Make "Sense" for Computing
 |
|
Lin Zhong
Assistant Professor of ECE
|
Decades of progress in semiconductor industry, as captured by Moore's Law,
have equipped modern computing systems with increasing processing power.
This is particularly true with mobile embedded systems such as our mobile
phones and digital cameras: the Apple iPhone, released in 2007, has more
processing power than the laptop PC we had 10 years ago. In contrast, their
ability to interact with or sense the physical world, including their human
users, improves with a much less satisfactory pace. Despite their great
processing power, modern personal computing systems spend most time idle,
waiting for their users to input instructions, instead of providing useful
services in an ambient fashion. Our recent work has focused on improving the
sense of computing systems and leveraging it for more efficient computing,
which is critical to thermal and battery-constrained systems. In this talk,
we will present three recent projects in this regard. The first project
infers user information from the graphical user interface and can put a
mobile system into a power-saving mode even between two user inputs, without
being noticed by the user. The second project employs freely available
context information to select wireless interfaces for energy-efficient data
communication. The third project leverages the ultra low-power motion sensor
available in many consumer electronics to drastically improve the efficiency
of video encoding.
Wednesday, October 15, 2008
3:00 p.m. - McMurtry Auditorium, Duncan Hall
2008 ECE
Affiliates Conference
Taming Complexity in Systems