2017 IEEE 13th International Conference on Wireless and Mobile Computing, Networking and Communications (WiMob)

Providing a Cloud-based Smart Meter Solution to

Control and Monitor Electrical Quantities of
Industrial Machines

Marcelo Dornbusch Lopes∗ , Leonardo R. P. Rauta§ , Benjamin W. Mezger∗ , Michelle S. Wangham∗

∗ Laboratory of Embedded and Distributed Systems and 4Vision Lab
University of Vale do Itajaí, Itajaí - SC, Brazil
§ Federal Institute of Santa Catarina (IFSC) - Campus Gaspar, Gaspar - SC, Brazil

Email: [email protected], [email protected], [email protected], [email protected]

Abstract—The remote monitoring and control of machines are Most of the IoT applications need to process and deliver
essential in industrial environments. Emerging communication a great amount of data coming from several heterogeneous
technologies such as Web of Things and Machine to Machine devices. The concept of Cloud of Things (CoT), which is the
Communication can meet this demand for automation in Industry combination of Cloud Computing with IoT and can meet these
4.0. This paper aims to introduce a solution for continuous and demands. The CoT uses the virtualization ability in the IoT
remote monitoring of electrical quantities in smart industrial
context and exposes sensors and actuators (things) features
environments with three-phase systems. The proposed solution
uses a service-oriented architecture and makes use of a Smart available through services. In the Cloud, it is possible to run
Gateway, RESTful web services, and cloud computing. The heavy computational tasks, while devices can focus on sensing
solution was integrated with a real case study and evaluated by and activity[7].
software testing. The results obtained demonstrate the feasibility
of the solution, and the accuracy of measurements persisted in
The heterogeneous of industrial machines create challenges
the cloud. for M2M communications and CoT. For example, it is common
in the apparel industry to have different machines, from dif-
Keywords—Remote Control and Monitoring Electrical Quanti- ferent manufacturers, which are using different communication
ties, M2M, Web of Things, Cloud Computing standards. To guarantee interoperability, there is an increasing
trend to treat IoT as Web of Things (WoT). WoT uses Web
standards and protocols to provide information sharing [8], [9].
I. I NTRODUCTION
The University of Vale to Itajaí, along with ABC Enter-
Industry 4.0 is no longer a ‘future trend’ for many prise1 , an apparel machinery manufacturer, has established
companies. Companies are combining advanced connectiv- a partnership with the objective to develop smart industrial
ity and advanced automation, cloud computing, sensors, environments in research and development (R&D) projects.
computer-powered processes, machine to machine communica-
tion (M2M), intelligent algorithms and Internet of things (IoT) The continuous and remote monitoring of electrical quan-
services to transform their businesses [1]. These paradigms tities in industrial environments and the machine energy con-
allow the interaction of devices (machines) in an automated sumption measurement are some of the main needs reported
manner, with or without minimum human intervention [2], [3]. by ABC Enterprise. The malfunction of the electrical grid can
cause machine failure, which in turn, can delay production and
The remote control and monitoring of accurate information even damage to the machine. Today, ABC Enterprise provides
on machine operation are necessary to efficiently manage technical assistance to its clients whenever there is a problem
industrial environments, as well as to increase the production reported to the Enterprise. ABC Enterprise sends a technician
performance. to measure the electrical quantities and produce a technical
report. This solution does not provide continuous monitoring,
The concept of M2M communication consists of a group energy consumption and is not cost effective. The energy
of information communication technology that is able to consumption is a critical factor to analyze production costs
measure, deliver, synthesize information and react to events since it allows the apparel factory to calculate real production
autonomously. Examples of such concept are telemetry read- costs of its machine operation. For example, the cost of a
ings of status of oil and brakes of cars on the move, health specific piece of garment.
state measurements of blood pressure and heartbeat of the
elderly, occupancy measurements of parking in cities, remote This article aims to describe a solution developed by
metering of water consumption, and real-time monitoring of partner institutions to remotely and continuously monitor the
critical parts of a piece of machinery [4], [5]. electrical quantities used by industrial machines and that
enables Machine to Machine (M2M) communication. This
The IoT concept provides everyday objects the ability to solution is an industrial appliance named Smart Meter and
self-identify, sense, process and connect with other objects and is based on an architecture presented in [10], which builds
Internet services. Additionally, IoT devices can be ubiquitous,
context sensitive and able to create smart environments [6]. 1 This Enterprise wished to remain anonymous.

