Received December 21, 2020, accepted January 5, 2021, date of publication January 18, 2021, date of current version

January 29, 2021.

Digital Object Identifier 10.1109/ACCESS.2021.3051937

A Smart Home Architecture for Smart Energy

Consumption in a Residence With Multiple Users
SÉRGIO H. M. S. ANDRADE 1,2 , GUSTAVO O. CONTENTE1 , LUCAS B. RODRIGUES1 ,
LUIGUY X. LIMA1 , NANDAMUDI L. VIJAYKUMAR3,4 , AND CARLOS RENATO L. FRANCÊS2
1 Computer Engineering Area, Estácio Nazaré University, Belém 66055-260, Brazil
2 Postgraduate Program in Electrical Engineering, Federal University of Pará (UFPA), Belém 66075110, Brazil
3 National Institute for Space Research (INPE), São José dos Campos 12227-010, Brazil
4 Institute of Science and Technology, Federal University of São Paulo (UNIFESP), São José dos Campos 12247014, Brazil

Corresponding author: Sérgio H. M. S. Andrade ([email protected])

This work was supported in part by the Coordination for the Improvement of Higher Education (CAPES) under Finance Code 001, and in
part by the PROPESP/UFPA-Dean of Research and Graduate Programs of the Federal University of Pará (UFPA).

ABSTRACT Smart Energy Control Systems (SECS) has been increasingly implemented in the Smart
homes scenario, due to the possibility of conditioning and controlling residential energy consumption,
thus contributing to reducing energy losses and unnecessary electrical consumption. With the evolution of
embedded systems in conjunction with the Internet of Things (IoT), Smart Outlet (SO), and devices that
promote the Users Indoor Identification (UII) environments, they have assumed fundamental roles in the
acquisition of data from electrical devices and also in the mapping of the individualized consumption of each
resident throughout the day, providing essential information for SECS Systems as a way to assist in energy
balance with minimal impact on the daily usability of electrical equipment. However, in most of the works
that propose these types of assistance to SECS based on SO and UII, they have a massive implementation of
sensors throughout the residence, misinterpretation of the data generated by the residents, difficulty in the
identification of multiple residents. Thus, the present work proposes an evolution of an SECS architecture
called SmartCom, with the implementation of accurate identification of electrical equipment through Near
Field Communication (NFC)-based SO (data transfer between the appliance and the SO) and of multiple
inhabitants through Wi-Fi handover using smartphones, with the least possible impact on the user’s comfort,
as well as in the building structure, achieving a rebalanced residential energy consumption 87.3% of the time
it was used.

INDEX TERMS Smart energy control systems, smart outlet, user’s indoor identification, SmartCom.

I. INTRODUCTION Regulatory Authority (such as in the areas of maintenance,

The electric power systems have undergone significant control, and supervision), while also fulfilling the follow-
changes in the last few years, owing to the introduction of ing prerequisites: management, controlling, reliability, clean
new business formats as in the electrical power distribution sources of renewable energy generation, cost-effectiveness
system, for example. These formats allow end users to be and interoperability. These attributes are all required within
included in the dynamics of Smart Grid (SG) networking. the SG domains [1].
This interaction results in a complex scenario concerning the Overall, an SG is a network that can manage electrical
services as it raises some serious challenges, for example, the equipment and systems in various fields and bring security,
question of how to combine hardware and software features. viability, efficiency, and quality of service in an intelligent
This is due to the heterogeneous environments in which and trustworthy way. There are seven interconnected areas
the power companies and power systems operate, which is, in those fields [2]. The first four (distribution, transmission,
in turn, a key factor when ensuring a suitable level of security. end-users, and generation on a large scale) are responsible
Therefore, it is recommended that a strategy that is suited for generation, transmission, and distribution. To provide full
to all aspects of the electricity area is to be adopted. Moreover, management between customers and the Advanced Meter-
the new model must satisfy the requirements of the Electricity ing Infrastructure (AMI), for example, data exchange, these
fields must ensure bi-directional communication. The energy
The associate editor coordinating the review of this manuscript and managing market, service delivery, and energy supply refer
approving it for publication was Macarena Espinilla . to the last three (operations, market suppliers, and services).
This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 License.
VOLUME 9, 2021 For more information, see https://creativecommons.org/licenses/by-nc-nd/4.0/ 16807
 S. H. M. S. Andrade et al.: Smart Home Architecture for Smart Energy Consumption in a Residence With Multiple Users

may be out of range due to the short range of the ZigBee

protocol, resulting in less reliable collection of data.
Therefore, this paper is an evolution of [6], [7], which now
allows the identification and tracking of multiple users by
internal Wi-Fi handover by making use of smartphones, and
through the use of SO technology using NFC identification
to extract accurate data from consumption, electric current,
voltage and identification of electric household appliances.
In this new architecture it is also used the LoRaWAN proto-
col, which is a long range and reliable protocol to control and
FIGURE 1. IoT architecture in the context of SHs. gather data of household appliances.
This new proposal seeks to assist multiple users to achieve
the ideal electrical energy consumption regardless of their
The term ‘customer domain’, [3] refers to energy managing location in the house and what appliance they are currently
on the client side, along with the organization and control using, utilizing a recommender system to preserve electrical
of electrical devices to find stability and improve energy energy as well as user’s comfort.
consumption at home. The purpose of this article is to tackle
one of the nine requirements for SG applications (energy
efficiency and demand response), which is exactly the one II. RELATED LITERATURE
used in the SH scenario. Within the overall SH scenario, studies in the literature have
Based on the development of SH applications, a new struc- recommended several kinds of HEMS to reduce energy con-
ture was designed that associates data technology solutions, sumption costs. This reduction can be achieved through a
innovative communication systems, and indoor identification new exclusive feature for control and by monitoring or even
with users and sensors to build a wide range of modern SH making use of decision-making techniques solutions to assist
applications connected to the IoT. The working principle of in optimizing power consumption.
an architecture present within the scenario of an smart home Several studies have been carried out that adopt a single
is demonstrated in Figure 1, in which it is illustrated the data approach to regulate an efficient measuring system required
transfer between local server to cloud server and the remote for residential homes. Some of those studies are similar to
access using a mobile application. those discussed in this article, like [8] and [9] for example.
With all devices connected to the internet as a result of Outlined in [10] is a HEMS that is capable of handling house-
the IoT, the user is allowed to connect to home appliances hold appliances and lighting (through the use of consumption
both internally via an intranet and externally via the internet, data collection). As examined in detail in [11] and [12], these
opening up a range of possibilities for intelligent services in HEMS models can be used for monitoring and informing
the residential scope. The following requirements are needed users about the home energy usage rates through I) A mobile
to provide users to monitor, access and remotely control application, II) The WEB storage system and III) Sensors
devices using the Internet [4], with an objective to achieve scattered around the houses.
efficiency and viability. Among the SH domain solutions already mentioned, some
Considering the above, this article has created a new Home are based on computational intelligence. In order to automati-
Energy Management Systems (HEMS) architecture. This is cally optimize the temperature and consequently save energy,
known as smart management consumer architecture and has fuzzy logic was used combined with thermal sensors to adjust
been based on the rules set out in [5] for interoperability and the area temperature, and standards were established because
viability of SGs into the context of IoT in SHs technology. of the guidelines recommended by the residents [13]. In [14],
The HEMS architecture is adjustable and includes impor- fuzzy logic is applied to decrease the energy requirements of
tant technological resources, such as remote-control capa- a home with regard to several specifications, such as outside
bility, interoperability, reuse, modularity, flexibility, data temperature, adjusted schedules, battery state- of- charge, and
security, and the possible use of computational decision- a wide range of prices and preferences for uses of electricity.
making techniques. Additionally, the scheme has as an objec- Fuzzy logic can also be used in a context-responsive sce-
tive to incorporate elements beyond the SH field such as nario for recognition. Fuzzy logic in computational intelli-
SG. It creates a single integrated interface system with an gence is capable of operating with an approximately 95%
interoperable layer, providing a web service with the tool by rate of accuracy and is slightly faster than other architectures,
using an AMI applied in the cloud. as explained in [15].
For example, in [6], [7], it is proposed solutions to identify Looking to control and optimize energy consumption,
only a single user’s consumption inside a residential home, other solutions of this kind are also used. A Bayesian Net-
while most of the residential homes have multiple inhabitants. work (BN) is adapted and used to control power supplies
The ZigBee protocol is also used in [6], [7] to control and designed for comfort in one of these solutions, as observed
gather information of house appliances, but distant appliances in [16]. This structure will probably form a link between the

