and therefore improve reliability. Nevertheless, several of them not suitable for real-time embedded since they incur significant overheads, other may be less intrusive but at cost being too specific a...
This paper is aimed at studying fault-tolerant design of the realtime multi-processor systems and in particular concerned with dynamic mapping scheduling tasks on embedded systems. The effort concentrated strategy having reduced complexity guaranteeing that, when a task input into system accepted, then it correctly executed prior to deadline. chosen method makes use primary/backup approach this describes its refinement based reduction windows within which primary backup copies can be...