photo



Lin Gu

Assistant Professor

Department of Computer Science and Engineering
Hong Kong University of Science and Technology
Clear Water Bay, Kowloon, Hong Kong

Office: 3562 Tel: +(852)2358-6991
l i n g u @ c s e . u s t . h k (Remove the spaces)


Quick links

COMP 362
COMP 660L
baijia.info (paper discussion)
Virtk (a.k.a. t-kernel)
VDB debugging utility
VigilNet
Non-academic links

Lin Gu is an assistant professor at the Department of Computer Science and Engineering. He received B.S. from Fudan University (1991), M.S. from Peking University (2001), and Ph.D. from the University of Virginia (2006). His current research interests include large-scale distributed systems and cloud computing, operating systems, wireless sensor networks, and energy-efficient computing.


Courses

COMP 362    Computer Communication Networks II, Fall 2009
COMP 660L Topics in cloud computing, Fall 2009
COMP 381    Design and Analysis of Computer Architectures, Fall 2008


Research projects

System software for cloud computing
Novel solutions are needed in the system software design for cloud computing systems. Neither traditional OS services nor existing search-oriented designs can provide the functionality, performance, and usability for the next-generation computing. Our poject studies how to design cloud-based system software, including the architecture, storage solutions, auto-parallelization utilities, and network organization.

Virtk (a.k.a. t-kernel): OS design for general low-power systems
Traditional OS services meet enormous difficulty in low-power embedded systems because of the stringent resource constraints. It is a research challenge to design OS services that are portable among a large spectrum of embedded platforms. I am working on a new OS kernel that supports virtual memory, preemptive scheduling, and reliable OS protection without traditional hardware support.

Lightweight detection and classification in wireless sensor networks
While a variety of sensors have been incorporated into a spectrum of wireless sensor network ( WSN) platforms, traditional signal processing algorithms, however, often prove too complex for energy-and-cost-effective WSN nodes. It is a challenge to to design efficient sensing and classification algorithms that achieve reliable sensing performance on energy-and-cost-effective hardware without special powerful nodes in a continuously changing physical environment. We have developed a lightweight detection and classification architecture and successfully used it in VigilNet.

Energy efficiency and power management
Novel hardware, architecture and middleware solutions are needed to make wireless sensor network sustain a long period of time. I am working on implementing radio-triggered hardware which greatly enhance the network nodes' wake-up capability and enable efficient power management services.

Network protocols
TDMA based MAC and data-link layer protocols. Low-latency routing protocols.


Selected publications (For all publications, visit Complete publication list.)

Conference Papers

Journals


Miscellaneous


ACM

IEEE

CRA

 

 

 

 

TinyOS

Slashdot

 

 

 

 

 

Notes