Abstract: Internet of Things (IoT) is extension of current
internet to provide communication, connection, and inter-networking between various devices or physical objects also
known as “Things.” In this paper we have reported an effective
use of IoT for Environmental Condition Monitoring and
Controlling in Homes. We also provide fault detection and
correction in any devices connected to this system automatically.
Home Automation is nothing but automation of regular activities
inside the home. Now a day’s due to huge advancement in
wireless sensor network and other computation technologies, it is
possible to provide flexible and low cost home automation system.
However there is no any system available in market which
provide home automation as well as error detection in the devices
efficiently. In this system we use prediction to find out the
required solution if any problem occurs in any device connected
to the system. To achieve that we are applying DataMining
concept. For efficient data mining we use Naive Bayes Classifier
algorithm to find out the best possible solution. This gives a huge
upper hand on other available home automation system, and we
actually manage to provide a real intelligent system.
Keywords’ IoT; Home Automation; Data Mining; Wireless
sensor Network; Na??ve Bayes Algorithm; AVR Microcontroller
I. INTRODUCTION
For many years Home Automation is mainly used as a
feature of science fiction writing, but it’s become practical
since the early of the 20th Century that is because of the
introduction of electricity and rapid improvement in
information technology.
Home automation or smart homes is described as a
technology which is used within the home environment to
provide comfort, security, convenience, and energy efficiency
to its user or occupants. By inclusion of the Internet ofThings
(IoTs), the research and development of home automation are
going to become more and more popular. Different wireless
technologies that supports remote data transfer, control and
sensing such as RFID, Wi-Fi, Bluetooth, and also cellular
networks have been evolved to add intelligence at various
levels in the home [3].
Internet of Things (IoT) is nothing but physical items
talking to each other, machine-to-machine communications and
person-to-computer communications will be extended to
‘things’.[4] Extension of the current Internet which providing
communication, connection and inter-networking in between
the devices and physical objects, or also known as Things, is a
growing trend that is often called as the Internet of Things.The
Internet of Things (IoT), also referred as the Internetof
Objects, that’s going to change everything which also include
ourselves. IoT is the next evolution or generation of the
Internet, it’s like taking a huge leap in its ability to collect,
analyze, and distribute data which ultimately we can turn into
information then knowledge and finally into wisdom [9].
The Internet of Things (IoTs) is nothing but connecting
everyday objects like Internet TVs, smart phones, sensors and
or actuators to the Internet where this devices linked together
intelligently which enables new forms of communication
between people and things, and between things themselves [8].
Now anyone, can have connectivity with anything for
anytime and anyplace and it is possible to create these
connections which extend and possibly create an advanced
dynamic network. Modern advances in electronics and
communications Technologies have lead to the miniaturization
and improvement of the performance of computers, sensors
and networking. These changes have given rise to the
development of several home automation technologies and
systems. According to, home automation can be useful to those
who need to Access home appliances while away from their
home and can incredibly improve the lives of the disabled.
In this paper we are primarily focusing on, the use of IoT
for the advance, flexible, energy efficient and self-learning
home automation system. For that we also using data Mining
Concept. Data Mining is subfield of computer science which is
used to discover pattern in large data sets by using combination
of some methods like statistics, machine learning, artificial
intelligence and database systems. In simple words Data
Mining is data or knowledge discovery. It involves analyzing
data by applying some conditions and converting it into useful
information or knowledge.
II. MOTIVATION
To improve standard of living it is needed to change
home environmental condition according to the mood of the
habitants without any interruption. In some cases physically
disable or handicapped people are not able move much from
one place so for them it is very difficult to access regular
domestic appliances. For them it is essential to develop a
system which requires less human interaction. We need energy
efficient, flexible system which also detect the fault in the
devices automatically and notify the related technician and user
about the problem automatically. To provide all this facilities
in developing countries like India we need a much smart
system which provide all the above facilities in low price and
less energy consumption.
