Analyzing Air Quality using the Internet of Things

wittyexecutiveRéseaux et Communications

17 févr. 2014 (il y a 3 années et 8 mois)

413 vue(s)

DUSTI N
FRANZ

Analyzing Air Quality using the
Internet of Things

Research Project



Purdue University Calumet




Professor Ricardo Calix

Internet of Things (
IoT
)


What is it?


Uniquely identifiable objects or
things

in an Internet
-
like
structure


What can it do?


Internet of Cows


Reduce Poverty gap in India


Environmental Monitoring

What things?


Arduino

Microcontrollers connected to
environmental sensors


Why Environmental Monitoring?


Important in NWI



Why
Arduino
?


Arduino

is cheap


It’s fun!

What exactly are you monitoring?


MQ7


Carbon Monoxide (CO)


MQ131


Ozone (O3)






Matches type of data collected by the Indiana
Department of Environmental Management (IDEM)

Reference Model

Reference model

-

Measuring

Arduino Uno with Networking

Physical
Arduino

w/ Sensors

Arduino

IDE


Program structure


#Include Libraries


//Define Global Variables


Setup()

{

//initialize connections

}


Loop()

{

//Read and send data

}

Reference model

-

Connecting

HTTP


REST(Representational State Transfer)


Easy to use APIs


HTTP method


PUT


Why not GET?


Status Codes


Data Formats


CSV


JSON

Reference model

-

COSM

Cosm


A twitter for data


Sharable


Searchable



Similar

problems

Reference model

-

Server,
Cosm

and other APIs

Using the
IoT

for the
IoT


What can we do with just two features?


Also have time and location!


More features!


Weather API


Temperature, Humidity, …

Python Script


To gather all of this information
into a single location



Grabs data from
cosm



every 5
mins


Weather API has restrictions


Updated every 30
mins


JSON



Outputs to CSV file



def

getTemp





grab weather data





return weather data





def

writeToCSV





grab
cosm

data





call
getTemp

every 30
mins





write data to
csv





def

doWork





call
writeToCSV

for each feed





##main##





call
doWork

every 5
mins


Reference model

-

Analyzing

Problem statement


We’ve collected data from different locations in
northwest Indiana and also collected polluted data
from car emissions



Use
Weka

to use this data to try to predict polluted
data
vs

non
-
polluted data

Dataset


MQ131


MQ7


Polluted


latitude


Longitude


temperature
in

Fahrenheit


dew point
in
Fahrenheit


Precipitation
in the last hour in
Inches


pressure
in
in Hg


Relative humidity


Date
and Time


WEKA

Naïve Bayes

Ranker

Problems



Sensors often have a warm up period




Sensors are sensitive and can easily produce bad
data



Getting polluted data

Moving Forward


Need more data of different types (Polluted data)


Need more features to better analyze data



Hope to submit to a journal



Related research


Automatic

Semantic

Content Enrichment of Sensor Network
Based Information