Building Impossible Systems

Speaker:	Dr. Lin GU
		Department of Computer Science & Engineering
		Hong Kong University of Science & Technology

Title:		"Building Impossible Systems"

Date:		Monday, 13 Oct 2008

Time:		4:00pm - 5:00pm

Venue:		Lecture Theatre F
		(Leung Yat Sing Lecture Theatre, near lifts 25/26)
		HKUST

Abstract:

In this talk, I will introduce three systems -- the Extended VigilNet, the
t-kernel, and the Vivaldi/8. They are all related to wireless sensor
networks. The Extended VigilNet has been completed, the t-kernel is an
ongoing work, and the Vivaldi/8 project is just starting. But there is a
common theme in these projects -- they attempt to solve some difficult
research problems so that we can build a system previously considered
impossible.

The Extended VigilNet is a self-organizing surveillance sensor network
that can detect and classify four types of targets -- people, people with
ferrous objects, small vehicles, and large vehicles. The amount of data it
processes in one minute exceeds what some other types of sensor networks
may collect in their whole lifetime of several months. Moreover, all the
processing is done in the network with strictly resource constrained
devices.

The t-kernel is an operating system that uses pure software methods to
efficiently implement OS protection and virtual memory. Many
microcontrollers used in very-low-power embedded systems lack the hardware
features (privilege, address translation logic, etc.) traditionally
required for OS protection and virtual memory. Hence, systems with these
microcontrollers had to use simplistic operating systems, giving rise to
many difficulties in programming and system management. The t-kernel
removes this limitation and enables very-low-power systems to use advanced
OS features.

The Vivaldi/8 project aims at developing a sensor network system that can
operate for multiple years with "continuous vigilance", meaning that the
system monitors unsynchronized external events in a "non-stop" manner.
Such a long lifetime more than doubles what the best currently known
system can achieve, and will significantly enhance the functionality
of sensor network systems.

***************************
Biography:

Lin Gu received B.S. from Fudan University, M.S. from Peking University,
and Ph.D. in Computer Science from the University of Virginia. Centering
on networked embedded systems, his research interest includes operating
systems, energy efficient computing, and distributed systems. He was a
major contributor to a series of large-scale wireless sensor networks in
DARPA's NEST project. He wrote a new OS kernel, the t-kernel, to overcome
the limitations of traditional sensor network operating systems. He
published a number of academic papers, including two "Best Paper Award"
winners. Prior to joining HKUST, Lin Gu worked at Google on global scale
web systems. He is now working at the CSE Department of Hong Kong
University of Science and Technology (HKUST) as an Assistant Professor.