|
Research
With continued scaling of IC technology, power efficiency and reliability are emerging as major challenges for the electronics and computing industry. In particular, the computation and storage demands have far exceeded the available resources in mobile and deeply embedded devices. Our research focuses on the following two areas: (1) We explore multi-level (device, circuit, architecture) hardware design solutions for developing power efficient and reliable VLSI Circuits and Systems. The developed methodologies and techniques are applicable to a wide range of applications in low-power mobile multimedia, high-performance computing with multi/many-core microprocessors, and application-driven design and (2) We focus on intelligent energy-efficient storage system for big-data applications by introducing data-awareness and user-awareness to hardware design process in mobile and and deeply embedded devices (e.g. mobile devices, unmanned aircraft systems, and Internet-of-Things).
Active Projects
Data-Mining Assisted Application-Driven Intelligent Memory System (NSF CCF #1514780)
 Low power embedded memory techniques have been widely investigated in the literature for general-purpose applications, but the improvements in power efficiency are often achieved with significant design complexity and power penalty for voltage regulations or boosting circuits. In this project, we use advanced data mining techniques particularly suited to mobile video data applications into the hardware design process to achieve intelligent memory with high power efficiency. We have used rule mining technique to perform statistical pattern analysis on mobile video storage data and obtained Cr and Cb patterns and based on them, we have developed intelligent memory with high power efficiency and zero area overhead.
Luminance Adaptive Mobile Video Storage and Processing System
 Nowadays, mobile video streaming enables people access to digital video content anytime, anywhere. We explore hardware power saving opportunities
by introducing user-experience-awareness into hardware design process. We have developed a Viewing luminance Context Aware SRAM (VCAS) to maximize the power efficiency while satisfying the user’s expectation, by introducing hardware-failure in viewing surroundings with high noise tolerance. We have built the video processing system and fabricated IC chips to test the effectiveness of VCAS. Most recently, we are collaborating psychologiests to study the impact of luminance on user's viewing experience and connect it to low-power design process.
Bring Offline Mining to Online Learning System: Intelligent Efficient Memory for for Embedded Real-Time Learning
 Enhancing energy efficiency and performance of embedded storage is of paramount importance to support online machine learning. While considering various sources of uncertainty and balancing multiple objectives of a system, the hardware implementation of machine learning needs continuous model updating and intensive memory access, and embedded memory is critical for the overall performance and energy efficiency. On-chip SRAM dominates the area and power consumption of the entire system at 56% and 60%, respectively. In this project, we face the storage challenge of embedded real-time learning systems by introducing synaptic data awareness using offline data mining techniques to extract useful knowledge from data itself for hardware optimization.
Real-Time On-Board Image/Video Processing/Storage/Detection System for Unmanned Aerial Vehicle (UAV)
 Unmanned Aerial Vehicles (UAVs) are emerging as a cost-effective and robust tool for many applications such as precision agriculture and oil/gas industry. However, current UAVs suffer from short flight life, huge image/video data size, and slow processing speed. We are working on power-efficient and high-speed on-board image/video processing, storage, detection systems, for unmanned vehicles targeting for different applications such as structure monitoring (e.g. wind turbines, railways, bridges and pipelines) and precision agriculture (e.g. crop population, livestock health monitoring, smart bin monitoring system).
More Details for Unmanned Aerial Vehicle (UAV) project
Chip Galary
We tape out and test several IC chips every year to verify the effectiveness of techniques we developed. Several recent chip micrographs are shown below.
Our Sponsors
|