16808 VOLUME 9, 2021

S. H. M. S. Andrade et al.: Smart Home Architecture for Smart Energy Consumption in a Residence With Multiple Users

condition of the equipment and the environment of residential communications (ZigBee, Bluetooth, and RF communica-
homes. tion) together with machine learning to try to predict, through
An SG solution is established on cloud computing and probability calculations using Markovian queuing theory,
Web Services is put forward in [17], using IoT sensors how many people will be present, in which location and
and devices with Artificial Intelligence (AI). In the study, what would be the ideal temperature, having their data sent
a LoRaWAN connection is established to control some of the to a database for the interpretation and optimization of the
home appliances while sensors and actuators are used to mea- HVAC to be made. It was concluded that among the tested
sure environmental variables. The traffic data is collected and IoT communications, LoRaWAN performed better with 60%
stored in the cloud server, by determining the interoperability more covered area compared to short range communications,
measurement units and hence allowing the users access via also achieving 0% packet loss.
notebook or smartphone. The aim of the paper in [22], is to evaluate the performance
In [18], an architecture is designed based on the concept of of LPWA (Low – Power Wide Area) technology based on
IoT to enhance remote access systems while simultaneously a LoRaWAN solution for supporting SG applications, espe-
a) monitoring and managing a user’s home and b) providing cially smart metering. It is based on a very simple wireless
a more comfortable and secure system. The users will be infrastructure and on a hierarchical structure that is suitable
granted access to the web or mobile applications that can for many distributed measurements and IoT applications.
make use of cloud services, including heuristic systems for Since the network is based on a star topology, it can make
SHs. an interconnection between the end nodes and the base sta-
In [19], the author finds a solution for power consump- tions/gateways that collect and forward messages to and from
tion management and optimization to improve the security the Internet toward one or more Network Server(s). The
of the user’s home based on IoT. This solution includes a Application Server secures the user data sent to/received from
relay module used to manage the device, security sensors the end device.
for smoke detection, and a central device (in communication In [23], there is a discussion of an electrical outlet formed
with the internet) and the communication between the user by the ATmega328 microcontroller and based on the ZigBee
and the device takes place through the Internet and mobile Protocol, which has devices to establish wireless communica-
applications for home management. tion between sensors, sockets, and the web server. In addition,
An intelligent home application based on the concept of it carries out the function of measuring voltage, together with
the IoT is also developed in [19], although the application current and power consumption, and making this information
is deployed for remote access through the Internet or locally available to the consumer, who can monitor real- time power
without the need for Internet access. This solution can be consumption through devices with a web browser and inter-
used for mobile devices that have access to a central host. vene by interrupting the electrical transmission to the plugs,
This application can be used to manage numerous home if necessary.
appliances and control security. A framework is established in [24], that is capable of
A monitoring architecture was developed in [20] with the determining single-user interactions in a multi-inhabitant SH
purpose of energy saving in the residential environment, through the use of multiple sensors installed in different
using IoT and Machine Learning techniques. The physical parts of the house. This solution utilizes the Floyd-Warshall
part of the monitoring system makes use of a non-invasive algorithm to calculate the distance between the sensors and
current sensor connected to a Raspberry Pi 3 A+, which the Sequential Importance Resampling (SIR) algorithm to
transmits information that is then stored in the Pi. The Pi determine the actual number of inhabitants and their initial
transmits this data (date, time, globally active and reactive location, based on the sensor’s response.
power, voltage, current, and active power readings from A mechanism was developed to identify the location of
3 different appliances) to the IoT middleware layer, where users through infrared sensors in [25]. The project achieved
the data is processed and stored on a cloud server. Finally, good results but failed to achieve one of its main objectives;
the processed data is displayed in an application layer for this was because it was impractical owing to the price and
better visualization. Mean squared error (MSE) and R2 tests number of sensors that had to be distributed throughout the
were used, by which it was determined that the developed house, as they were unable to cover a large area.
LSTM (Long Short-Term Memory) model had an accuracy Because of the cost and other problems arising from the
of approximately 84%. excessive use of sensors, it was decided that the best solution
It is discussed in [21] the use and importance of IoT was the use of intermediate communication and tracking
communication in buildings and in universities, working through a smartphone, which is the most widely used device
together with machine learning to conserve energy, moni- in the world and which most people carry around for almost
toring through the use of temperature and CO2 sensors to twenty-four hours a day.
determine how many users are present in order to manage A Hierarchical Dynamic Bayesian Network (HDBN)-
the HVAC (heating, ventilation, and air conditioning). The based model is used for activity recognition. This model
communication between the sensors and the database will employs inter-and-intra individual correlations and con-
be done using both long range (LoRaWAN) and short-range straints, at both micro-activity and macro activity levels,

VOLUME 9, 2021 16809

 S. H. M. S. Andrade et al.: Smart Home Architecture for Smart Energy Consumption in a Residence With Multiple Users

to recognize individual activities accurately. With the aid of

mobile applications, this project allows the users to make use
of their smartphones, along with the sensors, to determine
which part of the house they are currently located in and
whether they are together or not.

III. BACKGROUND
A. COMPUTATIONAL DECISION-MAKING TECHNIQUES
Specific biological system models are referred to as the
concept of Computational Intelligence (CI) and decision-
making techniques, which has a type of efficiency technique
in order to help users, allowing systems to perform smart
decision-making actions. The type of insight instigated by
computational decision-making represented has provided rise
to the following standards:
An interoperable model (this presumes that operated intel-
ligence such as neural networks arises from the dynamics
of connections between nodes and layers known as artificial
neurons).
Symbolic (consists of a set of rules and procedures and a
set of symbols to form structures, such as decision trees).
Fuzzy (this allows the classification of a specific com- FIGURE 2. Fuzzy inference machine structure.
ponent associated with a particular class, that is, to assess
in case of the aspect belongs to a class of greater or lesser
significance, such as Fuzzy Logic). a membership function while restarting it in the following
Evolutionary computation (based on processes existing instead.
in nature essence, which can develop, and have adaptable In [6], [7], the fuzzy inference machine was implemented
management, self-organization, and genetic algorithms). for residential systems optics for energy management. This
Because of these characteristics, they are validated for method obtained good results and, therefore, was used in this
use in functions that include monitoring, automation, man- work to assist in the implementation of handover technology
agement, and actuation, which are the key features that are and smart metering.
important for SG applications and SHs, a bigger importance As shown in Figure 2, the structure of the fuzzy inference
is laid on decision-making systems based on fuzzy in this machine has two input variables, which are the Consumer
article. While also allowing fuzzy-based systems to provide Index and the Monthly Percentage. These will be the input for
as possible options to systems which have the resources the 26 fuzzy rules present in the inference machine that will
explained in this article, some factors, such as agreeable be output after defuzzification, the Forecasting Consumption
computational effort, include solutions that establish good variable that will allow the inference of the user’s consump-
assistance for simplicity of rules and decision-making. tion forecast.
These fuzzy techniques can handle the complex variables The structure will then be able to use, in fuzzy logic, all
by sorting out and/or forecasting the outcomes to help in the fuzzy sets of the response parameters. Such a conclu-
the decision-making process, to notify residents about a pos- sion is because of a set of Fuzzy rules that were previously
sible increase in electricity consumption while also apply- defined and are viewed from the standpoint of a professional,
ing the optimal strategies for power consumption, based on as shown in the chart below.
the parameterizing which occurs when a smart consumption For instance, Figure 2 shows how the input value for the
architecture is configured. Power Consumption Indicator can be converted into a deriva-
The fuzzy logic is one of the methods most often discussed tive set with no values for Average Energy Consumption
in the professional literature while working with an auto- (A.E.C) and Above Average Energy Consumption (A.A.E.C),
mated system, and it also provides the means to communicate A.A.E.C having escalating levels such as 20% A.A.E.C, 40%
with the processes that are intrinsically similar to a com- A.A.E.C, 60% A.A.E.C, 80% A.A.E.C and 100% A.A.E.C.
puter [26]. It entails separating constant numeric inputs into These will receive a grade of 1 for A.E.C and >1 if it is inside
various situations. These states, either distorted or overlap- the A.A.E.C consumption range.
ping, are defined through the use of Membership Functions The framework is established by integrating all the val-
(MF). The state of a variable does not change suddenly during ues that are entered. The structure sets regulations within a
the overlapping procedures, but it slowly loses its value in specific situation. For instance, if the Power Consumption

