3rd International Conference on Internet of Things and Connected Technologies (ICIoTCT), 2018

Design and implementation of a smart energy meter based on internet of

things and neural network approach
Srishti Priyaa, Archit Srivastavab, Sumit Kumar Jindala,*, Sarat Kumar Sahoob
a
School of Electronics Engineering, VIT University, Vellore-632014, India
b
School of Electrical Engineering, VIT University, Vellore-632014, India

Abstract: This work is about optimizing the existing meters that provide meter reading at the end of the month to the consumers with the help of a person
who notes them down manually. This paper mainly highlights the existing methodology of electricity billing that takes place in the households and
proposes a method with the aim of requiring minimum human interference and very high efficiency by incorporating data transfer via the internet. Thus
without man-power power companies or organizations are able to provide services to its customers by feeding data in an integrated energy management
system. The same data will be made accessible to the consumer and billing authorities over cloud. This way the customer is fully aware of his daily
consumption. Energy theft is prevalent in the societies and by using this concept we can efficiently mitigate the same. The option of online payment is also
included thus making it very convenient for the consumer to pay their bills. Further, using neural networks we have also analysed the trend of energy
consumption or requirements of energy in the household. High fidelity billing and its related remote metering services are successfully implemented
through this project.

Keywords—Beaglebone black, Neural network toolbox, MATLAB, Energy meter, Firebase real-time database

for low power consumption in the smart metering system (S.Arun &
1. Introduction Sidappa Naidu, 2012).

Internet of Things (IoT) is an emmensely spreading phenomenon that The proposed system is an IoT based design and implementation of
enables data transfer and control of gadgets using internet. It gives human Energy Meter which monitors real time energy being consumed using
the immense power of getting or transferring information to objects that Beaglebone black and LDR and displays it on a webpage by counting
are not in close proximity or actually not physically seeing the object pulses being generated. This system aims at making the whole system of
and/or making the object work and promotes automation. Especially with electricity billing convenient with the help of ever evolving technology. In
internet being so easily available and with the exponential spreading of this paper the problems associated with the whole process of generation of
Wi-Fi and wireless devices, IoT has become one of the best areas for units, unawareness of the people living in the house to the method of
future innovation. It also has been widely accepted that over fifty billion payments are discovered along with analysing of data and proposing
gadgets will be using this technology by the next two years (Darshan Iyer possible solutions for the same. By being aware of the amount of units
N. & K. A. Radhakrishna Rao, 2015). Electricity is one of the main that are being generated in the household a person will be aware of his
necessity of every single household and till today all we know is the right billings and will also be able to monitor if he has left a device
feedback received monthly in the form of units of energy consumed by the switched on while he leaves the house since his meter will keep running.
house. Only when a person is fully aware of the meter readings
continuously, will he be motivated towards reducing the consumption. Section 1 provides an overview of the motivation behind this concept
This will also make consumer come in direct contact with regulating while Section 2 identifies the target areas for which a solution is proposed
authorities (Anmar Arif, Muhannad, Al-Hussain, Nawaf Al-Mutairi, along with its implementation in Section 3, the results of which are
Essam Al-Ammar, Yasin Khan & Nazar Mallik, 2013) and decrease contained in Section 4 along with conclusion in Section 5.
chances of blunder. This will also motivate people towards energy
conservation and hopefully a greener tomorrow. Power theft is also 2. Background
associated with electricity and there has been some work done in that area
to detect the same and cut energy supplies on such lines (K. P. A. Identifying Issues and Challenges
Satheyamoorthy, 2013). There are various communication modules used
for the transfer of data over internet like ZigBee Module which accounts Currently there is no option available where a consumer is able
to regularly track the readings of the units of electricity that the

ELSEVIER-SSRN INFORMATION SYSTEMS & EBUSINESS NETWORK ISSN: 1556-5068

650
HTTPS://WWW.SSRN.COM/LINK/3RD-ICIOTCT-2018.HTML

Electronic copy available at: https://ssrn.com/abstract=3170196

3rd International Conference on Internet of Things and Connected Technologies (ICIoTCT), 2018

