Português
English |
 |
 |
 |
 |
 |
 |
 |
Embedded
systems and wireless sensor network infrastructure design
WSN Management
MannaSim - Simulation
Optimization
Energy Map
Application Development
Protocols
Robotics
Security
EMBEDDED
SYSTEMS AND WIRELESS SENSOR NETWORK INFRASTRUCTURE DESIGN
(Claudionor José Nunes Coelho, Diógenes Cecílio da
Silva Júnior)
In this research project we are developing
fault tolerant hardware for wireless sensor nodes based on reconfigurable
architectures for special applications in wireless sensor networks. We
are also designing its operating system and middleware for application
development. The sensor node hardware is composed
of a microcontroller and a logic programmable device (FPGA – Field-Programmable
Gate Array). The sensor node will achieve fault tolerance by the insertion
of hardware and software assertions [Nacif et al., 2003]. Assertions generate
hardware and software interrupts that are managed by the hardware and
operating system [Sica et al., 2004].
The system design is divided into
three stages, some of them running concurrently. The first stage has the
goal of understanding the current technology. In order to facilitate this,
we acquired TinyOS/Mica Mote wireless sensors, and development kits designed
by University of California at Berkeley and sold by Crossbow Inc. The
Mica2 Mote is a low energy consumption sensor node that uses TinyOS, a
simple and compact operating system, eventbased, and designed for WSNs.
This kind of network is characterized by intensive concurrent operations
with minimal hardware requirements and minimum energy use.
In the second stage, we designed the
BEAN (Brazilian Energy-Efficient Architectural
Node) sensor node and our research contributed towards a master’s
thesis [Vieira, 2004b]. A BEAN prototype is currently being constructed.
The BEAN uses a Texas Instruments
MSP430 microcontroller with complete analogue to digital conversion capabilities,
making
sensor integration easier. We used CC1000 as the radio transceiver.
In the third stage, we are designing a new sensor node called Killer BEAN,
which uses, in
addition to the microcontroller, a high-density logic programmable device
– FPGA, a compact flash card unit (memory or peripherals), and a
large Flash/RAM memory capacity. The objective of this node is to receive
other special purpose processing elements, for example a send/receive
cryptographic module. This device integrates a set of functionalities
and computational resources that are not present in current commercial
sensor nodes.
To help wireless sensor node application
development/management an OS called YATOS (Yet Another Tiny OS) and a
middleware called WISDOM (WSN Development cOde
Middleware) were developed [Vieira, 2004a]. The OS provides application
services and
hardware interaction, controlling, monitoring and managing processor/peripheral
status.
WISDOM inserts an abstraction layer between the application and HW/SW
computational
platform. Currently, middleware related activities include the design
of a visual tool and a
new programming model, which are able to start from a module-based specification
and then proceed to generate application source code for a specific computational
platform. This approach helps sensor node programming and generates efficient
source code.
It is worth to mention that the technology
to design and build wireless sensor nodes
hardware is available for sale and tends to become more accessible with
the large-scale
production of different micro sensors. But wireless sensor design is highly
dependable on the application. The interaction between biology, medicine,
physics, and other research groups will help in the specification of different
practical real applications for WSN nodes.
WSN
MANAGEMENT
(Antônio Alfredo F. Loureiro, José Marcos Silva Nogueira,
Linnyer Beatrys Ruiz)
The task of designing solutions for
WSN specific problems is not trivial. The integration of these solutions
to promote the network productivity and QoS is even a greater challenge.
This activity deals with this challenge: to study the WSN management problem,
understanding the necessities, requirements and questions related to it
in order to identify the differences between the WSN management and traditional
network management. The definition of a management framework for WSNs
is the main contribution of this activity. This is a new research area
that only recently started to receive attention from the research community.
The WSN management must be simple,
adherent to network idiosyncrasies including its
dynamic behavior, and efficient in the use of scarce resources. In this
activity, we introduced a WSN management framework that includes a management
architecture for WSNs, called MANNA, which is organized in information,
functional, and physical architectures, and the proposition of an organization
for WSNs management from three management dimensions, namely management
levels, management functional areas and a novel management dimension called
“WSN functionalities”. The traditional management dimensions
are revised for WSNs.
