Wireless
Networking
Group Research
The research in Wireless Networking Group spans over various issues in
sensor networks and more general wireless networks. Sensor network is a
new computing paradigm that has not been understood much. Our goal is
to gain more in-depth understanding about many aspects of the sensor
network; eventually we may have a panoramic view about how to design
sensor network protocols effectively. More precisely, we have been
working on the following aspects.
NEW!!!! Current
Research Projects.
Snoogle
We use google everyday to search information on the Internet, but how
do we search the objects in the physical world? For example, how do we
find Thomas Jefferson's statue in College of William and Mary, how do
we find your favorite CD with a specific title, how do I find where I
put the tax return form, my computer architecture book? You use
Snoogle. Snoogle is a search engine for the physical world. With the
textual annotation to the objects by attaching RFID and sensors, the
searching capability is extended to more tangible world we live in. We
are not quite there yet, but Snoogle gives the architecture and
framework for an experimental searching system for physical object
searching.
Microsearch
William Blake in his famous poem wrote:
To see a World in a Grain of Sand
And a Heaven in a Wild Flower,
Hold Infinity in the palm of your hand
And Eternity in an hour.
Can we really search a world in a grain of sand? Nowadays, search
pretty much means google. You use google desktop search to search
whatever stored in your computer, but how far can this go? How small
the computer can be so that you can still do search? Can we search the
dancing angels on the tip of a needle (just joking)? Microsearch is a
textual search system for tiny devices. It can run on small sensors
(eventually smart dusts).
Sensor and RFID security
Someone asked the question: "How do the bacteria solve security
problems?" This question indeed fascinates us. Honestly we don't know
the answer, but we do know a bit how to do security for an ant or a bug
if they carry small sensors or RFID tags. We have built a very
efficient public key suite for sensors, which takes only around 1
second to do normal cryptographic operations. What does that mean? It
means, when an ant comes back home, it waits one second to be
authenticated so that it will be allowed to enter. Ants move slow, so
we think one second is pretty good for them. If they carry RFID tag,
and have no Internet connection, but they do need RFID readers, they
can use our new authentication and search protocols. More amazingly,
ants can do identity-based cryptography now. If one ant intends to send
a message to his girlfriend, he does not need to know her public
key--he merely uses her name to derive her public key. What a
convenience for ant empire!
Sensor Network Storage
Storage
Sensor Network Security
We are currently exploring new methods in enhancing sensor network
security. We have developed a suite of public-key based security
primitives for TelosB and Mica Motes.
Partial Coverage
Sensor network coverage is to guarantee that obejects present in the
network field will be detected. Partial coverage has not been
investigated much while most of the people have worked on how to
provide full coverage in a sensor network. Partial coverage, compared
to full coverage, can increase the sensor network lifetime at the cost
of decreased sensing quality. Many times, network lifetime is more
important than the sensing quality. How to balance the lifetime and
different sensing qualities is our main research. We have worked on a
framework to analyze the fundamental bounds of the partial coverage
performance. We have given a simple but powerful random sensing
schedule protocol that yields a very good performance. To achieve a
good worst case performance, we have designed several wave-sensing
schedule protocols as well that can guarantee the maximal detection
time for any object in a network field. Wave-sensing schedule takes the
analogy of a line sweeping through the field for object detection to
enforce the sensors touched by the sweeping line to wake up and go back
to sleep after the line passes.
Sensor Network Navigation
Sensor network navigation is a project to integrate several
functionalities of a sensor network, including sensing, communication,
data disssemination and query, into a useful application. The network
uses very simple computation and communication to distribute the map
information in the network. A user in a sensor field can query the
network about where to go by avoiding the dangerous areas (fire,
landmine, enemy, etc.). We analyzed the protocol, and tested the
navigation protocol on a sensor network platform composed of 50 Motes.
The recent improvement is to extend the original protocol to a more
cost-effective Voronoi-diagram-based protocol, which requires less
network search space for path optimization. Another interesting
extension is task scheduling and assignment through the coordination of
the sensor network.
Sensor Network Clock Synchronization
Clock synchronization is a basic service for sensor network
applications, used in vehicle tracking, localization, power saving, and
so on. We designed very simple localized and fault-tolerant protocols
for global clock synchronization for the entire network. The
protocols are based on a diffusion process in which each node only
conducts computation by knowing its neighbors, that is, exchanges its
clock reading with its neighbors by considering the clock reading
difference. We also show how to achieve synchronization in a localized
way in the presence of the Byzantine fault. In this framework, sensor
nodes may lie about its clock reading and may also corrupt the message
relayed through it.
Power-aware Routing in Ad-hoc and Sensor Networks
Energy consumption is one of the most important performance metrics in
wireless ad-hoc and sensor networks. In this study, we proposed
power-aware routing protocols to maximize the lifetime of a network.
The algorithm does not assume prior knowledge of the message sequence
because for ad-hoc network applications this sequence is dynamic and
depends on sensed values and goals communicated to the system as
needed. It selects a compromise between the path with the minimum power
consumption and the path that maximizes the minimal residual power in
the nodes of the network. The goal is to avoid the nodes that have been
depleted of power and at the same time choose a path that has less
aggregate power consumption. We also extended the algorithm to a
hierarchical structure and designed an algorithm that reduces amount of
control messages.
Mobile Sensor Networks/Connectivity Maintenance in Disconnected
Networks
Randomly deployed networks and mobile networks undergo disconnection or
network partition all the time; maintaining connectivity is crucial to
the normal function of a network. We proposed to use mobility to
achieve network connectivity or information delivery even though time
delay is introduced in this scenario. We consider two types of
mobilites: controlled mobility and natural mobility. Controlled
mobility means a network node can move under its own control, while
natural mobility is characterized by a node that can only move by the
control of its environment.