appliances consume. Power organisations hire workers for facility of online payment of bills in their login and through the
going to each household and collect reading of meter at regular app and e-mail service once the user crosses a critical value of
intervals. Thus making it labour intensive and increasing the usage which he can set himself. Therefore, the user will be
able to check his usage in real time. It’ll a also encourage the
liabilities of the organisation. Theft detection is another major
user to use less electricity so as to save money since he can see
problem faced by the companies. It is prevalent in almost
it that his usage is less or more compared to his neighbors.
every part of India and is one of the major reasons of revenue This system will have a switch or a relay which will be
and power loss (Shaik Salman siddiq, Shaik Reshma, K. controlled through the main server giving companies ease to
Anuja, P. PriyankaCh. Karuna Pravaha, 2017). turn off the electricity transmission autonomously from the
main server if the consumer doesn’t pay the bills within the
specified time. On the website along with the real time meter
Another concern here is the travelling vehicles required
readings the user can also view his previous readings or the
by these professionals that also negatively impacts the energy consumed in the past along with the graphical
environment. If the user is on a vacation, it’s not possible for representation of load and energy graph. Using nonlinear auto
regressive (NAR) neural network time series we predict the
meter reader to take readings (Chaithra Pallavi & D Sumana
future unit consumption of the user. The block diagram of the
M, 2015). Sometimes, many of the people are not available at
solution proposed in the paper is presented in Fig 1.
homes during the visit of the meter reader or might be doing
personal chores. Thus, the entering of an unknown person in For theft detection one of the best possible solutions is
the premises of their house is considered a breach of privacy equating the power received to that supplied (V. Preethi & G.
Harish,2016). If there is any discrepancy in the amount
and makes them uncomfortable. While noting down the
received the possibility of theft should be analyzed. One of the
readings there is a chance of the meter reader making blunder.
smart energy meter units will be connected to the pole system
Such errors could lead to both underpayment or overpayment in addition to the one’s connected in consumers’ premises.
of bills. On the other hand, large amount of economic losses is ΣPsent(Pole)=Pconsumed+LossNo Theft
incurred by companies due to non- payment of electricity bills. ΣPsent(Pole) Pconsumed+LossTheft
Many times consumer deliberately doesn’t pay the monthly
bills and it accumulates over a period of time to a large sum
3. Methodology and implementation
which ultimately becomes a problem for the companies to
collect as well as the consumer to pay. People sometimes leave The system shown in figure 1 can be broadly divided into 3 layers:-
on the appliances while they go out and is a major power and  Acquisition Layer: It consists of light dependent resistor
electricity wastage that takes place. Energy wastage thus needs (LDR) sensor to measure the number of pulses from the energy
to be reduced and an efficient way of doing the same is meter LED. The sensor data is used to calculate other
parameters. The data is then sent to a cloud database.
proposed. Most of the time user doesn’t have a track of the
Beaglebone Black is used to capture data and pushing data to
amount of energy he’s using. This leads to unexpectedly high the cloud database.
usage of electricity at times which results in increased charges.  Record Layer: Data from sensor is stored in Firebase Realtime
Database, which is a NoSQL cloud database.
B. Proposed Solution(s)  Application Layer: This layer provides a graphical user
interface to display the acquired data to the end user. This shall
Our solution is an IoT based Smart Energy Meter. It consists be achieved by developing a website to visualize the
of a website with a user login wherein user can check his usage parameters.
and the amount he needs to pay along with the deadlines,

ELSEVIER-SSRN INFORMATION SYSTEMS & EBUSINESS NETWORK ISSN: 1556-5068

651
HTTPS://WWW.SSRN.COM/LINK/3RD-ICIOTCT-2018.HTML

Electronic copy available at: https://ssrn.com/abstract=3170196

 3rd International Conference on Internet of Things and Connected Technologies (ICIoTCT), 2018

Fig 1- Block Diagram

For the implementation (Fig 2) in this system three bulbs are connected to
a meter using Beaglebone black and pulse count is obtained using Light
Dependent Resistor (LDR). The dataset used here onwards, is the one
generated from the three bulbs via pulses and tabulated in table 1.

Fig 3- Database

Fig 2- Implementation

Step-1: Achieved real time sensor data (LDR) is used to count pulses and
the data is uploaded the on Firebase Real-time Database. Python script
running on Beaglebone Black was used to push data to the database.
Hence, real-time sensor data was obtained. Fig. 3 shows the database.

