Database Integration using Heterogeneous Sources in Wireless ...

1

Blake Burns

Texas A&M University
-

Corpus Christi

Anne Edmundson

University at Buffalo

Dr. Longzhuang Li

Faculty Mentor

Texas A&M University
-

Corpus Christi


2

2

Overview


Abstract


Background


Objective


Red Tide


Importance of our Research


Approach


Project Implementation Details


Challenges


Future Works


References


3

3

Abstract


Using marine wireless sensor networks to collect
meaningful data for future analysis in predicting
the presence of red tides.



Selected attributes for data collection:



chemical oxygen demand


temperature


salinity


pH


water transparency


tidal currents



wind


precipitation


sun light intensity


chlorophyll concentration


dissolved oxygen


dissolved nitrogen and
phosphorus


4

Background
(part 1/2)


Wireless Sensor Networks


Consists of spatially distributed autonomous sensors to
cooperatively monitor physical or environmental conditions, such
as temperature, sound, vibration, pressure, motion or pollutants.


TinyOS


Operating system for wireless embedded sensor networks


Minimizes code size because of memory constraints


TinyDB


Query processing system used on network of TinyOS sensors


Given a specific query, TinyDB collects data from sensor nodes


TOSSIM


Simulates a complete TinyOS sensor network

5

Background
(part 2/2)



Wireless Sensor Network purposes:


Equipped with capabilities to measure/change environment


Sense, process, and communicate data


Wireless Sensor Network applications:


Environmental


Marine monitoring


Landslide detection


Medical


Monitor vital signs


Military


Smart Uniforms


Event monitoring for enemy detection


5

6

6

Objective



Our goal

to create a
uniform interface

to access to
multiple autonomous heterogeneous structured
data sources that will help to predict red tide

7

7

Objective Details


We are creating an interface that will forward a
query to multiple databases

and provide results in
a uniform manner for the specified information
regarding red tides



There are multiple ways a user may define their
query: by attribute, by date, or by node.


8

Red Tide
(part 1/4)


What is red tide?


Red tide is a naturally
-
occurring,
higher
-
than
-
normal concentration
of the microscopic algae
Karenia
brevis.


This organism produces a toxin that
paralyzes fish causes them to
suffocate. When red tide algae
reproduce in dense concentrations
they are visible as discolored
patches of ocean water, often
reddish in color.

9

Red Tide
(part 2/4)


Consequences


Disturbs marine ecosystem


Affects fishes, oysters, mussels and whelks


Significant because humans consume them


Existing approach


Satellite imagery


Satellites only see ocean surface


Weather prevents frequent coverage


Clouds and fog obscure visible and infrared data


Expensive for environmental monitoring



10

Red Tide
(part 3/4)


Why wireless sensor networks?


Real
-
time monitoring


Collects surface and sub
-
surface information


Not too expensive


Capable of remote monitoring in any environment


11

Red Tide
(part 4/4)


Predicting red tide


Measure temperature, dissolved oxygen content, salinity


Algae absorb oxygen so low levels of oxygen show
possible red tides


Variations in temperature are observed


Measure chlorophyll which is the indicator of red tide
algae


12

Importance of our Research


Our research will reduce the difficulty of
processing the coastal data through our uniform
interface that can access all the data related to the
coastal systems.



This will also help in detecting the presence of red
tide and predicting future red tides.

13

13

Approach
(part 1/5)


Due to limited resources, these attributes were
simulated using TOSSIM on TinyOS.



TinyDB was utilized to filter, and aggregate data from
wireless sensor nodes.



Restrictions with TOSSIM only allow one attribute to
be simulated in each network; therefore, 12 networks
were simulated and three TinyDBs were used, each
holding data from four networks.

14

14

Approach
(part 2/5)


This was only possible because each network had
data in common with the other networks: a node
ID.



Using the given framework for TinyDB, an
application was created that allowed the user these
capabilities.

15

Approach
(part 3/5)


This interface acts as the communication between
the user and the wireless sensor network.



This implementation provides a user with a
flexible means to gather information from
multiple marine wireless sensor networks.

16

End Users





Applications

Global Uniform


Interface

TinyDB

#3

TinyDB

#2

TinyDB

#1

Approach
(part 4/5)

WSN or
TOSSIM

TinyDB GUI

TinyDB Client API

DBMS

Sensor network

Approach
(part 5/5)

TinyDB query

processor

0

4

0

1

5

2

6

3

7

JDBC

Mote side

PC side

8

18

18

Project Implementation Details



Three separate main components



Attribute specific query


Map feature to query by specific node


[NYI] date range querying


Java based Graphical User Interfaces



19

19

The Interface
(part 1/3)


Main GUI Interface


Select which type of querying to do




20

20

The Interface
(part 2/3)


Attribute specific query


Select any number of attributes and get all
available data for those attributes



The following slides are some screenshots of the
attribute specific query in action using simulated
data (not accurate).



21

21

22

22

If no
attributes
were selected
the window
below appears

If there were attributes selected
then the below window appears
displaying the data from the
database for the selected
attributes

23

23

24

24

The Interface
(part 3/3)


Node specific query
(Map Interface)


Allows the user to select a node from a map to
query and retrieves the selected nodes data.



The following slides are of the node specific query
in action on simulated data (not accurate).



25

25

26

27

28

28

Challenges


Took far too long to get to implementation


First 4
-
5 weeks only reading and software tweaking of
TOSSIM and TinyDB.



TOSSIM would not generate custom data


Even still it will only generate one custom attribute
per run through.



TinyDB would not store data


We had to modify the main program to store the data
in a file.





29

29

Future Work
(part 1/2)


Incorporate time as a factor in submitting a query


Morning, afternoon, evening options



Increase flexibility of interface


Gather data from TinyDBs as well as other databases


Select multiple nodes from the map



Present the data in a more organized and logical
manner


30

30

Future Work
(part 2/2)


Make the application available via the internet


Allows for easier access



Deploy application onto a smart phone


Information is always available to the user and more
accessible



Eventually deploy sensor nodes to collect data and
use the application for these nodes


31

31

References


S. Chawathe et al.
The TSIMMIS Approach to Mediation: Data Models and
Languages.

In
Proc. 10
th

Meeting of the Information Processing Society of Japan,
1994.


Ibrahim, R. Kronsteiner, and G. Kotsis.
A Semantic Solution for Data Integration
in Mixed Sensor Networks.

Computer Communications
, 28(2005) 1564
-
1574.



A. Zafeiropoulos, N. Konstantinou, S. Arkoulis, D. Spanos, and N. Mitrou.
A
Semantic
-
based Architecture for Sensor Data Fusion.

In
the Second International
Conference on Mobile Ubiquitous Computing, Systems, Services, and
Technologies
, 2008.



S. Mihaylov, M. Jacob, Z. Ives, and S. Guha.
A Substrate for In
-
network Sensor
Data Integration.

In
the 5
th

Workshop on Data Management for Sensor Networks
,
2008.



I. Botan, Y. Cho, etc
. Design and Implementation of the Maxstream Federated
Stream Processing Architecture.

ETH Zurich, Technical Report, June 2009.



N. Tatbul.
Streaming Data Integration: Challenge and Opportunities.

In
the
Second International Workshop on New Trends in Information Integration

(NTII),
March 2010.


32

Acknowledgments


Dr. Dulal Kar


Dr. Longzhuang Li


Dr. Ahmed Mahdy


Huy Tran


Bhanu Kamapantula


Tinara Hendrix and Ashley Munoz


National Science Foundation


33

33

Questions?