978-1-5386-3839-2/17/$31.00 ©2017 IEEE

 2017 IEEE 13th International Conference on Wireless and Mobile Computing, Networking and Communications (WiMob)

a smart industrial environment, and that can be used before In [14], a cloud-based architecture is presented to control
and after the installation of the industrial machine. Our Smart and disseminate the collected data in a Smart Grid through a
Meter solution uses Smart Gateways, Cloud Computing, and SCADA (Supervisory Control and Data Acquisition) system
Web of Things technologies. As a proof of concept, a prototype based on LabVIEW. To monitor the Smart Grids, the solution
was developed, integrated into a real case study and evaluated. uses PLCs (Programmable Logic Controller) that allows the
communication via Ethernet. The SCADA system is also used
The remainder of this paper is structured as follows. as a Gateway to collect PLC data and send them to a data
Section II presents and compares some related works. Section persistence service in the cloud, which makes them available
III describes the provisioning architecture smart industrial through a Web service. To communicate the LabVIEW ap-
environments in the apparel industry. The focus of Section IV plication (SCADA) with the PLC are necessary a personal
is the description of the proposed industrial appliance, Smart computer (PC).
Meter. The use of the Smart Meter in the case study (ABC
Enterprise) is described in Section V. Moreover, this section In [15], the authors describe a middleware for industrial
presents results and discusses the applicability, data accuracy networks to control and monitor industrial machines using
and performance of the considered solution. Finally, Section object virtualization and data fusion. This middleware pro-
VI outlines some conclusions and future works. vides communication between the industrial machine with the
SCADA system and the gateway to the Cloud. The gateway
II. R ELATED W ORK creates a virtual representation of the machine and send it to
the layer above. This layer is responsible for managing these
This section analyses five related works that deals with virtual representations and creating applications to control and
the Web-based remote monitoring of industrial environments monitor the machines. It is also possible to develop business
considering the following characteristics: (i) if they make use intelligence applications in this layer, which can extract the
of RESTful Web Services; (ii) if they make use of cloud knowledge about machine operations. This solution supports
persistence services; (iii) if they make use of Smart Gateways; M2M communication using a multiprotocol gateway that can
(iv) if the Smart Gateway (SG) is embedded; (v) if M2M use wired (Wired Industrial Automation Devices - WIAD, e.g.
communication is used; and, (vi) if it provides monitoring or PROFIBUS, WorldFIP and CAN) or wireless (Wireless Sensor
control of the machines as well. and Actuator Networks - WSANs).
In [11], a monitoring infrastructure of the electrical quan- The Table I compares the correlated works with the one
tities was proposed. In this infrastructure, sensors attached presented in this paper (Smart Meter). [14] resembles the one
to the grid are used, and the data collected by the sensors presented here, once it monitors the Smart Grids, disseminates
are sent to a hub through an SPI (Serial Peripheral Interface) the collected data and uses a persistent cloud platform. How-
communication. This hub stores the collected data and makes ever, this work uses a costly solution.
the data available through an embedded Web server. The data is
not persisted in a cloud and the solution does not make use of
RESTful Web services. As a future work, the authors suggest TABLE I: Comparison between related works and proposed
the implementation of the IEC61850 protocol, in which it is solution
generally used for control, protection, alarm, and monitoring of REST Cloud Smart Embedded Monitor (M) /
energy substations and it will allow integrations with actuators. API usage Gateway SG Control (C)
[11] - - Yes Yes M/C
The work described in [12] proposes the development of a [12] - - Yes Yes M
residential energy metering system in real-time. In this work, [13] - - Yes Yes M/C
[14] Yes Yes Yes - M/C
energy meters are used by different devices, connected to a [15] - Yes Yes Yes M/C
central hub, which makes the collected data available through This work Yes Yes Yes Yes M/C
an embedded Web service. The communication between the
devices and the central hub uses the CAN (Controller Area [11], [12], [13] are similar works to this one due to the
Network) protocol. The work does not store the collected data use of Smart Gateways to monitor electrical quantities and
in the cloud, nor describes if there is any persistence service to disseminate the collected data through a web environment,
within the hub. Furthermore, the work does not foresee the use although they do not make use of the Cloud or RESTful
of actuators, only of low-cost meters. services.
In [13], the authors describe an architecture that uses Gate- [15] proposes a middleware to monitor and control indus-
ways to gather information through the optic sensors available trial machines using Service Oriented Architecture (SOA) to
for the gas and energy supply network. The work focuses exposes objects capabilities using web services, thus support-
on monitoring these supply networks. The optic sensors are ing syntactic and semantic interoperability among different
attached to the valves responsible for such distribution. The technologies, including SCADA systems. This solution uses
information gathered by these sensors are sent to the Gateway a web environment to manage an industrial environment.
via the Modbus communication protocol through a wireless However, it does not describe the use of RESTful services
connection. The work uses a simple web server and does not either the use of the Cloud.
use RESTful Web services, it does not persist data in the cloud
and the authors do not comment whether they are stored in the III. S MART I NDUSTRIAL E NVIRONMENT
Gateway. Moreover, the work does not predict that this solution
is able to control the gas valves, only the monitoring of these Many industrial machines do not have connectivity with
valves. the Internet through an IP protocol (Internet Protocol) due to
 2017 IEEE 13th International Conference on Wireless and Mobile Computing, Networking and Communications (WiMob)