ELSEVIER-SSRN INFORMATION SYSTEMS & EBUSINESS NETWORK ISSN: 1556-5068

652
HTTPS://WWW.SSRN.COM/LINK/3RD-ICIOTCT-2018.HTML

Electronic copy available at: https://ssrn.com/abstract=3170196

 3rd International Conference on Internet of Things and Connected Technologies (ICIoTCT), 2018

consumption value and hence a plot of cumulative unit consumed with

Step-2: Now, having created a website to visualize all these parameters respect to time is observed. The latest value corresponds to the total unit
stored in the cloud database. The website is synced to the database and consumption by the user up to that time.
updates all the parameters in real-time. HTML, CSS, JavaScript and
Firebase are used for developing the website. Fig. 4 shows the units
consumed and connected watt.

Fig. 6 shows the connected load graph with respect to time. It shows the
total watt of appliances connected to the energy meter at any instance of
time.

Fig 4- Webpage for live monitoring

Fig. 5 shows the cumulative unit consumption graph against time. The
number of units consumed keeps on adding to the previous unit

Fig 5- Units consumed per minute from Webpage

Fig 6- Load consumed per minute

ELSEVIER-SSRN INFORMATION SYSTEMS & EBUSINESS NETWORK ISSN: 1556-5068

653
HTTPS://WWW.SSRN.COM/LINK/3RD-ICIOTCT-2018.HTML
 3rd International Conference on Internet of Things and Connected Technologies (ICIoTCT), 2018

Fig. 7 shows the location of the installed meter on the webpage. This
location of the user is set by the user himself and can be used to tabulate
or compare the trend of energy consumption over regions and those
provide necessary arrangements.

Fig 7- Geographical location

Fig 9. E-mail being sent from the host to customer
Step-3: Next, the user might be interested in knowing the predicted unit
consumption and hence the bill amount. To accomplish this, Neural consumption and bill amount. Fig. 9 shows the mail received notifying
Network Time Series in MATLAB is used. The firebase database contains about the units consumed and price for a short period of time. It is
data in JSON format. Therefore, first data is read from the database URL assumed that the electricity rate is 5/- per unit and the calculated the price
into MATLAB and then the JSON data is parsed to a .csv file. Then, is shown in the figure.
loaded the data and used Non Linear Autoregressive (NAR) neural
network to predict the unit consumption time series. Fig. 9 shows the
model training, validation and testing graphs as well as error graph.

Fig 8- Neural Network Analysis

Step-4: Python script running on Beaglebone Black uses SMTP server

library to send mails to the user notifying him the daily energy

ELSEVIER-SSRN INFORMATION SYSTEMS & EBUSINESS NETWORK ISSN: 1556-5068

654
HTTPS://WWW.SSRN.COM/LINK/3RD-ICIOTCT-2018.HTML
 3rd International Conference on Internet of Things and Connected Technologies (ICIoTCT), 2018

and the load involved is also tabulated and presented in the form of a
Table 1- History Table from Webpage history table in table 1.

Date Time Pulse Units Watts

21/01/18 9:29:44 8 0.0025 150 4. Results and discussion

21/01/18 9:30:49 9 0.0053 168.75
The system efficiently provides the desired results and successful
21/01/18 9:31:49 10 0.0084 187.5 transmission and remote viewing of data that in this case is the units of
energy consumed over the internet which can be accessed by power
21/01/18 9:32:49 9 0.0112 168.75
organization and by the customer consuming electricity. Real-time
21/01/18 9:33:50 9 0.0140 168.75 tracking of data using a webpage and individual tracking of unit, load etc.
is also facilitated. Moreover, e-mail facility has also been successfully
21/01/18 9:34:50 11 0.0175 206.25
demonstrated. Possible future unit consumption prediction is also
21/01/18 9:35:52 9 0.0203 168.75 achieved using neural network time series function. Finally, a tabulation
21/01/18 9:36:52 9 0.0231 168.75 of all data is done.