Some of the contributions and results
of this work have been published as a doctoral thesis [Ruiz, 2003], a
journal [Ruiz et al., 2003c], a international book chapter [Ruiz et al.,
2004b], papers in international conferences [da Silva et al., 2003, Ruiz
et al., 2003a, Ruiz et al., 2003b, Ruiz et al., 2004c, Ruiz et al, 2004d,
Ruiz et al. 2003e, Silva et al., 2004b, Vieira et al., 2003c], papers
in national conferences [Ruiz, 2004, Ruiz et al., 2004a, Ruiz et al.,
2003d, Loureiro et al., 2002, Loureiro et al., 2003a, Loureiro et al.,
2003b, Vieira et al., 2004, Siqueira et al., 2004, Vieira et al., 2003b,
Silva et al., 2004c, Braga et al., 2004]. To the best of our knowledge
and in accordance with Mehmet Ulema [Ulema, 2003], the framework proposed
in the thesis [Ruiz, 2003] is the only integrated management solution
for WSNs that has been proposed in the literature so far.
MANNASIM
- SIMULATOR
(Antônio Alfredo F. Loureiro, José Marcos Silva Nogueira,
Linnyer Beatrys Ruiz)
The goal of this activity is to develop
a simulation environment that extends the Network Simulator Tool (NS-2)
[Network Simulator, 1999] capabilities and to incorporate the hardware
functionalities foreseen for WSNs. The proposed simulation environment
has been designed to be easy to use, extensible and accessible to public
in general.
Simulation is an important tool in
the search for a better understanding and knowledge of a given system.
Simulation using computers is an important tool for the analysis of complex,
expensive, and difficult to build systems. It is a technique that helps
the development and analysis of new ideas and algorithms to be used in
real world problems. To build a system without a simulation analysis is
not a good idea, mainly when there is none or little knowledge about the
technology that is being used, which is the case for WSNs.
The development of a simulation framework
for WSNs can assist researchers in
experiments and in the development of concept proofs before investing
in the development of real scenarios. In the beginning of our research
project and development activities, there was no simulation tool specific
for WSNs. Therefore, to perform the experiments, we have
designed and implemented a simulation module that extends the Network
Simulator (NS-2)
functionalities. This module, called MANNASim, was firstly developed as
part of a doctoral
thesis [Ruiz, 2003] with the help of two research assistants. This is
a Free Software project and it is being used to explore new ideas from
different groups in the SensorNet project.
OPTIMIZATION
(Geraldo
Robson Mateus)
In WSNs, there are several optimization
problems including density, routing, connectivity, coverage, and transmission.
These problems can be solved considering static and dynamic scenarios.
The density, coverage, and connectivity problems try to find the subset
of sensor nodes that must be active to assure the sensor field coverage
and the network connectivity, minimizing the energy consumption. The routing
problem deals with data delivery of sensed info collected by actives nodes
to an access point, considering two topologies: flat and hierarchical.
In the former, the routes can be established between active nodes and
a gateway (sink) node. In the latter, each group chooses a leader that
will receive all data from its group members and will disseminate it to
a gateway node. All problems mentioned above can be formulated as mathematical
models that can lead to the development of algorithms, making the area
very rich for applying combinatorial optimization techniques.
Contributions of this activity include
three master’s dissertations [Nakamura, 2003, de
Oliveira, 2004, and Menezes, 2004], and conference papers. In [Nakamura,
2003], it is
presented two models of mixed integer linear programming that deal with
coverage and
connectivity problems in WSNs. The models propose a node-scheduling scheme
during predefined time periods to save energy, but at the same time assuring
the best possible coverage and the network connectivity. In [de Oliveira,
2004], it is proposed a solution to the coverage and routing problems
in hierarchical WSNs using mixed integer linear programming, and an algorithm
based on Langrangian Relaxation to solve larger problems. In [Menezes,
2004], it is presented a mathematical model and algorithms based on Langrangian
Relaxation to density, coverage, and connectivity problems in flat WSNs.
ENERGY MAP
(Antônio Alfredo F. Loureiro, Raquel Mini)
The
efficient use of the limited energy is the key challenge in the design
of WSNs. Due to the paramount importance of the energy conservation, all
protocols for these networks have to be designed taking into account their
energy consumption. In this activity, we study four aspects related to
energy in WSNs: (i) energy dissipation model, (ii) and (iii) prediction-based
approaches to construct the energy map based on a probabilistic and budget
models, and (iv) a new routing algorithm for WSNs based on the energy
map.