16810 VOLUME 9, 2021

S. H. M. S. Andrade et al.: Smart Home Architecture for Smart Energy Consumption in a Residence With Multiple Users

Indicator has a combination of 1 "A.E.C." and "20% (c) Energy management on demand (Profile 3): The user
A.A.E.C." >1, all the rules, including "A.E.C." or "20% will have complete autonomy to choose which appli-
A.A.E.C.", will be checked. These rules will result in fuzzy ance should have its consumption reduced to maintain
output values that are then assembled in order to create a the estimated consumption limit.
fuzzy set for each resulting variable, with the weighting factor It is possible to assist the end consumer if they exceed their
normally varying between 0 indicating no result and 1 indi- daily average rate of energy consumption while looking for
cating the maximum result. Ultimately, the fuzzy set acquired a way to preserve user service and all the essential amenities
is used to calculate the outcome of the output variable in the in order to guarantee a lower impact from the usage of appli-
defuzzification phase. ances. With that in mind, the plans explained above take into
When determining the optimization goal for energy con- consideration the extent to which the comfort that each piece
sumption, the Indicator of Comfort (IC) is calculated from of equipment in a household offers to each user. This paper
the results of the fuzzy logic. will use the first approach as a reference point.
When establishing the standard for the comfort zone with
regard to household appliances and the level of human com- B. HANDOVER
fort, the formula for the Indicator of Comfort (IC) was Apart from identifying the total energy consumption in an
expressed as follows: SH, one of the main objectives of this architecture is also to
share this information with all the users through the use of a
IC = (EUH / NOH)∗ WD (1) mobile application, by giving them an individual electricity
where EUH is equipment usage hours, NOH is the number consumption limit for each appliance which could be shared
of hours while at home and WD is the number of days in a if two or more people are in the same room.
week. It was decided that the best solution was the use of inter-
Hereafter, the implementation assumes one of the follow- mediate communication and tracking through a smartphone,
ing approaches: which is the most widely used device in the world and which
(a) Balancing supply and demand in Reducing the excess most people carry around for almost twenty-four hours a day.
of Energy Consumption (Profile 1): Excessive con- With this idea in mind, a system was devised that resembles
sumption is offset by the amount available for other handover.
appliances based on the value of the total consump- Essentially, handover is a process of exchanging access
tion of each appliance. This new rate of consumption points (AP) based on signal quality identification. For exam-
amount is calculated employing the following formula: ple, if there are two APs, and the signal a smartphone is
currently connected to is weak, the handover tends to switch
ECP = (KEC ∗ 100) / HEC (2) from that AP to the one with the best signal. Handover is
CRP = (EC ∗ ECP) / 100 (3) essential not only for comfort but also for applications that
use real-time data exchange, for example, phone calls like
NEC = (KEC – CRP) (4)
these require a quick data exchange, and when on the move,
where in Formula 2, the ECP (Equipment Consumption a manual exchange is impracticable.
Percentage) is the percentage of consumption equip- The handover technology is of great importance to this
ment that is given with regard to a) potential energy; project, as it enables multiple users to be identified in real
b) KEC (KWh Equipment Consumption), the con- time, together with a description of where this person is,
sumption of the given equipment in KWh; c) HEC what appliance is currently in use, and how much energy that
(House Equipment Consumption), the value in KWh person is consuming, regardless of the location.
of the total energy consumption of the residence. A micro network can be created at the user’s home that
In Formula 3, the CRP (Consumption Reduce consists of Arduino microcontrollers, together with the ESP
Percentage) represents the percentage by which a 8266 that will simulate an access point (AP). This microcon-
given appliance will reduce its exceeded consumption; troller will be used to identify particular rooms in the house
EC (Equipment Consumption), consumption exceeded by Received Signal Strength Indication (RSSI)and determine
from the average consumption. how many users are in those rooms and who exactly they
And finally, in Formula 4, NEC (New Equipment are. This search will be carried out by mapping the signal
Consumption) will be the new consumption assigned strength, and this signal will be restricted to the area of each
to each appliance, in order to reestablish the energy room through a microcontroller that will be positioned in the
balance. center of the room.
(b) Reducing the excess of electricity consumption based This mapping will be done in each room at a time, with the
on a Priority Sequence (Profile 2): An energy con- help of a small tutorial that will require the user to stand:
sumption limit set for the most widely used appliances (a) Under the microcontroller.
is going to be kept and the excessive consumption is (b) In the corners of each room.
reduced from the least often used appliances to main- (c) Between the microcontroller and the exit door of the
tain the estimated consumption limit. room.

VOLUME 9, 2021 16811

 S. H. M. S. Andrade et al.: Smart Home Architecture for Smart Energy Consumption in a Residence With Multiple Users

(d) And finally, by the exit doors, in order to provide each

room dBm limits to the HEMS.
The information will be sent to a database along with the
identification of the user, by using the MAC address of the
smartphone, together with the information about each part
of the residence, such as the bedrooms and suite bathrooms,
so that consumption can be shared more easily.
The way the handover process works in this project is as
follows:
(a) In each room of the house at a central point (or close to
it), there will be a microcontroller that will be an AP.
(b) Each environment that has an AP based on the micro-
controller, will be mapped by the dBm (RSSI) levels
present at its ends through the software developed for
android devices. This mapping will be done by the user,
simply and intuitively.
(c) The smartphone that belongs to a given user, present in
a given environment, will send a Wi-Fi scan that will
enable to connect to the nearest and with the strongest
microcontroller AP signal, which in this case, will be
the one located in the environment in which the mobile
FIGURE 3. RSSI signal attenuation exemplification due to the natural
device is found. Thus, if a mobile device is connected barriers of a building.
to an AP and it is within the dBm limits pre-established
for that environment, the HEMS system will indicate of antennas (AP) throughout the residence, has also grown.
that the user is located in that specific environment. Therefore, facilitating the inclusion of APs based on micro-
(d) When the user moves to another room, the dBm signal controllers, such as network expanders and user identifiers.
from the AP of the previous room that the User was
connected to tends to weaken. With this, the software C. INTERNET OF THINGS
will make the analysis identifying that the user is out- Internet of Things is a network of appliances and devices,
side the pre-established RSSI limits for that particular which detect and share information about the physical world
environment, thus causing his disconnection from the in real time, they can be objects, buildings, machines, vehi-
old AP, and the realization of a new Wi-Fi scan in cles, and other physical systems with built-in computing,
search of an AP with the stronger signal, consequently communication, and sensory resources [29].
implying handover between APs. Therefore, the system In order to give the end users access to all devices con-
will identify the user’s new location. nected at home, IoT solutions are used for home control,
The natural barriers of a building, such as a masonry, plas- which includes communication and information technologies
terboard, and others, which promote the division of rooms in a and may also involve the remote control of appliances in real
residence, end up helping in the indoor positioning of the User time, as well as information sharing.
through RSSI and in the handover of AP, since the intensity Since it is divided into layers that represent the operation
of the Wi-Fi signal within an analyzed environment it is much and management architecture, the IoT can connect elements
stronger than when analyzing the same when leaving the from the real world and include others in the communication
environment. This phenomenon is depicted in Figure 3 occurs system that can intelligently process its detailed informa-
due to the attenuation of the Wi-Fi signal promoted by these tion and autonomous decisions [30]. These layers can be
natural barriers of the building, which cause a decrease of described as collectors, actuators, and sensors in the lower
around 10 dBm. layer, supported by the network layer, which has an essential
Thus, the moment the user leaves the environment in which role for transmission and security of the data provided to the
he is connected to the AP, the attenuation of the signal makes upper layer, that are responsible for creating smart strategies
the HEMS system clearly realize that due to its low connec- linked to rules and business applications. The Middleware
tion intensity with that particular AP, the user is no longer Layer has been designed to provide management services to
inside the room. the lower layers and data storage.
Therefore, this implementation denotes have a differential In addition, there are great volumes of data that could be
compared to other projects using sensors [24], [25], [27], [28], collected from various segments such as SOs and handover,
or even using only the MAC Address, when analyzed in the which will provide consumption data and user location. It is
user identification part with little change in the environment, necessary to combine the IoT with cloud services in order
since having a homogeneous distribution of the Wi-Fi signal to be able to analyze and process data, providing useful
in homes is already desired today, as well as the number information for the final users.