21/01/18 9:37:54 10 0.0262 187.5

21/01/18 9:38:54 11 0.0296 206.25 5. Conclusion

21/01/18 9:39:55 8 0.0321 150
This system stands out because of its low cost, easy real-time
21/01/18 9:40:55 9 0.0350 168.75 implementation, ease of settlement and error detection, elimination of
21/01/18 9:41:56 11 0.0384 206.25 human involvement, online payment and ease of monitoring the entire
process from a far-off location. The implementation of neural networks in
21/01/18 9:42:56 9 0.0412 168.75
an energy meter system is completely a whole new innovation giving the
21/01/18 9:43:57 10 0.0443 187.5 user an idea of the future readings of a device. With the rate of progress of
smart devices, the way of living is made easier, same is the case with this
21/01/18 9:44:57 9 0.0471 168.75
system also which along with that also inculcates awareness about energy
21/01/18 9:45:59 10 0.0503 187.5 conservation by means of this system. In rural areas also, due to the low
21/01/18 9:47:00 9 0.0531 168.75 cost and very limited setup requirements the system can be easily put to
use. Although internet accessibility and power could be considered
21/01/18 9:48:02 11 0.0565 206.25 constraints but by the expansion of the telecommunication companies and
21/01/18 9:49:03 11 0.0600 206.25 more people now getting access to internet resolving the former limitation
with free internet or very low cost internet services, every rural household
21/01/18 9:50:03 8 0.0625 150
now also has access to them. In case people living in rural areas can’t
21/01/18 9:51:06 10 0.0656 187.5 monitor their consumption data on a website or an app there is still a
provision of sending monthly bills through SMS that can be extended.
21/01/18 9:52:06 9 0.0684 168.75
Also by using neural networks allows us to overcome errors in
21/01/18 9:53:07 10 0.0715 187.5 measurement and getting accurate output. Thus through the means of this
21/01/18 9:54:08 4 0.0728 75 paper highly accurate or high fidelity billing is achieved by Beaglebone
black and neural network along with remote services like billing etc. being
21/01/18 9:55:10 7 0.0750 131.25 provided with the help of IoT.
21/01/18 9:56:12 7 0.0771 131.25

21/01/18 9:57:13 6 0.0790 112.5

6. Literature Survey
21/01/18 9:58:14 4 0.0803 75
In paper by Dr. K. P. Satheyamoorthy (2013) the consumers are
The count of pulses produced by three bulbs, time along with energy units notified about daily consumption at twenty-four interval by the means of

ELSEVIER-SSRN INFORMATION SYSTEMS & EBUSINESS NETWORK ISSN: 1556-5068

655
HTTPS://WWW.SSRN.COM/LINK/3RD-ICIOTCT-2018.HTML
 3rd International Conference on Internet of Things and Connected Technologies (ICIoTCT), 2018