The information about the remaining
available energy in each part of the network is called the energy map
and can aid in prolonging the lifetime of the network. We can represent
the energy map of a WSN as a gray level image, in which light shaded areas
represent regions with more remaining energy, and regions short of energy
are represented by dark shaded areas. Using the energy map, it is possible
to determine if any part of the network is about to suffer system failures
in near future due to depleted energy.
We propose mechanisms to predict the
energy consumption of a sensor node to construct the energy map of a WSN
using a probabilistic model. If a sensor can predict efficiently the amount
of energy it will dissipate in the future, it will not be necessary to
transmit frequently its available energy. This node can just send one
message with its available energy and the parameters of the model that
describes its energy dissipation. With
this information, the gateway can update its local information about the
available energy of this node. We analyze the performance of the probabilistic
model, and compare it with a naive approach with no prediction. Simulation
results show that the use of the prediction-based model decreases the
amount of energy necessary to construct the energy map of WSNs.
Due to the paramount importance of
energy conservation, it is highly desirable to define the amount of energy
each protocol can spend to perform its goal. Using this idea, we can associate
a finite energy budget for each network activity, and ask this activity
to achieve its best performance using only its budget. We show how this
principle can be used to construct the best energy map of a WSN. Our goal
is to construct the best energy map using only a defined amount of energy.
Simulation results show that we can approach the performance limits using
the proposed finite energy budget model.
We propose a new routing algorithm
for WSNs that combines concepts presented in
trajectory-based forwarding (i.e., a route is embedded in each packet)
with the info provided by the energy map to determine routes in a dynamic
fashion. Simulation results revealed that the energy spent with the routing
activity can be concentrated on nodes with high-energy reserves, whereas
low-energy nodes can use their energy only to perform sensing activity
or to receive information addressed to them. In this manner, the network
lifetime is extended.
Contributions of this activity include a
doctoral thesis [Mini, 2004] and some publications such as [Mini et al.,
2003, Mini et al, 2004a, Mini et al., 2004b].
In
this activity, we developed an environmental monitoring activity to be
used as a proof of concept of the WSN technology. In June 2004, we performed
an experiment at the UFMG campus that showed the integration of some of
the research results of the activities already developed. We continue
to work on this application and plan to build new ones.
Due to the great variety of WSNs applications,
it is important to have tools that help
application developers with their work. We have designed and implemented
a tool called
BeanWatcher [de Aquino, 2003] that helps the development of WSN monitoring
applications.
It runs in a portable mobile device
and can be used as a remote data monitor or even actuator.
Contributions of this activity include a master’s dissertation [de
Aquino, 2003] and
other publications [de Aquino et al., 2003a, de Aquino et al, 2003b].
PROTOCOLS
(Antônio Alfredo F. Loureiro, Linnyer Beatrys Ruiz)
A
fundamental problem in any computer network is the routing of data packets.
Routing
protocols for regular wireless networks are inefficient for WSNs due to
their particular
characteristics. For instance, in WSNs, transmission range is often limited,
which restricts a
direct communication between a node and the gateway, and the network is
typically data
centric and does not use an addressing mechanism. Any algorithm for this
kind of network
must be energy efficient. As sensor networks are built for a specific
purpose, it is possible to design protocols optimized for a given application,
thus reducing the energy consumption of the network. This activity has
proposed and evaluated different routing protocols.
ICA [Maia et al., 2004] is a hierarchical
multi-hop algorithm that presents a good trade-off between delivery rate
and energy consumption. CHDOS [Maia et al., 2003] explores the idea of
a chain of nodes to increase the network lifetime. PROC [Macedo et al.,
2004b] is a routing protocol for homogeneous WSNs that periodically send
data to the network, such as networks for environmental and industrial
control. STORM/AD [Nakamura et al., 2004] is a superimposition solution
that defines a substrate plane with a self-organizing algorithm that builds
a directed acyclic graph, and a superimposition plane with an adaptive
diffusion dissemination algorithm that evaluates the available paths choosing
the best one according to the application characteristics provided by
the network designer. Multi [Figueiredo et al., 2004] allows different
dissemination strategies that can be employed in different scenarios.