16812 VOLUME 9, 2021

S. H. M. S. Andrade et al.: Smart Home Architecture for Smart Energy Consumption in a Residence With Multiple Users

A widely used technology in the IoT scenario is the (b) Sectional: As this is an open architecture, the func-
cloud computing service, which, in addition to providing tionalities discussed in this project are not applicable.
shared storage, resources, information, and software, pro- If the communication protocol is maintained, it will
vides remote access for all the appliances connected in the be possible to extend other management modules (for
network [31]. example, lighting management).
The main goal of this paper is to develop an architecture (c) Adaptable: Implementations can be made in specifi-
that can assist the user in making decisions about their con- cations other than those listed in this project (paper).
sumption through a mobile application in which, together It should be noted though, that the solutions described
with IoT technology, the appliances that are consuming elec- in this paper are mostly those discussed in the published
tricity and how much is being consumed can be identified literature.
through the implementation of SOs. (d) Scalable: This allows new modules to be included,
It is also possible to identify multiple users by imple- while also extending the network management.
menting a system with handover characteristics. Having these (e) Accessible: Providing that there are prior permission
data stored in the cloud, the user can have this information and Internet access, there is no restriction of access
available on the web and also through the application itself, that prevents users from controlling their monitored
being able to observe in real time his/her energy consumption appliances, regardless of their location.
and rebalance it when necessary. In addition to tackling the problem of how consumers
To make sure that the system can work properly and that should be responsible for home management, this archi-
any user, with or without prior knowledge, can understand tecture also designs middleware that is loosely associated
how to use it, it is necessary to: with publishing/subscription capabilities (IoT services). This
(a) Download and register the user account in the mobile serves utility purposes and allows communication between
application. systems through an exchange of messages, as well as result-
(b) Map the limits of each room in the house based on ing in different solutions that can assist interoperability at
physical limits such as walls and doors, so that the various categories of the middleware client. This means that
energy study and indoor location can be done. the system can improve scalability without being restricted to
(c) Define their monthly electricity consumption limit, a specific solution.
in order to establish the maximum consumption that the The architecture set out in this paper permits the use
house can have. of heuristics based on computational decision-making tech-
If this limit is exceeded, the user will be notified by the
niques together with the adoption of smart management ser-
mobile application through a pop-up notification and will
vices. As a result, the evaluation of the variables obtained
be offered to choose one of the three profiles previously
from the instrumentation and extraction standards can assist
mentioned to rebalance their energy consumption.
in characterizing the consumption of each consumer, and
IV. THE PLANNED ARCHITECTURE eventually be applied to the optimized model designed for
This architecture sets out an innovative model based on IoT efficient measurement. Therefore, this architecture is not
services of interoperability for SHs. This can be achieved restricted to extract and study basic patterns of energy con-
by defining a middleware procedure based on REST API sumption for each monitored user but also be extended to
(Representational State Transfer), which integrates company acting as a communication channel between varied elements
monitoring systems with the measurement techniques avail- of data traffic.
able to the consumers. Figure 4 represents the structured model that supports the
The architecture also allows the control of alternative structured model that supports the architecture set out in this
sources of energy (distributed generation) and the automation paper. Each layer represents a specific role that includes the
of home appliances through smart devices and controlling the operational features of the architecture which can be adjusted
rate of message consumption, for example by setting out rules to improve adaptability and scalability. Also, Figure 4 illus-
for scheduling daily activities, as well as the administration trates many of the essential components that are required
of the real time energy consumption of household appliances. for the implementation and interoperability of the proposed
The goal of the architecture is to maintain the underly- architecture.
ing principles for the development of SH interoperability The measurement node proposed, as a new version of the
applications in a dependable, sectional, adaptable, scalable, old one present in [7], has the function of checking the data of
and accessible way since these are solutions that are widely electronic equipment through voltage and current sensors to
used in SH domains. It also lays down the requirements obtain consumption data, as well as extracting other data such
for providing cloud management services and is carried out for example, in the precise identification of which electronic
generically to ensure other effective solutions could be found equipment is actually connected to that outlet using NFC
in the future. The planned architecture characteristics are: technology.
(a) Dependable: This architecture integrates a method for Each SO, as well as each connector of electronic equip-
both the verification and safety of data at the applica- ment, will have an NFC microchip. On the electronics side,
tion layer of the network. through the software developed for android, the user will be

VOLUME 9, 2021 16813

 S. H. M. S. Andrade et al.: Smart Home Architecture for Smart Energy Consumption in a Residence With Multiple Users

There is a separation between networks where:

(a) For end users, there is an alignment with open standards
(such as Ethernet and Wi-Fi) already recommended by
the SG through which they can make use of to request
utility services.
(b) For the power companies, a network that efficiently
manages the requested service by adhering to a closed
standard is used, that is less likely to be vulnerable, such
as PLC or 4G.
The external communication interface helps users to han-
dle the data and the external environment in the central unit.
To ensure the network access technologies are transparent
they need to be subjected to inspection applications. The
key feature when providing access to data in heterogeneous
domains is communication, which entails establishing real-
time access for users no matter where they are and a deter-
FIGURE 4. Proposed architecture components diagram. mining factor in making this architecture work.
This is defined as the ‘‘main administration" or ‘‘aspects of
able to insert the electronics data as brand and model in the administration’’. These include data, adapters, connections,
present NFC tag. contexts, security and rules, all of which require the use of
On the NFC side present in the SO, it will receive the the architecture to function properly.
precise data (brand and model) of each device that will be Several connections of the measurement nodes and the
connected to it and, together with the voltage and current data, tracking nodes are installed on the consumer side, and the
they will be transferred via LoRaWAN protocol, which is also connection administration is used to manage the communica-
another evolution of the SmartCom [7], for the central unit. tion between the central unit and these nodes. This involves
Another evolution present in the work was the tracking the inclusion of adapter management, which permits commu-
node that also communicates with the central unit, which will nication between the middleware types, as the communica-
have the main function of identifying as well as locating the tion interfaces must be installed via Web Service. In addition,
user within his/her residence, by indoor handover methods since connection management is a means of restricting traffic,
that will take into account the signal strength RSSI in relation supervisory messages can be prioritized as needed, given an
to the APs. urgent request from the utility’s monitoring system.
The interoperability of all the components in this architec- For managing environment variables in the architecture
ture, whether consisting of hardware or software, is of great and communication ports, the responsibility is at the context
importance to ensure they are following accessible require- management, including message priority classes for control,
ments that might need to be adjusted to the exact features of encryption rules, and so forth. However, an identification of
a particular scenario or external and internal communication each device monitored by a measurement node is needed for
interfaces as well as this, the communication protocols must a more effective and transparent means of control.
be compatible with the REST API. It is owing to the interop- Each device in the architecture requires an identification
erability of the hardware and software components, as well as protocol and its category (if necessary) so that it can be acti-
the adoption of open standards, that these can be customized vated by the surveillance system. Defining rules and strate-
to suit the needs of the user or manufacturer. gies for optimizing the consumers’ energy when developing
the identification field is severely important and provides a
A. MEASUREMENT AND TRACKING NODES AND clear view of consumption patterns.
CENTRAL UNIT LAYERS Fuzzy logic is a decision-making technique required in this
Elements with communication interfaces are provided for paper to be an accountable implementation for developing
the central unit, measurement, and tracking nodes to oper- optimization rules that can be applied to smart appliances or
ate the communication functions specified in the architec- extracting consumption patterns.
ture. An attempt is made to maintain the integrity of the As real-time location systems (RTLS) provide information
system by creating a transmission channel for the features that is private to the resident, not only because of the location
of measurement, to prevent any loss of data. LoRaWAN but also because of the MAC that is required for user verifica-
is a wireless network protocol (IEEE 802.11ah) for low tion, this information will not be sent to a possible communi-
frequency networks and sensors and is generally utilized cation center with the energy retainer. However, the function
to create a network platform for home security, indus- of the REST API will be to make the databases ‘‘talk to
trial, and IoT applications. Moreover, if there is a bottle- each other’’ so that the fuzzy logic can work correctly and
neck in the network, this protocol will be used as internal obtain the information from the measurement and tracking
communication. node. This will result in consumption profiles that have better