GSM along with load control using relay circuit. LCD is used to display Acknowledgements
also. In paper published by Darshan Iyer N and Dr. K A Radhakrishna
Rao in 2015, data is sent by using PIC18F46k22 which is a The authors would like to express their gratitude towards VIT University
microcontroller, a PLC modem is also being used to include a feature of that provided with all the facilities to overcome any hurdles that came
theft detection. This entire application required a very complex hardware along the way. The authors are also highly appreciative and would like to
and software components. In addition to this in 2016, Prince Sharma, thank their colleagues for their continuous motivation and support.
Rahul Agrawal, Rahul Choudhary and Rahul Jangid also suggested
supplying restricted amount of electricity for more savings. A paper by REFERENCES
Amrita Lodhi, Nikhil Kumar Jain and Prashant Chaturvedi in 2017 used
ARM which is a RISC based processor along with Arduino and Wi-Fi
Module is used to transfer data and implement IoT successfully. Here the Darshan Iyer N and Dr. K A Radhakrishna Rao,(2015). IoT Based Electricity
value of energy is first being digitalized and continuously being Energy Meter Reading, Theft Detection and Disconnection using PLC
modem and Power optimization, IJAREEIE, Vol 4, Issue 7, ISSN
transmitted to the official department and people. Tom Hargreavesn,
(Online) : 2278 – 8875.
Michael Nye and Jacquelin Burgess in 2010 studied the consumer
Anmar Arif, Muhannad Al-Hussain, Nawaf Al-Mutairi, Essam Al-Ammar,
behavior and analyze the effects of receiving feedback in terms of the Yasin Khan, Nazar Mallik, (2013). Experimental study and design of
energy they have consumed. It was evidently observed that the selection smart energy meter for the smart grid, International Renewable and
of target audience is utmost necessary and the information provided Sustainable Energy Conference, IEEE.
should be both informatory and unique to keep the audience motivated. K. P. Satheyamoorthy, (2013). Smart Energy Meter Load Control, IJAREEIE,
Thingspeak platform is used in paper by Shaik Salman siddiq, Shaik Vol 2, Issue 8.
S.Arun and Dr, Sidappa Naidu, (2012). Design and Implementation of
Reshma, K. Anuja, P. Priyanka and Ch. Karuna Pravaha to visualize the
Automatic Meter Reading System Using GSM, ZigBee through GPRS,
energy consumption using Arduino and LCD is also used to see the output
IJARCSSE Volume 2, Issue 5.
but is only limited to a small area. It used ESP8266 to send data over Wi- Shaik Salman siddiq, Shaik Reshma, K. Anuja, P. Priyanka, Ch. Karuna
Fi. MATLAB plotting was also done. Further Chaithra Pallavi and D Pravaha, (2017). Smart Electricity Meter using Wi-Fi, International
Sumana MS suggest that peak and non-peak hours should have different Journal of Research in Advent Technology, Vol.5, No.3, E-ISSN: 2321-
tariff rates and also uses RF modules for connection of various meters to a 9637.
collector . Further advancements are shown in paper by Karthiga S, Chaithra Pallavi, D Sumana M, (2015). Automated Controlling of Smart
Meters, International Journal for Computer Applications, Volume 121 –
Vignesh S, Vignesh S, Alex v Stanislavous and Kiran Thomas. The main
No.4, 0975 – 8887.
advantage is using ZigBee module for communication that reduces the V. Preethi and G. Harish,(2016). Design and implementation of smart energy
complexity and power consumption since it works on low power but is meter, International Conference on Inventive Computation Technologies,
area limited. In continuation to ZigBee, S. Arun and Sidappa Naidu also IEEE.
use GSM thus encompassing the drawback or range and making it suitable Prince Sharma, Rahul Agrawal, Rahul Choudhary, Rahul Jangid, (2016).
for long distance also. V. Preethi and G. Harish, also gives the feature in Wireless Smart Energy Meter, International Journal of Electrical and
Electronics Research, Vol 4, Issue 2.
which billing can be done in two ways pre-paid or post-paid. In the paper
Amrita Lodhi, Nikhil Kumar Jain, Prashant Chaturvedi, (2017). IOT Based
published by Anmar Arif, Muhannad Al-Hussain, Nawaf Al-Mutairi,
Energy Meter Rating, Vol 9, Issue 01, IJEEE.
Essam Al-Ammar, Yasin Khan and Nazar Mallik, Arduino Mega is used Tom Hargreavesn, Michael Nye, Jacquelin Burgess, (2010). Making energy
to improve the efficiency of Smart Grid of which the energy meter is a visible: A qualitative field study of how householders interact with
part. Md. Masudur Rahman, Noor-E-Jannat, Mohd. Ohidul Islam and Md. feedback from smart energy monitors, Energy Policy 38, 6111-6119,
Serazus Salakin used an optocoupler is used to shield the device Elsevier.
containing GSM-Arduino shield and performing GPRS based Karthiga S, Vignesh S, Vignesh S, Alex v Stanislavous, Kiran Thomas.
Automatic Energy Calculation Through Wireless Smart Meter Using
communication from any interference like voltages or currents. Nitesh
ZigBee, International Journal of Computer Science & Communication
Rawat, Sonia Rana, Bhuvesh Yadav and Neha Yadav suggest installing
Networks, Vol3(2), 117-125.
RF transceivers instead of GSM in order to reduce the overall price. Md. Masudur Rahman, Noor-E-Jannat, Mohd. Ohidul Islam, Md. Serazus
Finally in 2015 another module was introduced i.e. XBee by Himshekhar Salakin, (2015). Arduino and GSM based smart energy meter for
Das and L. C. Saikia that not only does all the things ZigBee can but also advanced metering and billing system, International Conference on
has a better range and throughput. Electrical Engineering and Information Communication Technology,
IEEE.
Nitesh Rawat, Sonia Rana, Bhuvesh Yadav, Neha Yadav, (2016). A review
paper on automatic energy meter reading system, INDIACom, IEEE.

ELSEVIER-SSRN INFORMATION SYSTEMS & EBUSINESS NETWORK ISSN: 1556-5068

656
HTTPS://WWW.SSRN.COM/LINK/3RD-ICIOTCT-2018.HTML
 3rd International Conference on Internet of Things and Connected Technologies (ICIoTCT), 2018

Himshekhar Das, L. C. Saikia, (2016). GSM enabled smart energy meter and Environment, IEEE.
automation of home appliances, International Conference on Power and

ELSEVIER-SSRN INFORMATION SYSTEMS & EBUSINESS NETWORK ISSN: 1556-5068

657
HTTPS://WWW.SSRN.COM/LINK/3RD-ICIOTCT-2018.HTML