Energy Conservation in a Smart Home via an Embedded Platform

International Journal of Computer Applications Technology and Research
Volume 5– Issue 5, 241 - 244, 2016, ISSN:- 2319–8656
241
Energy Conservation in a Smart Home via an Embedded
Platform
Rahul Sivaram
MIT College of Engineering
Ex-Serviceman Colony
Pune, India
Vikash Kumar
Pune, India
Shashank Kumar
Pune, India
Sahil Budhwani
Pune, India
Abstract: Home automation is a popular field of application which combines the concepts of embedded systems and pervasive
computing to control a living space for modern times. A smart home system aims to free up the user to carry out his or her daily tasks
without worrying on the trivial aspects of personal home management, while at the same time providing the user with more control of
the home than ever before. The implementation comprises of a large number of sensors which can be used to control or monitor
objects distributed in three-dimensional space around the house. The sensors can be specialized in measuring temperature, humidity,
pressure, light, noise, dust air, and upon intelligently computed triggers, compel the system to perform a specific task, or a set of tasks.
In this project, a solution to transform a normal house into a smart, autonomous house in order to reduce the energy consumption of
the household is proposed. This can be realized with the help of wired sensor networks where we control and interface with electronic
appliances via a Raspberry Pi, an embedded development platform. The embedded platform can be controlled remotely to turn
switches on and off, and eventually aims to become as autonomous as possible via the help of a Bayesian Network. The base workings
of the project are aimed at being as scalable as possible, to be able to potentially fit into industrial as well as office environments in the
future.
Keywords: Energy Conservation, Embedded Platform, Bayesian Belief Network, Pervasive Computing.
1. INTRODUCTION
In the 21st century, the advent of the internet and advances in
the power and miniaturization of computing devices has made
it possible to implement computers wherever we can imagine.
The range of possibilities has given birth to the idea of ‘Home
Automation’, in which computers and algorithms can be given
the responsibilities of operating a household, with minimal
Human interaction needed.
With the rise in popularity of cheap, embedded computers,
such as the Raspberry Pi [8] and the Beaglebone Black
computing has never been more accessible to the masses. It is
proposed in this paper to take advantage of the small form
factor, energy efficiency, computing power and GPIO
capabilities, to interface the embedded platform with
appliances and electrical sockets found in a household. These
platforms can then be programmed in a way to tailor to the
specific functions of the appliance it is connected to,
essentially adding a level of intelligence to the otherwise
“dumb” appliance. The addition of wireless modules to the
embedded board adds the provision to wirelessly network the
board, giving rise to the possibilities of an interconnected,
intelligent home. The optimization of this wireless network
could bring about a level of efficiency to a house/office
space/industrial complex (depending on the appropriate scale
factor) which could bring about tremendous gains in
automation of menial tasks, interconnectivity of the
application space, and energy usage savings.
Electricity is a precious resource which should be utilized
carefully, especially in the modern era where energy
requirements are rising at an alarming rate. It has become a
common agony for citizens of urban spaces to face the ill
effects of a poorly structured, operated and maintained energy
grid. These include problems such as load shedding, and even
complete blackouts. Some of the areas receive only an hour of
electricity a day. As we are restricted on the front of
producing electricity, this calls for a smart way to use
electricity in our living spaces to create an energy efficient
future. This is where our project comes in. We propose
making our homes operationally seamless so that not only do
we not have to worry about explicitly operating our home
appliances, we can be assured the home space is functioning
as efficiently and safely as possible [7]. A scenario in which
every appliance turns on automatically whenever a person
enters the space relevant to the context of the appliance,
depending upon interlinked sensors and historical data which
will act as the training set for our system. Since we will be
working on limited resources we shall consider only a few
appliances however the scope can be widened to
accommodate various devices and so that we can obtain