16814 VOLUME 9, 2021

S. H. M. S. Andrade et al.: Smart Home Architecture for Smart Energy Consumption in a Residence With Multiple Users

control of the stipulated limit defined by the end user, (which

will be discussed later).
The data management enables the measurement and track-
ing devices to be handled and stored, together with the
environment variables and data, (whether they are local or
remote), and these are obtained from remote databases in
the cloud. In the future, store rules will be enforced that are
based on metaheuristics (i.e., the management rules). Both
supervisors and consumers must be present in this layer.
The security administration system will be responsible
for creating security standards for transactions, certification, FIGURE 5. Measurement node prototype.
access, adversity in recovery, and encryption, which can
be applied to services implemented on the platform from
an application perspective. Since this design provides the management application for monitored appliances and acts
use of web services established on REST API, the secure as an intermediary for remote management requests.
utilization of frameworks is highly recommended, such as This layer also provides an efficient method of classifica-
the OAuth2 framework, to ensure that the information and tion based on energy consumption and management, anal-
access provided are safe and reliable. Also, issues such as ysis, and optimization rules that can be established for the
communication networks between supplier companies and instrumentation of each user. That is done by specifying meta-
end users or external issues are not addressed to this layer. heuristic services (through computational decision- making
However, this research study assumes that the network and techniques).
the utility have met the security requirements. The specialist Local and remote applications can use REST API requests
literature has revealed numerous ways to implement commu- to manage monitored devices and establish full control of the
nication networks for this purpose, as described in [5], for available services that are based on access profiles.
example.
The middleware layer can be used to switch to renewable
V. IMPLEMENTATION
energy sources for energy regulation or send control mes-
Measurement and tracking hardware are created for the val-
sages such as how to turn the appliances off remotely and
idation of the architecture, as well as its software that will
offer a view of all the components. It is also used to show
enable, automate, and control the residence.
how each user’s internal and external environment can be
used to store variables. Additionally, it is in charge of creating
interaction with cloud and web services. A. HARDWARE
As it uses its tools without interfaces to be integrated with Based on the features discussed earlier, the central node
other new applications more easily, it is a separate module and components must be adept to manage and monitor the
is under the supervision of the utility. Nevertheless, customers measurement nodes and also provide the installation of the
and supervisors can rely on REST API to perform manage- applications or other functions (both the middleware and
ment tasks at this layer. additional software created with the objective to store the
The measurement node designates the main layer that is data), while at the same time meet the demand of the
needed to capture all the measurement readings and acquire customer.
data from monitored devices. As well as this, it is responsi- After collecting data from the sensors, the measurement
ble for handling actuation and control messages, as well as node then transmits it via LoRaWAN to the central node.
collating the data sent to and from the network. It not only collects information but also disrupts household
appliances (power on / off). In this scenario, each outlet will
B. LOCAL AND CLOUD APPLICATIONS LAYERS be fitted with a measurement node that can read each device
A mechanism is provided by the local and web applica- connected to the electrical network for better measurement
tion layers to check and control components belonging to efficiency.
the architecture. These applications must work successfully As well as storing information such as current, voltage,
regardless of the platform and be compatible with the REST user’s electrical energy consumption, data, and usage time,
API services. the central node also performs a crucial role in monitoring and
The cloud layer provides a method of displaying infor- controlling any connected electrical devices. All the informa-
mation in any device in communication with the Internet, tion is encapsulated and sent via LoRaWAN communication
and thus centralizes services which are essential for remote and forwarded to a central unit.
access and monitoring the connected elements. This layer is The measurement node that is recommended identifies
responsible for the data transferred by the central unit storing the type of appliance connected to it, through the NFC tag on
all the data of the tracking and measurement nodes and which the device plug is depicted in Figure 5. The tag is read by the
database service will manage these data. It also provides a NFC reader that can be found in the SO, and the information

VOLUME 9, 2021 16815

 S. H. M. S. Andrade et al.: Smart Home Architecture for Smart Energy Consumption in a Residence With Multiple Users

FIGURE 7. Prototype of a tracking node.

FIGURE 6. Communication scheme of the smart outlet.

about the type of electrical equipment is conveyed to the

ATmega328 microcontroller.
The registration of NFC tags is done through the applica-
tion, having the ID and information of the appliances saved
in the database, so that when the SO is used, it is identi-
fied which appliances are consuming electricity. This also
receives the voltage and current data collected by the sensors
and processes this data based on a timeframe to calculate
energy consumption. Figure 6 shows the communication
scheme of the measurement node.
In the case of real-time tracking, the process of sending
FIGURE 8. Diagram depicting the web service layer (middleware)
information is similar to that of the measurement node and components.
this technology sends information to the middleware. The
nature of the information comprises where the user is and
In Figure 8, the interoperability between the measurement
the dBm (signal strength) of where she/he is located.
and tracking node is demonstrated. This interoperability will
As illustrated in Figure 7, the physical structure of the
be done through the database and the middleware in which
handover consists of an ESP8266 and an Arduino microcon-
the database will send the information received by the nodes
troller, which will be positioned in the center of each room in
so that the middleware can then process it and send it to the
the house in order to distribute the signal evenly to all ends.
application and web service.
As mentioned earlier in section 3, this hardware will aim to
This system includes features of the above-mentioned
promote the acquisition of the RSSI signal in decibels, as well
technologies, like the SO and handover, and uses REST as
as being also responsible for the user’s interconnection to
middleware.
the internet as a common Access Point. The user will leave
Together with residential metering equipment, the interface
a fingerprint, which is essential to accurately indicate the
of this middleware can connect applications for final users
location where the user is, thus capturing the signal strength
and utilities. Furthermore, highlighting numerous control and
information according to the distance.
management actions is made possible by using these middle-
ware features, such as, the necessity to access the measure-
B. SOFTWARE ment information recorded in a database by the central node
As a means of finding an SH solution for energy management, and set parameters for the transmission of this information to
this paper seeks to enable interoperability to occur between the management structure of the power utilities.
customer metering, real time tracking, and other management Nevertheless, depending on the type of user, some func-
applications. Hence, a system must be implemented with a tions will be restricted to ensure the safety of the residents,
layer based on the REST API that accepts interoperability the utilities are not permitted to turn off any electrical device
between these features. Overall, these applications should be which is controlled by the system, for example. This is exclu-
compatible with the Web Service technology chosen for this sively for the use of the residents, and users of the MAC
architecture. Address. Moreover, their location in the house is only visible

16816 VOLUME 9, 2021

S. H. M. S. Andrade et al.: Smart Home Architecture for Smart Energy Consumption in a Residence With Multiple Users

FIGURE 10. Above – Average Energy Consumption Indicator in Fuzzy

Logic.

FIGURE 9. An overall pyramid scheme for users.

to the residents. Instead, the energy supply of the utility can

be interrupted manually.
A procedure has been developed that is based on the fuzzy
logic system regarding applying a meta-heuristic for valida-
tion. The system may notify the customer of the necessity to
lower the energy consumption based on the input variables
(consumption indicator and rates) after which, the informa-
tion handling will be conducted in order to guarantee power
savings.
This implementation is necessary before the consump-
FIGURE 11. Monthly percentage input from a Fuzzy Logic perspective.
tion plan of each user can be defined since it relies on
the information collected from the measurement node and
the tracking node. Besides, the fuzzy logic must calculate the
consumption and produce results that can help each user,
regardless of the device and where he/she is located in the
residence.
The individual rate of consumption will be measured for
each individual. In Figure 9, the pyramids are formed utilizing
fuzzy logic and represent the most and least used appliance of
each user, the most often used being on the top and the least
used on the bottom.
The user pyramid schemes will serve as the basis for opti-
mizing the consumption defined in the profiles, as the ques-
tion of if each appliance can have its consumption reduced or
FIGURE 12. Exceeding consumption percentage.
not, depends on whether it is at the top or the bottom of this
pyramid. This optimization will be discussed later.
User pyramids will result in an overall pyramid represented The pyramids of the users will result in a general pyramid
in Figure 9 as ‘‘Combined Consumption’’ that also lists all the that defines the most widely used appliances in the house.
appliances used in the house and provides information about The information about the general customer that will be
which consume the most or least amount of energy. This made available to the energy utilities, will be taken from this
information from the overall pyramid will be made available pyramid, which is shown as a residence.
to the energy utilities, as it has information on the general It is also made clear that when two or more people share
consumption of the house. an appliance in the same room, the consumption will be
The pyramid data is extremely important to be able to divided between them for better comfort, distribution, and
rebalance the user’s consumption when it is necessary to consumption optimization.
use the profiles because a poor definition of the priority The Power Consumption Indicator (PCI) is calculated
of the appliances can directly impact the user’s comfort by based on the information provided by the automated housing
rebalancing inefficiently. system. This information includes: Use up to the current day

VOLUME 9, 2021 16817

 S. H. M. S. Andrade et al.: Smart Home Architecture for Smart Energy Consumption in a Residence With Multiple Users

FIGURE 13. Decision-making system flowchart [7].