III. LITERATURE SURVEY
Home Automation is nothing but automation of the home
that is automation of household activity or housework. Itcan
also include centralized lighting control, appliances,
ventilation, heating and air conditioning (HVAC), securityof
doors and gates. It helps to improved comfort, convenience,
security and energy efficiency. Home automation is useful for
elderly and disabled peoples to increase quality of life so they
becomes less dependent on caregivers. [5]
In recent years home automation popularity has been
increasing mainly because of its simplicity through smartphone
connectivity and higher affordability. In home automation
system various electrical devices in a house interacts with each
other by use of the information technology to increase energy
efficiency and security. Although there are some problems with
this system like complexity, high competition with other
vendors, incompatible standards and high expense which
results to this home automation system is limited to wealthy or
ambitious users only [6].
Many of the home automation systems that are
commercially available can be separated into two categories:
locally controlled systems and remotely controlled systems.
Locally controlled systems use an in-home controller to
achieve home automation. This allows users complete use of
their automation system from within their home via a
stationary or wireless interface. Remotely controlled systems
use an Internet connection or integration with an existing home
security system to allow the user complete control of their
system from their mobile device, personal computer, or via
telephone from their home security provider. [7]
In one study researchers introduces home automation
systems based on Bluetooth, that using Android Smart phones.
The devices that we used to access and control is physically
connected to a Bluetooth controller. The Smart phone is then
connected to it by using in-build Bluetooth connectivity which
control that devices. In some other case researchers also
provide network interoperability and one important feature that
is remote access to control home devices or appliancesusing
home gateways [2].
There are few examples where researchers have reportedan
effective implementation for Internet of Things that areused
for monitoring regular domestic conditions and also theytrying
to build on low cost ubiquitous sensing system. But it has some
issues like requirement of IPv6 connectivity which a major
concern in implementing this methods. Also storage
requirements and effective retrieval of data and security are
concern. [1] And other are just monitoring and controlling the
home environment more like just the switching functionality.
There are some research on data mining models for IoTs.
Where they propose four data mining models for the Internet of
Things, which are multi-layer data mining model, distributed
data mining model, Grid based data mining model and data
mining model from multi-technology integration perspective
[2]. We also get some information on Na??ve Bayes Algorithm
which is use for data mining [7].
There are a number of issues involved when designing a
home automation system. It should provide a user- friendly
interface on the host side, so that the devices can be easily
setup, monitored, and controlled. Furthermore the overall
system should be swift enough to realize the true power of
wireless technology. Lastly the system should be cost effective
in order to justify its application in home automation. To
minimize the shortcomings of each system and to overcome the
design issues previously mentioned, we integrates locally and
remotely controlled systems with the use of Cloud data
network. This allows the system to operate without the
dependence of a mobile provider, allows the system to be used
with various mobile phone platforms, and allows the system to
operate locally when phone or computer access is not available.
Cloud networking and data infrastructure allow individuals to
monitor, manage, and control their personal data points through
the Internet. One of the available services is Pachube[10].
Fig. 1. Block Diagram of Existing System [4].
IV. PROPOSED SYSTEM
A. Objectives
We sense environmental conditions by various sensors and
according to its values, we control various devices which are
connected to microcontroller through drivers and relay board.
One PC connected to microcontroller will monitor the sensors
values continuously and convert this values in graphical form.
We can manage to change settings by just changing the
threshold values in the PC.
If some device is not functioning after some time period
sensors will detect the faulty device ex. If light bulb is not
working after 5 second light sensor will detect the problem.
Then Home PC will inform it to cloud server then server apply
data mining on its available data sets to find out which
technician or service provider to call and it sends notification to
that technician and user via SMS or E-mail. For data mining
we are using Na??ve Bayes Classifier Algorithm. In thisway we
can predict the service provider which is most suitable and
most popular in our area for our problem.
