My research passion lies in resolving real world problems
and creating practical solutions that can shape the state-of-the-art and
assist the lives of others. In accordance with my research philosophy,
my research interests lie broadly in wireless and sensor
networking, distributed systems and
real-time computing.
Currently, my research is focusing on
Wireless Sensor Networks (WSNs),
a new information paradigm based on the collaboration of a large number of
self-organized sensing nodes. These networks form the basis for many
promising applications such as immersive gaming, intelligent battlefields,
hazard response systems, smart hospitals and learning environments. My
research is mainly system-oriented - building
practical systems. Specifically we
are aiming at four major interleaved efforts: 1) Integrated sensor
systems such as VigilNet , 2) sensor network
middleware service such as
localization,
networking,
coverage. 3) in-situ sensor
system modeling and enhancement, and
4) architecture,
system, language and development support for large-scale integrated sensor
network systems. The ultimate research goal is to contribute
to the design, implementation, deployment, use and maintenance of
practical sensor systems.
In
road networks, sensor nodes are deployed sparsely (hundreds of meters
apart) to save costs. This makes the existing localization solutions
based on the ranging ineffective. To address this issue, we introduce an
Autonomous Passive Localization (APL) scheme. Our work is inspired by
the fact that vehicles move along routes with a known map. Using
vehicle-detection timestamps, we can obtain distance estimates between
any pair of sensors on roadways to construct a virtual graph composed of
sensor identifications (i.e., vertices) and distance estimates (i.e.,
edges). The virtual graph is then matched with the topology of road map,
in order to identify where sensors are located in roadways. We evaluate
our design in local roadways and simulated environments, where we found
no location matching error, even with a maximum sensor time
synchronization error of 0.3sec and the vehicle speed deviation of
10km/h..
This system has
been reported in Infocom 2008. [PDF]
proposes two in-network distributed algorithms, namely Minimum Resource
and Optimal Area that aim to preserve personal privacy in such areas
while maintaining the monitoring functionality. Both algorithms reply on
the well established privacy concept of k-anonymity; Although both
proposed algorithms provide same privacy guarantees, the Minimum
Resource aims to do so with minimum possible number of exchanged
messages between sensor nodes while the Optimal Area algorithm aims to
maintain the highest quality of monitoring functionality. Furthermore,
to accommodate the system users mobility, we propose an incremental
maintenance scheme for both algorithms that aims to avoid redundant
reevaluation of privacy guarantees. The proposed system is evaluated
with a network of 39 MICAz motes on a physical test-bed, and an
extensive simulation of 1,000 sensor nodes. [Demo Video]
is
a large indoor sensor network test-bed,
supporting up to 360 nodes. The whole test-bed is composed of six 4 feet
by 8 feet boards. Each board in the system can be used as an individual
sub-system, because each board is designed to be separately powered,
separately controlled and separately metered. Each individual
board can support up to 60 nodes, therefore, the whole system can
support up to 360 nodes working simultaneously. In the first phase of
construction, three high-end HIT HITCPX1250 projectors are used to
generate event (it is capable to create mirage ). In the
second phase of construction, motorize objects are introduced to
create another sets of mobile targets. The ultimate goal of this testbed
is to allow researchers to conduct all kinds of system research locally
and remotely with realistic sensing modality as inputs. The first
phase of construction is finished during 2007. In the second phase,
mobility support will be added.
Multi-Sequence Positioning (MSP) is designed and implemented for sensor
node localization in outdoor environments. The novel idea behind MSP is
to reconstruct and estimate two-dimensional (or 3D) location information
for each sensor node by processing multiple easy-to-get one-dimensional
node sequences obtained through a loosely guided event distribution. We
have realized the MSP idea through two physical systems (indoor and
outdoor version) with totally over 60 MICAZ motes. This evaluation
demonstrates that MSP can achieve sub-feet-level accuracy, requiring
neither additional hardware on sensor nodes nor precise event
distribution. It also provides a nice tradeoff between physical costs
(anchors) with soft cost (events) while maintaining localization
accuracy.[Demo Video]
This system has
been reported in SenSys 2007. [PDF
]
Despite the well-known fact that
in reality sensing patterns are highly irregular, researchers continue
to develop protocols with simplifying assumptions of circular 0/1
sensing models. In this project, we design and implement two Sensing
Area Modeling (SAM) techniques useful in the real world. P-SAM provides
accurate sensing area models for individual nodes using controlled or
monitored events, while V-SAM provides continuous sensing similarity
models using natural events in an environment. Evaluation under
real-world settings reveal several serious issues concerning circular
models, and demonstrate significant improvements in several applications
when SAM is used.