242
optimized metrics, and perform critical analytics and improve
the learning algorithm used in the project to ensure optimal
functionality [5].
The paper proposes the use of a Bayesian Network as the
principle machine learning algorithm to be used. The selection
of this algorithm is due to the fact it is an extremely powerful
tool to model uncertainty in a situation. The algorithm does
this through the data present in a Joint Probability
Distribution, in which the probabilities of events are co-
related, and are used to teach a machine to come to a decision
about an event when there are a set of actions the machine is
allowed to take in the case of the event. This principle is used
to model the uncertain way a home environment operates.
even the average two bedroom home, in the perspective of the
appliances the home may contain, the living and usage
patterns of the occupants of the home, the considerations of
users of the home space who may not be permanent residents,
but rather frequent visitors such as family and friends, and
non-frequent users such as delivery men, semi-pertinent users
such as pets, makes modeling a deterministic “trigger”
mechanism for the appliances of the home quite difficult,
challenging the placement of this particular computational
problem in the appropriate complexity class. Using the
optimally designed Bayesian Network and historical data
collected from different input points into the system, it is
proposed to model this “trigger” mechanism [3].
2. RELATED WORK
In a paper described by the Silviu Folea Automation
Department Technical University of Cluj-Napoca Cluj-
Napoca, Romania[1], the fundamental concepts which
describe a framework that can be applied to develop a smart
home, and can be tweaked/expanded upon as necessary. It
also provides an introduction to the software ‟LabVIEW‟,
which can be used to capture and interpret data from various
presence sensors.
Alsheikh et al[2] in their paper take into account the
framework of Wireless Sensor Networks (WSNs) and
introduces pertinent topics such as issues with wireless
networks, organization of WSNs, and the specifications of
WSNs. The paper also discusses various machine learning
algorithms (Supervised and Unsupervised) which can be
applied to WSNs to optimize their functioning as well as bring
a degree of autonomy to them.
A paper authored by Nancy A. Roberts for the US Air force
Research Laboratory[3], introduces the implementation of the
concept of a Bayesian Network to model scenarios in a smart
living space for security purposes, illustrated in the paper via
a simulated ”Door Break-in”, in which complete knowledge
of the domain is unknown. More specifically, it proposes how
a Bayesian Network could use historical data to attempt to
predict future instances of the “Door Break-in”.
A paper by Dr. Leslie Haddon [5] deals with the less
technical, but nevertheless important topics regarding the
aspects of home automation such as the human attitudes
towards the concept of home automation, as well as how the
automation of another system such as electricity production
and traffic management would be applicable
3. SYSTEM ARCHITECTURE
As can be seen in the FIG 3.1, we propose a twofold approach
to interact with the embedded platform: one is via a
smartphone application which can access a dynamic webpage
encoded in a scripting language such as PHP, which is hosted
by a server being run in the embedded platform, implemented
using the appropriate framework such as Apache or Node.js.
FIG 3.1: SYSTEM ARCHITECTURE
This approach demonstrates the act of explicit Human
Computer Interaction (eHCI). The second approach is to have
the system continuously monitor surrounding activity through
a presence sensor such as a motion sensor. This would be the
act of implicit Human Computer Interaction (iHCI). The
system would be properly integrated through the combination
of hardwiring, as well as wireless interconnectivity. Wireless
connectivity is proposed to be implemented through an
appropriate set of media access control and physical layer
specifics, such as IEEE 802.11[4]. The appliance we consider
here is an energy efficient LED bulb.
The entire system’s support framework can be thought of as
everything but the mobile webpage and the actual appliance
being controlled.
To operate the system, the user would use a combination of
the options for iHCI and eHCI which are provided. However,
the ultimate goal of the proposed project is to make the
system as autonomous as possible, able to predict system
“trigger” events using a decision value stored in a structured
database, using a probability engine. This engine would
calculate probabilities based on the training data set it has
access to, as well as a dynamically updated probability
distribution. The attributes whose probabilities contribute to
the ultimate trigger decision the system makes (whether or not
to trigger), would include the quantification of common, or
even uncommon, daily activities which take place in a
household. However, to limit the scope of the attributes
considered, due to our project only being at an early research