(CC), current day of the month (MD), historical consumption Also, based on the excessive consumption shown by the
average (HCA), and the number of days in the current month unclear system, the factors which result in increased energy
(CMD). Based on this data, the indicator will be calculated usage may be included from the information observed for
by (5): every device. This means that the users can decide based
on the defined strategies of the profile, which elements
PCI = CC / (MD (HCA / CMD)) (5) to disable, or how to reduce their use. Figure 13 illus-
From the ‘‘Indicator of Comfort’’ formula (specified in the trates a flowchart used for explaining the stages of the
background section of this paper), it is possible to acquire the decision-making system.
values for PCI. They are divided by scales, where PCI = 1
suggests that the consumption is inside the typical range. It is VI. DEMONSTRATION AND RESULTS
shown in Figure 10 that when PCI > 1, it is separated into The operating process of the new architecture is demonstrated
scales ranging from 20% to 100%. in the flowchart format in Figure 14, being possible to verify
Similarly, in Figure 11, the input variable ‘‘Monthly the necessary steps for the correct functioning of the system
Percentage’’ is divided into 5 pre-established periods. through the APs, handover system and SO, to provide smart
The output variable "Excess consumption" may undertake energy management for the user.
six potential states from the starting point of the graph after The interoperability between the handover and the SO will
the Fuzzy machine has been completed, meaning that the identify which room the user is in and what is being consumed
consumption falls inside the average array. If this is within by him/her, determining the identification and the precise
a range from 20% to 100%, it indicates that the amount location so that their consumption is optimized in the best
consumed has been surpassed, as it is shown in Figure 12. way.
Additionally, when the Fuzzy inference system sends the In Figure 15, the system illustrates the user identification
final result, the occupant will then be instructed by a notifi- process and data acquisition from the system SO. This data
cation on her/his mobile application. The alert will show all will be acquired and sent to a server to be processed and
the devices that have exceeded the expected consumption rate later sent to the middleware, showing the information already
and the user can follow steps to provide an optimized usage processed in the application and the Web Service.
plan. Additionally, with the interoperability information, two
This system output completely depends on the "Index" and systems have been designed for the architecture in different
"% per month" input variables. The output is also changed domains, one being a system for mobile devices while the
when the input values change. other is a system for WEB applications. The objective of
The configuration management system can make decisions both systems was to show the viability and justification of
with greater precision based on data, to ensure that the users the planned architecture.
can be informed of the current rate of electricity usage by The two designed system prototypes require monitor-
checking the consumption of electronic devices. ing and administration utility that are consistent with the

16818 VOLUME 9, 2021

S. H. M. S. Andrade et al.: Smart Home Architecture for Smart Energy Consumption in a Residence With Multiple Users

FIGURE 14. Diagram of the proposed new architecture.

FIGURE 15. Tracking node and measurement node interoperability.

ones of the proposed architecture. These are not limited to FIGURE 16. The monitoring system – WEB case study.
the users’ access platform or the established types and the
access limitations are established in the validation scheme.
Figure 16 shows consumption in real time, giving the user the The second screen contains the application’s general
possibility to view the energy consumption of each device in options as well as some information about the user’s current
their environment via the web interface. location, current dBm, and MAC Address. This information
Figure 17 displays the two screens of the mobile service, is provided by the user tracking system. The mapping will
which include the initial screen and the ‘main menu’ screen be done to establish the signal strength inside each room and
with the ‘general options’. the software will determine the room the user is currently
The first screen requires user registration or log in to access in, based on the signal identification and taking into account
the application and web services. The user registration will the limitations of each room defined by the mapping done
not only serve to register the user in the system but also in case previously.
there is a problem that causes the user to change smartphone The section for registering an NFC tag information is
(for example technical problems or the loss of the device). available within the user and room settings menu, and
Therefore, the user can log in with his registered account the information will be sent to the cloud database after
normally on another smartphone and can continue using it. registration.

VOLUME 9, 2021 16819

 S. H. M. S. Andrade et al.: Smart Home Architecture for Smart Energy Consumption in a Residence With Multiple Users

FIGURE 19. Fan voltage and current obtained through the smart outlet.

FIGURE 20. Database containing the information obtained from tests

using the fan.
FIGURE 17. Mobile application login screen and menu screen.

FIGURE 21. Graph to visualize Consumption (kWh).

lation is imperceptible in the graph. The data in Figure 19 are

FIGURE 18. Smart outlet test using a fan.
shown in Figure 20, which are stored in a database with the
time, date, and power recorded from the fan test.
Other results can be obtained by manipulating the collected Figure 21 shows the consumption data of the appliances,
set, in addition, to be able to view the information through the which are extracted from the measurement node, using the
designed solutions, starting with the data locally stored (but NFC to identify which appliance was used and its consump-
only the month of measurement in the central unit) or data tion extracted by the electric current and voltage sensors.
stored externally. In order to represent the potential for energy consump-
Tests were carried out to check the accuracy of the mea- tion optimization, surveys and measurements have been con-
surement node. As illustrated in Figure 18, a fan is connected ducted to test the SG, involving the collection of entered data
to the SO. necessary for the authentication of the results.
As shown in Figures 19 and 20, it is possible to verify Based on the information gathered and displayed
the voltage and current values information taken from the in Figure 22, a smart approach has been employed to assess
database in real time, as well as the appliance that is currently the possible results for energy use optimization.
being used. In Figure 19, the current graph is represented by After profile 1 was successfully applied, it is noticeable a
a continuous line because in the scale presented, its variation decrease in the energy consumption of each appliance in pro-
is always within the 0.6 ampere range, thus the current oscil- portion to its overall consumption, until the excessive amount

16820 VOLUME 9, 2021

S. H. M. S. Andrade et al.: Smart Home Architecture for Smart Energy Consumption in a Residence With Multiple Users

FIGURE 22. Users energy consumption average rate.

FIGURE 24. Statistics and energy consumption data screens.

FIGURE 23. The balance of appliances energy usage based on overall

consumption.

of power consumption was reduced based on a 171 KWh

limit.
In Figure 23 this can be observed in the fan for example,
which had its limit reduced from 28 KWh to 18,1 KWh. Pre-
viously, the shower’s consumption amount limit was 15 KWh
but following the implementation of regulatory measures for
consumption, this amount has been reduced to 9.7 KWh, and
the rates of every other appliance have also been reduced.
Even though the consumption of the refrigerator is being
shown in the image, it will not get reduced as it is essential for
the user in their daily lives. Therefore, it is necessary to con-
tinue its constant use since the user keeps this equipment on
every day of the week. As a result, its operation regime is not
analyzed, that is, the next day it will continue with the same
FIGURE 25. Identification of multiple users in the same environment.
consumption even if there is excessive use of electricity and
the user utilizes the profile to perform the energy rebalancing.
The refrigerator, being the most essential appliance for the It can also be seen in the image to the left of Figure 24
user, in this case, will continue with its regular consumption when consumption is exceeded its pre-established
regardless of the scenario. value (171KWh). In the image on the right, it can be seen how
This is the objective of Profile 1, which makes a propor- energy consumption can be reduced since its pre-established
tional reduction of each one and manages to balance the consumption for the day is above the standards and needs to
consumption rates of all the appliances to ensure the end user be compensated the next day.
is provided with a suitable degree of comfort. It is demonstrated in Figure 25 that when two or more users
The system’s main goal is to establish the viability of the are in the same room, both are identified and located by the
designed architecture, as well as encourage its endorsement. handover system. Using the RSSI fingerprint, it verifies that
Figure 24 illustrates the monthly consumption, based on the the two are inserted within the pre-established dBm limits for
data collected through the use of SOs, as well as the results for the living room, with user 1 being identified at −25 dBm from
the estimated energy consumption after Profile 1 was applied. AP2 and user 2 being identified at −19 dBm from AP2.

VOLUME 9, 2021 16821

 S. H. M. S. Andrade et al.: Smart Home Architecture for Smart Energy Consumption in a Residence With Multiple Users

FIGURE 26. Sum of the two user’s appliance limit consumptions.

FIGURE 28. Remaining consumption hours for each user after two hours.

User 1 and user 2 used the same video game for two
hours simultaneously, resulting in a reduction of 2 hours from
8 hours, which would be the sum of the limits of the two users.
Therefore, after these 2 hours, it is identified through RSSI
that user 1 is no longer within the mapped limits of the living
room (AP2) as shown in Figure 27, occurring the proportional
division of the remaining 6 hours.
This division will take into account the initial contribution
of each user, so from these 6 hours user 1 will be entitled to
25% (1 hour and 30 minutes) and user 2 will be entitled to
75% (4 hours and 30 minutes) of the remaining total.