This system has
been reported in SenSys 2007. [PDF
]
defines a Unified Sensing Coverage Architecture, which features three novel ideas: Asymmetric Architecture, Generic
Switching and Global Scheduling. uSense provides sensing coverage
through a creative separation of scheduling from switching. We design
and implement sophisticated scheduling algorithms externally and
represent such intelligence with a lightweight generic switching
algorithm running at resource-constrained sensor nodes. As an instance
of these scheduling algorithms, we propose a novel two-level scheduling
algorithm, called uScan. We evaluate our architecture with a network of
30 MicaZ motes, an extensive simulation with 10,000 nodes. The results
indicate that uSense is a promising architecture to support flexible and
efficient coverage in sensor networks.
This system has been reported in ICDCS 2007
[PDF
] and
MobiCom SRC competition
2006
In this
project, we design and test a methodology for navigation of mobile
wireless sensor networks and fast target acquisition without a map,
called GraDrive. Our approach integrates per-node prediction with global
collaborative prediction to estimate the position of a stationary target
and to direct mobile nodes towards the target along the shortest path.
We demonstrate that a high accuracy in localization can be achieved much
faster than existing navigation models without any assistance from
stationary sensor networks.
This system won the
best paper
award in the 2nd International Conference on Mobile Ad-hoc and Sensor Networks (MSN
2006) [PDF
]
In this project we
design and implement a framework, called StarDust, for wireless sensor
network localization based on passive optical components. In the
StarDust framework, sensor nodes are equipped with optical
retro-reflectors. An aerial device projects light towards the deployed
sensor network, and records an image of the reflected light. An image
processing algorithm is developed for obtaining the locations of sensor
nodes. For matching a node ID to a location we propose a
constraint-based label relaxation algorithm. We propose and develop
localization techniques based on four types of constraints: node color,
neighbor information, deployment time for a node and deployment location
for a node.
uses the spatio-temporal properties of well
controlled events in the network (e.g., light), to obtain the locations
of sensor nodes. We demonstrate that a high accuracy in localization can
be achieved without the aid of expensive hardware on the sensor nodes,
as required by other localization systems. Through performance
evaluations of a real system deployed outdoors, we obtain a 20cm
localization error. A sensor network, with any number of nodes, deployed
in a 2500m2 area, can be localized in under 10 minutes, using a device
that costs less than $1000. To the best of our knowledge, this is the
first report of a sub-meter localization error, obtained in an outdoor
environment, without equipping the wireless sensor nodes with
specialized ranging hardware.
This system has been reported in
SenSys05 [PDF
][Demo Video]
is one of the major efforts in
the sensor network community to build an integrated sensor network
system for surveillance missions. The focus of this effort is to acquire
and verify information about enemy capabilities and positions of hostile
targets. Such missions often involve a high element of risk for human
personnel and require a high degree of stealthiness. Hence, the ability
to deploy unmanned surveillance missions, by using wireless sensor
networks, is of great practical importance for the military. In this
work, we design and implement a complete running system, called
VigilNet, for energy-efficient surveillance. It currently consists about 40,000
lines of NesC and Java code, running on XSM, Mica2 and Mica2dot
platforms. The complete system is designed to scale to at least 1000 XSM
motes and cover minimal 100x1000 square meters to ensure operational
applicability. We evaluate middleware and system performance extensively
on a network of 203 MICA2 motes.
Various aspects of the VigilNet system have been reported in
MobiSys04 [PDF],
SenSys05 [PDF] , Infocom05
[PDF], RTAS06 and TECS
[PDF]
In
road networks, sensor nodes are deployed sparsely (hundreds of meters
apart) to save costs. This makes the existing localization solutions
based on the ranging ineffective. To address this issue, we introduce an
Autonomous Passive Localization (APL) scheme. Our work is inspired by
the fact that vehicles move along routes with a known map. Using
vehicle-detection timestamps, we can obtain distance estimates between
any pair of sensors on roadways to construct a virtual graph composed of
sensor identifications (i.e., vertices) and distance estimates (i.e.,
edges). The virtual graph is then matched with the topology of road map,
in order to identify where sensors are located in roadways. We evaluate
our design in local roadways and simulated environments, where we found
no location matching error, even with a maximum sensor time
synchronization error of 0.3sec and the vehicle speed deviation of
10km/h..