Fig. 1: Overview of a Smart Industrial Environment

security restrictions, geographic factors or costs. In such cases, • Web RESTful services: provide the resources of the
it is possible to use a Smart Gateway as an intermediate device machines (electrical quantities) for the clients (other
to create these connections between the machines and the machines or applications), which allows the monitor-
Internet. These devices use an integration method of software ing (through HTTP GET) and/or controlling (through
and hardware capable of interpreting the messages originated HTTP POST or PUT) of these resources;
by the machines and directing them to the target destination,
whether it is a database, a server or another device. • Web RESTful Client: sends, periodically, the mon-
itored data through Smart Gateway to the Cloud
In our previous work [10], we present a solution based on persistence service.
a Service-Oriented Architecture (SOA) and one which makes
use of Smart Gateways and cloud computing to provide smart The JSON2 was used as the format in the exchanging of
industrial environment. This solution provides interoperability messages between smart gateways and the cloud. JSON is
between heterogeneous devices, data persistence and process- most suitable for using with embedded systems since it is
ing in the cloud, and remote monitoring and non-real-time less verbose and it uses less computational resources when
control through the Web. Next, the main components of the compared with the XML3 standard [16].
proposed solution in [10] are adopted in this current work, is According [10], the cloud is used for: (i) providing a REST-
described as follows. ful Web service for receiving the data from the Smart Meter;
(ii) persisting these data in a NoSQL (DocumentDB) database;
A. Smart Gateway or Smart Meter (iii) providing a Platform as a Service (PaaS), creating control
and metering Web applications; and (iv) provisioning of these
Smart Gateways are responsible for: (i) abstracting the Web applications. The PaaS enables companies to develop
proprietary communication protocols (or native APIs) of ma- customized web applications to remotely control and monitor
chines and making their features (resources) available via a industrial machines; the data and the application will be hosted
RESTful API for M2M communication; (ii) sending monitored on the Cloud.
data to a Cloud persistent Web Service; and (iii) allowing for
the remote monitoring and non-real-time control of industrial
machines for web applications hosted on the Cloud or mobile IV. S MART M ETER U SAGE C ASE
device applications [10]. The non-real-time control is used To provide the continuous monitoring of three phases
only to set the machine parameters and configurations. Figure equipment’s electrical quantities, and to prevent, and correctly
1 illustrates the modules that compose the Smart Gateway identify some of the causes of failure the architecture described
utilized in the current work. The proposed Smart Meter is an in Section III was used to develop the Smart Meter usage case.
industrial appliance that acts as the Smart Gateway defined in Figure 2 shows the Smart Meter (SM) solution.
[10].
An overview of the Smart Gateway modules (Figure 1) is
described hereafter: The digital meter device used in this work (DM6200 from
Schneider’s Electric4 ) measures voltage, current, frequency,
• Device Drivers: Controllers to abstract the industrial and power factor among other quantities. This meter also
machine protocols and APIs on the production floor provides one RS485 serial interface and supports Modbus
for the Interpreter module; communication protocol.
• Interpreters: create and constantly update a virtual 2 http://www.json.org/
representation of machines, based on monitored data, 3 https://www.w3.org/XML/

in one predefined model for each industrial machine; 4 http://www.schneider-electric.com