B. System Design
Fig. 2. Architecture of Proposed System
‘ We use different sensors (light, temp., level,
humidity) to collect the data to understand the
environmental conditions and also to detect any fault
in devices. It is necessary to act devices according to
the inhabitant requirement.
‘ Home PC continuously monitor sensors values and
control the devices accordingly. If problem found it
report to cloud server. Here user can modify some
settings and see the devices functionality and
working.
‘ We build one Registration Application where
Technicians, super market vendors and other service
provider will register on it. They provide their details
like type of service (Sales/Service) and their service
timing etc.
‘ Finally Cloud Server will apply data mining on data
sets. It also mail or SMS Technician and send details
to the Owner (mail or SMS). We can connect any
number of users on cloud server so it support multi
user system characteristics.
‘ Here we can use only one cloud server but we can
connect many numbers of users to it via pc, or any
android devices.
C. Algorithm
Na??ve Bayes Classifier Algorithm:
It is a simple probabilistic classifier algorithm, based on
applying Bayes rule with na??ve or strong independence
assumptions. Na??ve Bayes is used particularly when attributes
are independent of each other.
Na??ve Bayes always provide a more efficient solution. This
procedure is based on Bayes Rule, Probability is denoted by
P (C’W) Where,
P (C’W) =P (W’C) P (C) ‘P (W)
Here P (C): is Prior probability (Independent probability of C).
P (W’C): is likelihood probability (conditional probability of
W given C).
P (W): is evidence (Independent probability of W).
And P (C’W): is posterior probability (Cond. probability of C
given W).
.
Class labels in the proposed system dataset is whether to
Choose Technician or not and it is Yes or No. Attributes are
Types of Services and Service Timing.
D. Mathematical Model
We use Set Theory to explain our system.
S= { U, Sen, Ser, Tech, grab(), login(), reg(), setthval(),..}
Where S is our system which is set of some functions and
other sets.
U={ U1,U2,’.Un}
Where U is Infinite set of Users.
Sen={Sen1,’Sen8}
Where Sen is set of 8 servers as we are using 8 channelbuild
in AVR microcontroller.
Ser={Ser1} As we use only one server
Db ‘reg (U)
Auth ‘login (Uid,pass)
Senval ‘grab (C) Where C- channel No. (ADC-0-7)
Setthval (thval, Seni)
Scan (): Scan function used to scan the current sensors values
with set threshold values. Ex. Scan (Currval, Thval);
For Prediction we applying Data Mining on data Set, and for
that we are using Na??ve Bayes Classifier Algorithm.
After that we get results
A1 ’email alert
A2 ‘SMS alert
A3’Voice alert (Future Scope)
As we are using Na??ve Bayes Classifier Algorithm so we
cannot predict the exact time for problem completion butwe
know we can find the solution in nondeterministic polynomial
time so the above problem is NP-complete (NPC).
E. Implementation Details
Fig. 3. Data Flow Diagram (DFD1)
In the implementation hardware part of the system we are
using AVR ATmega-32 Microcontroller. It is 8 bit
microcontroller and has in-build ADC. We use MAX-232 IC
for the serial communication between microcontroller and PC.
According to sensors values we calculate which devices we
need to turn ON or OFF. For that purpose we use ULN 2803
Driver IC. The detailed Circuit diagram is shown in Figure 4.
Fig. 4. Circuit Diagram
V. RESULTS
In these papers we proposed a Novel technique that will
give us best result. First of all we collect differentsensors
values continuously and then compare this collected values
with the Threshold values which are already set according to
users requirements. For that we divide threshold values into
three categories which are shown in Table 1. We are also
converting this collected sensors values in graphical format for
users understanding as shown in Figure 3. By this values we
control the various home appliances. If error occurs while
controlling this devices, home PC will inform to cloud server.
Cloud server then apply data mining on available data sets and
send SMS or E-mail to technician and user about the problem.
TABLE 1 TRESHOLD VALUES SET
...(download the rest of the essay above)