The performance of protocols for WSNs
such as routing, data aggregation, and localization depends on the MAC
protocol employed by the sensor nodes. MAC protocols are responsible for
physical communication among nodes, and usually employ techniques to reduce
energy consumption. Most WSNs will operate unattended, and energy consumption
must be minimal. Also, protocols must account for frequent node failures.
MAC protocols for sensor networks differ from protocols used in ad hoc
networks as the amount of energy and the degree of change is much more
pronounced in WSNs. Furthermore, MAC protocols for sensor networks usually
do not account for mobility, as nodes are usually stationary.
In this activity, we are designing
and implementing a MAC protocol using power control
algorithms to increase the network lifetime. The idea is to dynamically
decrease transmission power to the minimum, increasing network capacity
and decreasing energy usage. A node will implement techniques for turning
frequently its radio on/off to decrease energy consumption.
ROBOTICS
(Guilherme Augusto Silva Pereira, Mário Fernando Montenegro Campos)
The
main goal of this activity is to develop algorithms and methodologies
to bring greater flexibility to WSNs by including the mobility factor.
The use of mobile robots in WSNs creates a wealth of interesting possibilities
and fascinating problems. The great synergy between the two areas, which
naturally complement one another, is the key that ushered in a broad spectrum
of applications.
Mobile robots are, in their own right,
autonomous networked systems that can localize
themselves in the environment and move following (pre)planned trajectories.
They are
capable of actively perceiving their surroundings using a collection of
different types of
sensor modalities. Based on the perceived information, mobile robots are
able to quickly react to unexpected and dynamic events. When interacting
within a sensor network, a mobile robot can be seen as powerful mobile
sensor node. They embed all typical node capabilities such as sensing
and network routing, but reach far beyond a node’s limitations.
This is due to the robots inherent ability to move about and execute a
myriad of different tasks.
Sensor deployment is among the many
tasks which robotics has a great impact on the
sensor network. Other relevant tasks include node power charging, node
replacement,
network connectivity maintenance and active data collection. Also, a mobile
robot may
concurrently act based on data gathered by the network of sensors, such
as search and rescue, monitoring, surveillance and people guidance. From
the robotics point of view, sensor networks may be considered as a natural
extension of the robot capabilities. Data collected from the network may
be readily used to augment the robot’s perception of the world and
to improve localization, map building and navigation.
Contributions of this activity include
two doctoral theses [Pereira, 2003, Tavares, 2004] and other publications.
In [Pereira, 2003], besides the main contributions on cooperative robotics,
a distributed algorithm for localization of a robot/sensor network is
presented. In [Tavares, 2004], it is studied the problem of sequential,
decentralized decision making of a team of autonomous agents, under stochastic
action, partial observability, and imperfect and limited communication.
Besides that, different sensor deployment algorithms are being studied
and implemented in a master’s dissertation [Pereira et al., 2004].
Security
(Antônio Alfredo F. Loureiro, Wong Hao Chi)
Security
is a critical issue in WSNs. Besides the challenges present in other networks,
those trying to secure WSNs face another serious obstacle: resource scarcity.
This additional challenge, together with the fact that it is difficult
to secure such networks, have attracted a significant number of research
groups, both in academia and industry, to make security in WSNs one of
their top research agendas. The security group within the SensorNet project
has kept up with the international community in this effort, and it is
the only security research group in the country that have focused on this
new type of network.
We have focused on prevention, detection,
as well as tolerance measures. Currently, a
number of theses, dissertations, and independent studies are being carried
within the security sub-group. More specifically, we have proposed key
distribution schemes for both flat and hierarchical networks, and have
developed associated security protocols. We have started formulating intrusion
detection models (both centralized and decentralized) for attacks against
WSNS in general, and specific detection schemes for detecting attacks
based on malicious manipulation of radio signal strength levels in communication.
Finally, we are developing redundancy/alternation mechanisms to make the
networks tolerant to attacks.
Contributions of this activity include
one doctoral thesis and three master’s dissertations, which are
work in progress, and some publications [Ferreira et al., 2004, Pires
Jr, 2004a, Pires Jr et al., 2004b, Oliveira et al., 2004, da Silva et
al., 2004]. Currently, a security protocol has been incorporated to our
monitoring application demo.