VII. CONCLUSION
This project has the purpose to develop an innovative solu-
FIGURE 27. System identifying the movement of a user to another room.
tion for an SH environment regarding the concepts of
decision-making techniques, IoT, device design, measure-
Figure 26 shows the estimated time to reach the limit for ments, interoperability, and easy -to -use applications.
consumption of television, video games, and air conditioning This architecture implementation has led to the following:
for the two users who are separated. It is noticed that user a hardware solution with the objective to identify the user
1 has less time to use the television and the video game than location and what appliance is she/he using and how much is
user 2, so user 1 joins user 2 who is present in the living she/he consuming, with an interoperable middleware solution
room. This results in the possibility that users can use these which provides solutions to help the consumer to optimize
appliances together, thus reducing energy consumption and her/his consumption. The architecture also includes several
increasing each user’s usage time. technological requirements, for example a high degree of
This strategy can be used so that instead of the user hav- flexibility and reuse, service transparency, availability of
ing to stop using the appliance, his/her consumption can be information, and modularity.
reduced by dividing proportionately with more people who In this paper, the focus of key aspects of the SH domain is
are also using the same appliances in the same room. the designed architecture. Its specific features and functions
It is demonstrated in Figure 27 that after 2 hours in the are not only suitable for monitoring and control contexts
same room, it is identified through the RSSI fingerprint that but also offer an interface for control through monitoring
user 1 is no longer in the living room, as his / her dBm value systems based on traditional solutions derived from the tool.
is −176 dBm in relation to the AP2 and is no longer within Nevertheless, this alone does not mean that the end-user
the pre- established limits to that room. The system then cannot find the solutions needed to control and monitor the
identifies that user 1 is now located in Bedroom 1, as his dBm home.
value for the AP1 is −22 dBm, being within the defined limit. When applying the architecture, it should be stressed that
User 2 is still within the limits defined for the living room, not the tracking of the consumption habits of different users
occurring the handover process. connected and actively monitor this is made possible using
Figure 28 shows the result of the proportional division a mobile or web application. Moreover, it is noteworthy that
between users when they are in the same room. To explain this the data generated by the architecture will allow for effective
division using the videogame as an example, user 1 and user mechanisms to be designed which constitute an effective
2 have their defined limit hours summed up, with the total of method of applying government policies that can assist in
this sum being 8 hours. User 1 contributes 25% (2 hours) of reducing energy consumption. Using efficient computational
this sum, while user 2 contributes 75% (6 hours). techniques to produce statistical data that can help reduce

16822 VOLUME 9, 2021

S. H. M. S. Andrade et al.: Smart Home Architecture for Smart Energy Consumption in a Residence With Multiple Users

electricity waste, the system can also provide benefits to the [10] J. Iqbal, M. Khan, M. Talha, H. Farman, B. Jan, A. Muhammad, and
environment. H. A. Khattak, ‘‘A generic Internet of Things architecture for control-
ling electrical energy consumption in smart homes,’’ Sustain. Cities Soc.,
In addition, several benefits are provided for an SG sit- vol. 43, pp. 443–450, Nov. 2018, doi: 10.1016/j.scs.2018.09.020.
uation, in particular for SHs, through the adoption of this [11] T. AlSkaif, I. Lampropoulos, M. van den Broek, and W. van Sark,
architecture. The benefits are as follows: ‘‘Gamification-based framework for engagement of residential customers
(a) An intuitive user interface for mobile applications to in energy applications,’’ Energy Res. Social Sci., vol. 44, pp. 187–195,
Oct. 2018, doi: 10.1016/j.erss.2018.04.043.
control and monitor the devices and rate of consump- [12] M. K. Annaqeeb, R. Markovic, V. Novakovic, and E. Azar, ‘‘Non-intrusive
tion of each appliance of the end-user. data monitoring and analysis of occupant energy-use behaviors in shared
(b) An ability to predict the final amount of energy office spaces,’’ IEEE Access, vol. 8, pp. 141246–141257, 2020, doi:
10.1109/ACCESS.2020.3012905.
consumption by using an smart module.
[13] S. Wan, P. Xu, K. Wang, and S. Li, ‘‘Estimation of distributed thermal
(c) A Design and control consumption system for both the boundary based on fuzzy clustering of temperature observable points,’’
customer and the power companies. Int. J. Heat Mass Transf., vol. 147, Feb. 2020, Art. no. 118920, doi:
10.1016/j.ijheatmasstransfer.2019.118920.
(d) A tracking system for residential homes with multiple
[14] K. T. Prajwal and V. S. N. V. S. Gupta, ‘‘Smart home energy management
residents to improve the management of electricity con- system using fuzzy logic for continuous power supply with economic util-
sumption. isation of electrical energy,’’ in Proc. 2nd Int. Conf. Inventive Syst. Control
(e) The development of a low-cost open standard solution (ICISC), Jan. 2018, pp. 274–279, doi: 10.1109/ICISC.2018.8399078.
[15] A. Patel and T. A. Champaneria, ‘‘Fuzzy logic based algorithm for context
for transmission and metering devices. awareness in IoT for smart home environment,’’ in Proc. IEEE Region
Every software component and device used in this project 10 Conf. (TENCON), Nov. 2016, pp. 1057–1060, doi: 10.1109/TEN-
is interoperable and scalable and they have all been open CON.2016.7848168.
standards-based, a key element being their low price. Such [16] K. Khan, T. Hossen, A. Savasci, L. Gauchia, and S. Paudyal, ‘‘Design of
a simplified hierarchical Bayesian network for residential energy storage
factors can allow the proposed architecture to be used in com- degradation,’’ in Proc. IEEE Power Energy Soc. Gen. Meeting (PESGM),
mercial areas on a large scale. Additionally, the architecture Aug. 2019, pp. 1–5, doi: 10.1109/PESGM40551.2019.8973603.
has already been successfully employed in real- world situ- [17] M. Shahjalal, M. K. Hasan, M. M. Islam, M. M. Alam, M. F. Ahmed, and
Y. M. Jang, ‘‘An overview of AI-enabled remote Smart- home moni-
ations and is thus suitable for use and managed to rebalance toring system using LoRa,’’ in Proc. Int. Conf. Artif. Intell. Inf. Com-
residential energy consumption 87.3% of the time it was used. mun. (ICAIIC), Feb. 2020, pp. 510–513, doi: 10.1109/ICAIIC48513.
2020.9065199.
REFERENCES [18] V. Govindraj, M. Sathiyanarayanan, and B. Abubakar, ‘‘Customary homes
[1] P. Kumar, Y. Lin, G. Bai, A. Paverd, J. S. Dong, and A. Martin, ‘‘Smart to smart homes using Internet of Things (IoT) and mobile applica-
grid metering networks: A survey on security, privacy and open research tion,’’ in Proc. Int. Conf. Smart Technol. For Smart Nation (Smart-
issues,’’ IEEE Commun. Surveys Tuts., vol. 21, no. 3, pp. 2886–2927, TechCon), Aug. 2017, pp. 1059–1063, doi: 10.1109/SmartTechCon.2017.
3rd Quart., 2019, doi: 10.1109/COMST.2019.2899354. 8358532.
[2] M. H. Cintuglu, O. A. Mohammed, K. Akkaya, and A. S. Uluagac, ‘‘A sur- [19] S. Somani, P. Solunke, S. Oke, P. Medhi, and P. P. Laturkar, ‘‘IoT
vey on smart grid cyber-physical system testbeds,’’ IEEE Commun. Surveys based smart security and home automation,’’ in Proc. 4th Int. Conf.
Tuts., vol. 19, no. 1, pp. 446–464, 1st Quart., 2017, doi: 10.1109/COMST. Comput. Commun. Control Autom. (ICCUBEA), Aug. 2018, pp. 1–4, doi:
2016.2627399. 10.1109/ICCUBEA.2018.8697610.
[3] A. Annaswamy, IEEE Vision for Smart Grid Control: 2030 and Beyond
[20] R. A. Rashid, L. Chin, M. A. Sarijari, R. Sudirman, and T. Ide, ‘‘Machine
Roadmap, IEEE Standard, Oct. 2013, vol. 1, no. 1, pp. 1–12, doi:
learning for smart energy monitoring of home appliances using IoT,’’
10.1109/IEEESTD.2013.6648362.
in Proc. 11th Int. Conf. Ubiquitous Future Netw. (ICUFN), Jul. 2019,
[4] M. Al-Kuwari, A. Ramadan, Y. Ismael, L. Al-Sughair, A. Gastli, and
pp. 66–71, doi: 10.1109/ICUFN.2019.8806026.
M. Benammar, ‘‘Smart-home automation using IoT-based sensing and
[21] A. Javed, H. Larijani, and A. Wixted, ‘‘Improving energy consump-
monitoring platform,’’ in Proc. IEEE 12th Int. Conf. Compat., Power
tion of a commercial building with IoT and machine learning,’’ IT
Electron. Power Eng. (CPE-POWERENG), Apr. 2018, pp. 1–6, doi:
Prof., vol. 20, no. 5, pp. 30–38, Sep. 2018, doi: 10.1109/MITP.2018.053
10.1109/CPE.2018.8372548.
891335.
[5] D. Mocrii, Y. Chen, and P. Musilek, ‘‘IoT-based smart homes:
A review of system architecture, software, communications, privacy [22] A. Haidine, A. Aqqal, and A. Dahbi, ‘‘Performance evaluation of low-
and security,’’ Internet Things, vols. 1–2, pp. 81–98, Sep. 2018, doi: power wide area based on LoRa technology for smart metering,’’ in Proc.
10.1016/j.iot.2018.08.009. 6th Int. Conf. Wireless Netw. Mobile Commun. (WINCOM), Oct. 2018,
[6] S. H. Andrade, E. da Luz Oliveira, R. Alfaia, A. Souto, C. R. L. Frances, pp. 1–6, doi: 10.1109/WINCOM.2018.8629693.
and N. L. Vijaykumar, ‘‘Adopting fuzzy technique to save energy in smart [23] N. Prayongpun and V. Sittakul, ‘‘Smart Plug for Household Appliances,’’
home control systems,’’ in Proc. IEEE 58th Int. Sci. Conf. Power Electr. in Proc. Adv. Intell. Syst. Comput., 2017, pp. 317–325.
Eng. Riga Tech. Univ. (RTUCON), Oct. 2017, pp. 1–6, doi: 10.1109/RTU- [24] M. A. U. Alam, N. Roy, and A. Misra, ‘‘Tracking and behav-
CON.2017.8124795. ior augmented activity recognition for multiple inhabitants,’’ IEEE
[7] E. D. L. Oliveira, R. D. Alfaia, A. V. F. Souto, M. S. Silva, C. R. L. Francás, Trans. Mobile Comput., vol. 20, no. 1, pp. 247–262, Jan. 2021, doi:
and N. L. Vijaykumar, ‘‘SmartCoM: Smart consumption management 10.1109/TMC.2019.2936382.
architecture for providing a user- friendly smart home based on meter- [25] K. Murao, T. Terada, A. Yano, and R. Matsukura, ‘‘Detecting
ing and computational intelligence,’’ J. Microw., Optoelectron. Electro- room-to-room movement by passive infrared sensors in home
magn. Appl., vol. 16, no. 3, pp. 736–755, Sep. 2017, doi: 10.1590/2179- environments,’’ in Proc. 1st Workshop recent Adv. Behav. Predict.,
.10742017v16i3965. Apr. 2012, pp. 1–12. [Online]. Available: https://www.ibr.cs.tu-bs.de/dus-
[8] Y. Li, Z. Sun, L. Han, and N. Mei, ‘‘Fuzzy comprehensive eval- beigl/Awarecast/awarecast2012_submission_5.pdf
uation method for energy management systems based on an Inter- [26] D. Dintsis and V. Simankov, ‘‘The control automation system method
net of Things,’’ IEEE Access, vol. 5, pp. 21312–21322, 2017, doi: based on combined implementation of fuzzy logic and discrete automa-
10.1109/ACCESS.2017.2728081. tion systems,’’ in Proc. Eur. Model. Symp., Nov. 2013, pp. 87–90, doi:
[9] M. Botticelli, L. Ciabattoni, F. Ferracuti, A. Monteriu, S. Pizzuti, and 10.1109/EMS.2013.15.
S. Romano, ‘‘A smart home services demonstration: Monitoring, con- [27] L. Wu, Y. Wang, and H. Liu, ‘‘Occupancy detection and localization by
trol and security services offered to the user,’’ in Proc. IEEE 8th monitoring nonlinear energy flow of a shuttered passive infrared sen-
Int. Conf. Consum. Electron., Sep. 2018, pp. 1–4, doi: 10.1109/ICCE- sor,’’ IEEE Sensors J., vol. 18, no. 21, pp. 8656–8666, Nov. 2018, doi:
Berlin.2018.8576259. 10.1109/JSEN.2018.2869555.

