A Wireless Sensor Network testbed for Information Fusion

foamyflumpMobile - Wireless

Nov 21, 2013 (3 years and 6 months ago)

58 views

Page
1

(
4
)

A Wireless Sensor Network testbed for Information Fusion












(updated
2013
-
11
-
21
)



The Distributed Real
-
Time Systems (DRTS) research group



Information Fusion

Infrastructure project (cgi3) of the

Information Fusion research
program


1.

P
roject O
verview

Wireless Sensor Networks
(WSNs)
play an increasingly
more important

role in surveillance and
situation awareness systems.
WSNs
monitor parts of the environment, and
need to
provide timely
alarms for
any
anomalies detected.
The

Common Goals and Infrastructure
project with
in

the
Information Fusion (IF) research program at University of Skövde, Sweden, aims at finding generic
properties of infrastructures
to be used for
IF. Many IF applications rely on
getting
timely updates
about

the environment,
with
information used for time critical decisions.
Typically infrastructures need
to provide real
-
time guarantees, while distributing only the useful information, in a very large
distributed
system of more of less
autonomous
nodes.


To en
able experiments with new IF algorithms and approaches
, we are building up a WSN test
bed for
our IF lab. In collaboration with our industrial partners we will evaluate
new
IF based approaches for
their business areas. This includes military vehicle recogni
tion sensors (Exensor), surveillance situation
awareness (S
AAB
), and soil monitoring for
real
-
time
precision agriculture (SLU/Agroväst).


For the summer of 2008
,

some parts of this larger project is
realized
as an integration project, to
interconnect
unit
s in
an initial version of a potential infrastructure for
information fusion, based on a
distributed database approach.

1.1.

Project Purpose

New t
est
bed and lab resources allows integration of
commercially available
WSN and IF equipment
,

with
specific sensors
,
equipment
and algorithms
used by our partners. Our collaboration will be
strengthened
by
that
our
common
and
new experiments and evaluations can be conducted together
in
one location
.



The major goal is to
be able to
present how typical components and a
pproaches
need to
be used and
combined, to meet the
complex requirements from typical IF applications.

1.2.

Motivation

We have proposed an approach for a WSN/IF infrastructure based on commercially available hardware,
including sensor nodes such as Crossbow’
s Telos and MICAz motes, and more powerful mobile nodes,
such as the Nokia N800/N810 tablet PC. Both type of units are small and mobile, and
have
a
n

operating time of about a week. The
se

development platform
s

are
publicly available and can be
easily
extend
ed

for specific pu
r
poses
.
We use a multi
-
tiered approach for scalability:
The limited processing
and storage resources of the sensor nodes are complemented by the more powerful tablet PCs. To cover
a surveilled area, we
have
propose
d

a two
-
tier approach th
at is more scalable

than a network only
consisting of sensors

[Mat07
, Mat08
]
. In this architecture
,

sensors perform simple monitoring of the
environment, and more processing
-
intensive computation
is
done at the upper tier. We will further
extend this appro
ach to allow more tiers, to process or fuse higher level information at further upper
tiers.


W
e will
specifically
evaluate a database approach to large scale communication in IF infrastructures.
Using a database for communication allows any node to safel
y publish and read information by
using
transactions, and the data becomes available
in the database
for all users of
the
a communication
infrastructure. A distributed real
-
time database with eventual consistency and virtual full replication
allows large s
cale systems to be built. In such systems users access their data locally as if the database
was entirely locally available, while using a fraction of the resources in communication storage and
Page
2

(
4
)

processing required from a centralized communication approach.

There is no need for unpredictable
remote operations during database transactions

since the database approach can replicate database
objects to all nodes where they are available, resulting in that all transactions use only local data
replicas
.


Distribut
ed
Information

Fusion
allows

fusion at lower levels, reducing the amount of data to propagate
in the network, and also dissemination to users without passing a central node. Local decisions may be
done based on lower levels of fused data, possibly more rel
evant to the local situation to control.

1.3.

Problem

It is a complex problem to p
ropag
ate
updates
timely
through a resource
-
constrained network, and
to
provide

the right abstraction level for information
at points of decision
. Further
,

implicit addressing o
f
dynamic
communication
links
is
a
complex

issue for the users of a distributed solution
.


It is difficult to understand generic requirements

on an information fusion infrastructure
, which
information fusion methods are generally useful and how

to apply th
em
, without experiments and
evaluations where requirements from different application scenarios are used. Deepening the
collaboration with different such scenarios allows a deeper understanding from joint development of
ideas and approaches.

1.4.

Partners and

collaborations

The project is currently supported by
S
aab

Microwave Systems (
Contact
Per Gustavsson)
;
Exensor
(
Contact
Thomas Jörgensen
)
;
SLU/Agroväst projekt Precisionsodling Sverige (POS
, ”Precision
Agriculture Sweden”
) (
Contact
Mats Söderström, Bo Ste
nberg)
.


These companies already support the
Information Fusion Program as external sponsors, but also want
to deepen the collaboration regarding

infrastructure issues, and also for future additional applications
for research projects.


