Skip to main content

Research

 

The SmartSystems laboratory conducts interdisciplinary research at the intersection of hardware systems, artificial intelligence, cybersecurity, and autonomous cyber-physical systems. Our work spans the full computing stack, from smart sensors and System-on-Chip (SoC) architectures to FPGA-accelerated cloud infrastructures, autonomous robotic systems, and zero-trust security frameworks.

 

1. FPGA Design and Optimization for Edge, Cloud, and AI Systems

We develop novel FPGA architectures, compilation frameworks, and hardware-software co-design methodologies for high-performance and energy-efficient computing across edge and cloud environments. Our research addresses resource virtualization, multi-tenancy, accelerator sharing, secure bitstream deployment, and workload-aware hardware adaptation.

Recent contributions include FPGA-accelerated cloud infrastructures supporting secure multi-tenant execution, trusted tenant-side bitstream synthesis, power-domain isolation through galvanic isolation mechanisms, and adaptive FPGA resource allocation. We further investigate AI acceleration through quantization-aware deployment, teacher-free knowledge distillation, chiplet-aware workload mapping, and reconfigurable architectures for next-generation edge intelligence.

Representative projects include:

  • · FPGA multi-tenancy and reconfigurable cloud computing.
  • · Secure FPGA virtualization and trusted compilation.
  • · HARE: Hardware-software co-design for panoramic UAV vision.
  • · D-VAEKD: Knowledge distillation for highly quantized FPGA AI accelerators.
  • · Edge-SiP Synthesizer for chiplet-based AI systems.

 

2. System-on-Chip Design Automation and Optimization

The laboratory develops methodologies, tools, and architectures for next-generation System-on-Chip (SoC) and heterogeneous computing platforms. Research focuses on design-space exploration, hardware-software co-design, security-aware synthesis, chiplet-based systems, and design automation techniques for complex embedded platforms.

Our work combines optimization algorithms, formal methods, and architectural modeling to automatically generate SoC configurations that balance performance, power, area, security, and reliability requirements. Particular emphasis is placed on emerging heterogeneous and chiplet-based architectures for AI-enabled edge computing.

Representative projects include:

  • · Security-aware SoC design automation.
  • · Multi-objective optimization of secure SoC architectures.
  • · OpenTitan-based trusted system design.
  • · Chiplet-aware workload mapping and heterogeneous integration.
  • · RTL verification and automated debugging methodologies.

 

3. Zero-Trust Security for System-on-Chip and Cyber-Physical Systems

We investigate security architectures that extend beyond traditional perimeter-based protection toward Zero-Trust principles for embedded systems, cloud infrastructures, autonomous platforms, and cyber-physical systems.

Building upon our Multi-Level Security (MLS) framework and the MEXT-SE design environment, we develop methodologies for secure-by-design hardware platforms capable of operating in adversarial environments. Current research focuses on Hardware Roots of Trust, resilient autonomous systems, secure FPGA clouds, trusted execution infrastructures, attack containment, and adaptive protection mechanisms.

Our objective is to create computing systems that remain trustworthy even in the presence of compromised components, malicious tenants, or hostile operating conditions.

Representative projects include:

  • · Zero-Trust architectures for autonomous systems.
  • · Security-aware SoC design automation.
  • · OpenTitan-enabled secure platforms.
  • · FPGA cloud security and trusted bitstream deployment.
  • · Resilient cyber-physical systems and autonomous fleets.
  • · Hardware Root of Trust and secure boot infrastructures.

 

4. UAV, Robotics, and Autonomous Cyber-Physical Systems

The laboratory develops intelligent sensing, perception, and decision-making architectures for autonomous robots, unmanned aerial vehicles (UAVs), and multi-agent systems operating in complex environments.

Our research combines distributed sensing, edge AI, computer vision, machine learning, and real-time embedded computing to enable robust human-robot collaboration and autonomous operation. Particular emphasis is placed on distributed perception, human intention prediction, collaborative robotics, panoramic UAV vision, and resilient autonomous platforms.

Representative projects include:

  • · Distributed smart-camera networks for 360° scene understanding.
  • · Human intention prediction for human-robot collaboration.
  • · Multi-robot coordination in manufacturing environments.
  • · UAV perception and panoramic vision systems.
  • · Real-time 3D scene reconstruction and tracking.
  • · Autonomous system resiliency and security.

 

5. Smart Image Sensors and Neuromorphic Vision Systems

Inspired by biological vision systems, we develop smart image sensors that integrate sensing and computation within the imaging substrate. Our architectures perform hierarchical, pixel-parallel processing directly at the sensor level, dramatically reducing data movement, latency, and power consumption.

The research combines reconfigurable computing, neuromorphic processing, deep learning acceleration, stacked silicon integration, advanced packaging, and in-sensor

computing. By moving intelligence closer to the point of data acquisition, these systems enable real-time machine vision for resource-constrained edge platforms.

Our architectures have demonstrated operation at 800 MHz while reducing dynamic power consumption by up to 96.9% through hierarchical processing and spatiotemporal redundancy elimination.

Representative projects include:

  • · Hierarchical smart image sensor architectures.
  • · Neuromorphic pixel-level processing.
  • · Stacked-silicon and 3D-integrated vision systems.
  • · In-sensor deep learning acceleration.
  • · MAGMA: Adaptive Gaussian Mixture Model Accelerator.
  • · Vision systems for low-light and extreme-environment imaging.
  • · Smart sensing for robotics and autonomous systems.

 

Cross-Cutting Research Theme

A unifying theme across all laboratory activities is the development of secure, intelligent, and adaptive computing systems, spanning the entire stack from smart sensors and edge devices to FPGA clouds, autonomous robots, and cyber-physical infrastructures. By integrating advances in reconfigurable computing, AI acceleration, system architecture, and zero-trust security, the laboratory aims to create next-generation platforms capable of operating efficiently, safely, and autonomously in increasingly complex environments.

 

FUNDING

Our research is supported by funding and donations from the following organizations:

  • · National Science Foundation
  • · Airforce Research Lab
  • · German Research Association
  • · German-French University
  • · European Union
  • · Xilinx
  • · Altera

COLLABORATION

We actively collaborate with the following organizations and companies:

  • · R-Dex Systems (www.r-dex.com)
  • · ZeSys (Center for Embedded Systems in Berlin, Germany) www.zesys.de
  • · Frohle Network in Germany
  • · Citi bank