Skip to content

Research

We design energy-efficient and reliable computing hardware, from sensing and event-driven processing to memory architectures and 3D integration. Our work combines architecture design, software models, and FPGA prototyping for edge AI and embedded systems.

Research areas

  • Event-driven computing: computation triggered only when data changes, including neuromorphic computing with spiking neural networks, approximate computation, and efficient learning and inference.
  • Memory and 3D integration: hybrid memory, stacked architectures, and efficient data movement.
  • Event sensing: asynchronous sensor processing, real-time perception, and low-power embedded implementation.
  • Reliable and sustainable computing: fault tolerance, thermal-aware design, and energy-aware operation.

Projects and research themes

  • Event-Driven Computing

    Low-power event-driven computing for IoT and edge devices. Our neuromorphic work investigates approximate neuron circuits and stacked memory to reduce energy use while maintaining accuracy.

    Explore event-driven computing

  • HYMEX

    Hybrid non-volatile memory and SRAM for energy-efficient neuromorphic AI. We study 3D memory integration and adaptive power management to reduce operating and idle energy.

    Explore HYMEX

  • Event Sensing

    Efficient processing of asynchronous event-camera data. Our work connects sensing, spiking neural networks, and embedded hardware for real-time perception.

    Explore event sensing

  • 3D-IC Design and Reliability

    Thermal-aware fault detection and recovery for TSV-based 3D networks-on-chip. This project provides foundations for reliable stacked computing systems.

    Explore 3D-IC research

Publications, prototypes, and opportunities