proposes two in-network distributed algorithms, namely Minimum Resource
and Optimal Area that aim to preserve personal privacy in such areas
while maintaining the monitoring functionality. Both algorithms reply on
the well established privacy concept of k-anonymity; Although both
proposed algorithms provide same privacy guarantees, the Minimum
Resource aims to do so with minimum possible number of exchanged
messages between sensor nodes while the Optimal Area algorithm aims to
maintain the highest quality of monitoring functionality. Furthermore,
to accommodate the system users mobility, we propose an incremental
maintenance scheme for both algorithms that aims to avoid redundant
reevaluation of privacy guarantees. The proposed system is evaluated
with a network of 39 MICAz motes on a physical test-bed, and an
extensive simulation of 1,000 sensor nodes. [
is
a large indoor sensor network test-bed,
supporting up to 360 nodes. The whole test-bed is composed of six 4 feet
by 8 feet boards. Each board in the system can be used as an individual
sub-system, because each board is designed to be separately powered,
separately controlled and separately metered. Each individual
board can support up to 60 nodes, therefore, the whole system can
support up to 360 nodes working simultaneously. In the first phase of
construction, three high-end HIT HITCPX1250 projectors are used to
generate event (it is capable to create mirage ). In the
second phase of construction, motorize objects are introduced to
create another sets of mobile targets. The ultimate goal of this testbed
is to allow researchers to conduct all kinds of system research locally
and remotely with realistic sensing modality as inputs. The first
phase of construction is finished during 2007. In the second phase,
mobility support will be added.
Multi-Sequence Positioning (MSP) is designed and implemented for sensor
node localization in outdoor environments. The novel idea behind MSP is
to reconstruct and estimate two-dimensional (or 3D) location information
for each sensor node by processing multiple easy-to-get one-dimensional
node sequences obtained through a loosely guided event distribution. We
have realized the MSP idea through two physical systems (indoor and
outdoor version) with totally over 60 MICAZ motes. This evaluation
demonstrates that MSP can achieve sub-feet-level accuracy, requiring
neither additional hardware on sensor nodes nor precise event
distribution. It also provides a nice tradeoff between physical costs
(anchors) with soft cost (events) while maintaining localization
accuracy.[
Despite the well-known fact that
in reality sensing patterns are highly irregular, researchers continue
to develop protocols with simplifying assumptions of circular 0/1
sensing models. In this project, we design and implement two Sensing
Area Modeling (SAM) techniques useful in the real world. P-SAM provides
accurate sensing area models for individual nodes using controlled or
monitored events, while V-SAM provides continuous sensing similarity
models using natural events in an environment. Evaluation under
real-world settings reveal several serious issues concerning circular
models, and demonstrate significant improvements in several applications
when SAM is used.
defines a Unified Sensing Coverage Architecture, which features three novel ideas: Asymmetric Architecture, Generic
Switching and Global Scheduling. uSense provides sensing coverage
through a creative separation of scheduling from switching. We design
and implement sophisticated scheduling algorithms externally and
represent such intelligence with a lightweight generic switching
algorithm running at resource-constrained sensor nodes. As an instance
of these scheduling algorithms, we propose a novel two-level scheduling
algorithm, called uScan. We evaluate our architecture with a network of
30 MicaZ motes, an extensive simulation with 10,000 nodes. The results
indicate that uSense is a promising architecture to support flexible and
efficient coverage in sensor networks. 
In this
project, we design and test a methodology for navigation of mobile
wireless sensor networks and fast target acquisition without a map,
called GraDrive. Our approach integrates per-node prediction with global
collaborative prediction to estimate the position of a stationary target
and to direct mobile nodes towards the target along the shortest path.
We demonstrate that a high accuracy in localization can be achieved much
faster than existing navigation models without any assistance from
stationary sensor networks. 
uses the spatio-temporal properties of well
controlled events in the network (e.g., light), to obtain the locations
of sensor nodes. We demonstrate that a high accuracy in localization can
be achieved without the aid of expensive hardware on the sensor nodes,
as required by other localization systems. Through performance
evaluations of a real system deployed outdoors, we obtain a 20cm
localization error. A sensor network, with any number of nodes, deployed
in a 2500m2 area, can be localized in under 10 minutes, using a device
that costs less than $1000. To the best of our knowledge, this is the
first report of a sub-meter localization error, obtained in an outdoor
environment, without equipping the wireless sensor nodes with
specialized ranging hardware.
is one of the major efforts in
the sensor network community to build an integrated sensor network
system for surveillance missions. The focus of this effort is to acquire
and verify information about enemy capabilities and positions of hostile
targets. Such missions often involve a high element of risk for human
personnel and require a high degree of stealthiness. Hence, the ability
to deploy unmanned surveillance missions, by using wireless sensor
networks, is of great practical importance for the military. In this
work, we design and implement a complete running system, called
VigilNet, for energy-efficient surveillance. It currently consists about 40,000
lines of NesC and Java code, running on XSM, Mica2 and Mica2dot
platforms. The complete system is designed to scale to at least 1000 XSM
motes and cover minimal 100x1000 square meters to ensure operational
applicability. We evaluate middleware and system performance extensively
on a network of 203 MICA2 motes.