243
stage, we propose to only consider a handful of contributing
attributes for the system. It can be intuitively hypothesized
that increasing the number of attributes considered would
increase the accuracy of prediction of the crucial trigger
decision, but consideration must be taken that some attributes
may conflict with others, repealing the notion that more data
is always better data, in the context of our project.
Our database design will is proposed to contain two tables: the
first one is the ‘input table’, which will consider attributes like
entry number, date, day, time range, time and also label
denoting whether at that instance of time the lights are turned
on or off. The second table will be the table considering the
results of the probability engine, and also a final trigger
decision based on how those probabilities relate to a preset
threshold. The prior probability will simply by calculated by
considering the duration for which appliance was on and
subtracting it from one will give the probability of the second
class then we can compute the likelihood that is the actual
probability whether at that at this instance of time the light
should be on or off.
4. THE BAYESIAN CLASSIFIER
Heterogeneous systems in the smart home consists of the
building automation system , energy management system, fire
alarm system, digital surveillance system and other network
based systems. Due to the involvement of many entities and
the environment being partially observable, the static decision
making algorithms cannot make decisions to an acceptable
degree. This calls for a smart learning algorithm which
accurately speculates the triggering of appliances.
A Bayesian classifier is a supervised machine learning
technique which can be employed for predictive models
where the environment is highly uncertain. In the context of
our proposed framework, this learning technique uses a set of
labeled data where information regarding a particular instance
of time i.e. that is the discretized time range is fed to the
system with a class label ‘yes’ or ‘no’ denoting for which
instances of time the lights should be turned on. It is based on
a probalistic model where we compute the probability to
trigger the appliance based on a belief network as shown in
Fig 4.1. It consists of a structural model with conditional
probabilities, where probability of a particular node depends
upon the probability of all the incoming nodes i.e. all the
parent nodes. It is only an early example of the model we will
use to visualize the environment, created using the Microsoft
Belief Networks Tool (MSBNx).
FIG 4.1: SAMPLE BELIEF NETWORK
Let us consider an example to determine in which class a set
of tuples belongs based on Bayes’ formula given as :
Here A denotes the class which can be {yes, no}
B denotes the tuple space {B1, B2… Bn}
P (A/B) is computed by our belief network for all possible
values class of A, and the maximum calculated value will be
used for prediction via the function:
5. CONCLUSION AND FUTURE WORK
Home automation is a technology which is still in its infancy.
Here we have provided some insight into the energy
conservation and automation potential of an intelligent
interconnected smart home via explicit Human Computer
Interaction (eHCI) i.e. mobile application and web
applications, as well as implicit Human Computer Interaction
(iHCI) implemented through motion sensors. Both of these
methods will present different pattern in home usage patterns,
which can then be modeled to predict and optimize energy
needs of a living space. There are many use cases for which
we require innovative solutions to use home automation
systems, such as whether a user will only conduct his or her
activities in a living space for a short period of time or a long
period of time. Depending on what the system can “observe”
about a user’s habits, the system can be programmed to take
appropriate actions. The problem lies in being able to
efficiently give true contextual awareness to a system.
Consider a case where someone is sleeping in the room with
lights turned off and another person enters the room [6]. An
unintelligent system will act the way it was meant to,
triggering the light, which is obviously a less than desired
action. The true problem is giving the system not just
awareness, but also the trait of precognition, to a certain
degree. Overcoming this problem could lead to massive gains

244
in the automation potential of a target environment, which
could translate to an energy efficient deployment scenario.
The future scope of this project is also quite promising, as the
basis of the project (the triggering mechanism) can easily be
scaled to accommodate the needs of any household appliance,
as well as any machine in a general sense after fine tuning the
belief network which powers the trigger logic. Subsequently,
if the system is implemented in a wide manner, such as with
the backup of a cloud computing environment, it would be
possible to expedite the training period of the system, by
performing cluster analysis of people with similar patterns of
living, and gain a large, more accurate, and quicker idea of
how and when the system should trigger. A cloud supported
framework would also let the user of the system be able to
remotely monitor and operate the system, via the internet.
Live energy statistics could be made available to the user to
give the user information to act upon, and a user could fine
tune the system further as he or she deems fit.
6. REFERENCES
[1] Smart Home Automation System Using Wi-Fi Low
Power Devices Silviu Folea Automation Department
Technical University of Cluj-Napoca Cluj-Napoca,
Romania Email: silviu.folea@aut.utcluj.ro
[2] Machine Learning in Wireless Sensor Networks:
Algorithms, Strategies, and Applications Mohammad
Abu Alsheikh1,2, Shaowei Lin 2 , Dusit Niyato 1 and
Hwee-Pink Tan 2 1School of Computer Engineering,
Nanyang Technological University, Singapore 639798
2Sense and Sense-abilities Programme, Institute for
Infocomm Research, Singapore 138632
[3] AFRL-IF-RS-TR-2003-25 In-House Final Technical
Report February 2003 Using Bayesian Networks and
Decision Theory to Model Physical Security. Nancy A.
Roberts
[4] Home Automation Analysis of Current Sandeep
Panigrahy, Saurabh Wahile Undergraduate Student,
India, Undergraduate Student, India, International
Journal of Advances in Computer Science and
Technology (IJACST)
[5] Home Automation: Research Issues Dr Leslie Haddon
School of Cultural and Community Studies Arts B
University of Sussex Falmer, Brighton BN1 9QN UK
Paper presented at the second EMTEL Workshop: The
European Telecom User. November 10-11th, 1995,
Amsterdam, The Netherlands
[6] http://jenson.org/easyhard/
[7] http://inhabitat.com/infographic-the-benefits-of-home-
automation-systems/
[8] https://www.raspberrypi.org/

Energy Conservation in a Smart Home via an Embedded Platform

More Related Content

What's hot (10)

Viewers also liked (16)

Similar to Energy Conservation in a Smart Home via an Embedded Platform (20)

More from Editor IJCATR (20)

Recently uploaded (20)

Energy Conservation in a Smart Home via an Embedded Platform