VOLUME 9, 2021 16823

 S. H. M. S. Andrade et al.: Smart Home Architecture for Smart Energy Consumption in a Residence With Multiple Users

[28] V. Bianchi, P. Ciampolini, and I. De Munari, ‘‘RSSI-based indoor local- LUIGUY X. LIMA was born in Belém, Pará, Brazil.
ization and identification for ZigBee wireless sensor networks in smart He received the bachelors’ degree in electrical
homes,’’ IEEE Trans. Instrum. Meas., vol. 68, no. 2, pp. 566–575, engineering from Estácio Nazaré University, in
Feb. 2019, doi: 10.1109/TIM.2018.2851675. 2018.
[29] S. H. Shah and I. Yaqoob, ‘‘A survey: Internet of Things (IOT) technolo-
gies, applications and challenges,’’ in Proc. IEEE Smart Energy Grid Eng.
(SEGE), Aug. 2016, pp. 381–385, doi: 10.1109/SEGE.2016.7589556.
[30] S. Verma, Y. Kawamoto, Z. M. Fadlullah, H. Nishiyama, and N. Kato,
‘‘A survey on network methodologies for real-time analytics of
massive IoT data and open research issues,’’ IEEE Commun. Sur-
veys Tuts., vol. 19, no. 3, pp. 1457–1477, 3rd Quart., 2017, doi:
10.1109/COMST.2017.2694469. NANDAMUDI L. VIJAYKUMAR received the
[31] T.-Y. Kim, S.-H. Bae, and Y.-E. An, ‘‘Design of smart home imple- degree in computer technology from the Techno-
mentation within IoT natural language interface,’’ IEEE Access, vol. 8, logical Institute of Aeronautics, in 1978, the mas-
pp. 84929–84949, 2020, doi: 10.1109/ACCESS.2020.2992512. ter’s degree in applied computing from the
National Institute for Space Research, in 1984, and
the Ph.D. degree in electronic and computer engi-
neering (area: informatics) from the Technological
SÉRGIO H. M. S. ANDRADE was born in Belém, Institute of Aeronautics, in 1999. He participated
Pará, Brazil, in 1989. He received the degree in in the Postdoctoral program at University College
control and automation engineering from the Insti- Cork (UCC), Coastal and Marine Research Center
tute of Superior Studies of Amazon, in 2011, and (CMRC), at University College Cork (UCC), Cork, Ireland, in the area of
the master’s degree in electrical engineering, with Coastal Computational Modeling. The Postdoctoral program was funded by
an emphasis in applied computing from the Fed- the European Union. He retired in May 2017, fearing that the Pension Reform
eral University of Pará, in 2017. He is currently would harm the benefits acquired. He currently maintains a link as a Volun-
pursuing the Ph.D. degree in electrical engineer- teer Research Collaborator with the Associated Laboratory of Computation
ing, with an emphasis in applied computing. He is and Applied Mathematics (LAC), National Institute for Space Research
also a Professor of Technical Schools, Colleges, (INPE). He has experience in the field of Computer Science, with an empha-
Universities, and Graduate Studies. He is currently a Professor with the Fac- sis on Performance Evaluation, Software Testing with Formal Specification,
ulty of Automation Engineering and Computer Engineering, Estácio Nazaré Time Series Analysis, and Computational Modeling of Coastal Environment.
University. He is a Coordinator of Control and Automation Engineering and Since July 2019, he has been serving as a Visiting Associate Professor
Telecommunications. He is a Commentator on CBN Amazônia radio about with the Institute of Science and Technology, UNIFESP, São José dos
Automation and Robotics. He is also a Partner-Owner of Salt Engenharia e Campos.
Automação LTDA. He is a member of the Editorial Board of the Journal of
Autonomous Intelligence. He is also a Reviewer of the Journal Solar Energy. CARLOS RENATO L. FRANCÊS received the
bachelor’s degree in computer science from the
Federal University of Pará, in 1995, the master’s
degree in science from the Department of Comput-
ing and Computational Mathematics, University
of São Paulo, in 1998, and the Ph.D. degree in
GUSTAVO O. CONTENTE was born in Belém, computer science and computational mathematics
Pará, Brazil, in 1997. He is currently pursuing from the University of São Paulo, in 2001. He has
the degree in computer engineering course with a 1-D Research Productivity scholarship. He did a
Estácio Nazaré University. Postdoctoral Internship at the Institute of Systems
and Computer Engineering, Technology and Science (INESC-TEC), Porto
/ Portugal, in 2012. He is currently a Full Professor with the Faculty of
Computer Engineering, Federal University of Pará, where he was the Presi-
dent of the State of Pará Data Processing Company, from January 2007 and
December 2010 and, until 2011, he was a member of the Computing Advi-
sory Committee the National Institute of Educational Studies and Research
(INEP-MEC). He was the Dean of Graduate Studies, Research, and Tech-
nological Innovation, Federal University of the South, and the Southeast of
Pará (Unifesspa), from 2013 to 2016. He was Pro-Tempore Rector of the
Federal University of the South and the Southeast of Pará (Unifesspa), from
LUCAS B. RODRIGUES was born in João Pessoa, June 2016 to October 2016. He has experience in the areas of Telecom-
Paraíba, Brazil, in 1998. He is currently pursuing munications Systems and Computer Science: - In Telecommunications Sys-
the degree in computer engineering course with tems: Planning, optimization and performance evaluation of Smart Cities,
Estácio Nazaré University. wireless networks, heterogeneous networks - HetNet, 4G / 5G, PowerLine
Communication-PLC, optical clouds, among others. - In Computer Sci-
ence: Computational Intelligence and extraction techniques, correlation and
pattern recognition in social, economic, and environmental systems, such
as: energy, public health, social security, public security, deforestation, and
deforestation and economics.

16824 VOLUME 9, 2021