We also collaborate

with University of Latvia with a group that specialize in sensor integration and
software/hardware development for sensor platforms. Further, we have collaboration with the WSN
group at University of Virginia. This group is one of the top three groups in
the US for state
-
of
-
the art
research in WSNs.

1.5.

Current funding

The project is a part of the
Information Fusion research profile
,
a research program
at the University of
Skövde
within the area of information fusion from databases, sensors and simulations

(
http://www.infofusion.se
), sponsored by a number of partners (
http://www.his.se/infofusion/partners
).


The
Wireless Sensor Network testbed

project has so far received direct funding for purchase of
testbed
equipment, such as
50
sensors and
4
tablet PCs.
A
dditional funding is approved

for
project man
-
hours,
for
deepened collaboration

with the partners in
the three areas of
military vehicle recognition,
surveillance situation awareness, and soil monitorin
g.

2.

WSN summer project

2008

1.6.

Introduction

To
deepen co
llaboration between research groups at University of Skövde and the participating
companies of the IF program, a
WSN
summer project is initiated, where the intended

outcome is a joint
application for funding of one new PhD posit
i
on and one Post
-
doc positio
n

for the Information Fusion
Infrastructure Project (cgi3).


Page
3

(
4
)

By demonstrating the capabilities of a distributed information fusion setup,
the feasibility of such an
approach is
evaluated
.

A demonstrator
may well motivate further collaboration in utilizing
such
distributed fusion.

1.7.

Aims and objectives

The aim of the WSN summer project is to develop the parts missing for a WSN testbed based on
Telos/MICAz motes and Nokia N810

units used as
database nodes. The
overall
WSN
network
is
organized as sub networks
where the nodes in each sub network
propagate sensor updates to
one
database node
,

by communicating to
one sensor gateway connected to
each
database node
. The sensor
gateway act
s

as a sensor network edge node
,

for updates to be stored at the
database node
.

D
ifferent
sensor sub networks use database nodes

(each one with its sensor gateways
connected to it
)
.


The sensor data is made available in the OS of the N810
in such a way that updates are

store
d

in
distributed database at the N810, where each N810 makes

up one node in the distributed database.

Any
distributed open source database available for extension can be used

at the database node
,
and it is
important that detached replication can be build on top of it, for propagation of updates between
database no
des.
As a part of the project
,

a

suitable database is
chosen and extended

with detached
replication
.


Objective 1:

Connecting the gateway sensor node to the N810 (both in hardware and software)


Objective
s
2:

Objective 2a:

Implement a distributed database

on N810, and store incoming sensor updates in the
distributed database.

Objective 2b:

Visualize sensor data at the N810 unit, first as raw data collected, later cohesive
presentation of
related data

sensed.

Objective 2
c
:

M
ake
all
sensor data objects avai
lable
to all nodes,
to be allocated and used at
any
other
database node,
rather
tha
n

at
the database

node
where the
sensor
is
connected

via its gateway node
.

Objective 2d:

Visualize the content of sensor data at other database nodes, as a cohesive view of
the
“mission” where sensors are used.

Simple fusion of related data is done at database nodes.


Objective 3:

Connect sensors from collaborating

comp
anies to Telos or MICAz nodes, so that real
values are used from these particular sensors in the database s
ystem

1.8.

The platform

The
hardware of the platform is chosen to be an open platform, where all source code is
available and
extensible, and where efficient development tools can be used.


Sensor nodes:

Crossbow is a major supplier of sensor hardware and
much
research
work in the area is based on
hardware from Crossbow. Our test bed uses TelosB nodes that can be connected to any USB port
and
appear
there
as a virtual port that can be access from the OS directly
, for sensor values or for
programming
. The MICAz n
odes are more versatile when integrating with other sensor hardware, but
requires separate programming equipment.


Database nodes:

Nokia tablets currently contains the N700, N800 and N810 model, and uses a the Linu
x

distribution
from the
Maemo project. The

units have WiFi, USB,
web camera,
battery and
a relatively large touch
screen. N810 also has GPS and a hardware keyboard.


The Maemo project has now released the “OS2008” distribution
, as is also the name of the OS when
packet as a product
.

The Maemo proj
ect is a open source project, where
the
source code is available

for
extension and interfacing
.



Page
4

(
4
)

3.

References


[Mat07]

G. Mathiason, S.F. Andler, S.H. Son and L. Selavo (2007) Virtual Full Replication for Wireless Sensor
Networks, Proceedings of the 19th
ECRTS (WiP), Pisa, Italy, July 4
-
6 (
http://feanor.sssup.it/ecrts07/
)


[Mat0
8
]

G. Mathiason, S.F. Andler, and S.H. Son (2008). "Virtual Full Replication for Scalable and Adaptive Real
-
Time Communication in
Wireless Sensor Networks", Proceedings of the Second International Conference on
Sensor Technologies and Applications (SENSORCOMM 2008), Aug 25
-
31, Cap Esterel, France
(http://www.iaria.org/conferences2008/SENSORCOMM08.html)