 2017 IEEE 13th International Conference on Wireless and Mobile Computing, Networking and Communications (WiMob)

Fig. 2: Overview of the smart environment, shows the interconnection between the devices and services.

from the monitored machines, having different communication

interfaces (Ethernet, WiFi, and serial interfaces). These com-
munication features allow its use with various machines in the
industrial sector.
Attached to the BBB, a cape serial developed by Logic
Supply6 was used, which is an expansion board that adds some
interfaces to the hardware solution (three RS232 gates, one
CAN bus gate and one RS485 gate).

A. Controller Module (Device Driver)

The Controller Module (CM), representing the Device
Driver in the presented solution, is responsible for managing
the Modbus message exchange between the Smart Meter and
the DM6200 device (Figure 2). It was developed in C++ and
made available as a shared library (libsmartmeter.so). When the
Fig. 3: Electric scheme between the monitored machine and connection is established, two concurrent execution processes
the DM6200 device are activated in a publish/subscribe pattern, providing two
processes (as shown in Figure 2):
• Reader: It is an independent process that dispatches
The Schneider meter offers high accuracy with 1% of mar- frequent queries using RS485 interface to the DM6200
gin of error in current and voltage readings, and also provides device. Each query performs two actions in sequence:
the following measurements: (i) RMS values of Phase-to-Phase (i) it reads all the electrical quantity set provided by
and Phase-Neutral voltage and current; (ii) Currents through the DM6200 (through the Modbus protocol); and (ii) it
the CT’s (Current Transformer), in which can reach up to 75A calculates the real power, reactive power, and apparent
peak to peak in the monitored machine of the ABC Enterprise; power values. These power values are not provided
(iii) number of interruptions in energy supply. by the DM6200. At the end of these two actions, a
timestamp is generated for each specific query, which
These device meter features allow a wide electrical mea- in turn is added to the list of available queries for
surements in three phase systems. The schematic diagram conversion. This data is available for the Converter
in Figure 3 shows a connection made between the ABC sub-module.
Enterprise machine and the DM6200 to collect electrical data
during the machine operation. • Converter: This second process consumes the query
results generated by the Reader sub-module. Each
The DM6200 communication feature allows its integration query result in this sub-module is parsed in a JSON
as a sensor in the Smart Meter solution. The Smart Meter soft- format and called Complete Query (CQ), then it is
ware is embedded in the BeagleBone Black (BBB)5 , which is a inserted in a second list. The second list is available
single board computer open source. BBB works independently for the Interpreter module through the JNI interface.
5 http://beagleboard.org/BLACK 6 http://www.logicsupply.com/cbb-serial/
 2017 IEEE 13th International Conference on Wireless and Mobile Computing, Networking and Communications (WiMob)

TABLE II: Voltage Relative Values where V x and Ax represents values of voltage and current of a
specific phase, respectively. The Total Apparent Power (APA)
Resource Description is calculated by the Equation 2:
VLN Average of Voltage between Phases and Neutral
VLL Average of Voltage between Phases AP A = ACA/P F T (2)
VUN Unbalance of Voltage between Phases
VNN Neutral Voltage where ACA is the Real Power Average, and P F T is the Total
V12 RMS Voltage between Phase 1 and 2.
V23 RMS Voltage between Phase 2 and 3 Power Factor.
V31 RMS Voltage between Phase 3 and 1
V01 RMS Voltage between Phase 1 and Neutral The Equation 3 calculates the Real Power:
V02 RMS Voltage between Phase 2 and Neutral
V03 RMS Voltage between Phase 3 and Neutral T P x = AP x ∗ P F x (3)
VA1 Voltage angle in Phase 1
VA2 Voltage angle in Phase 2 where AP x is the Apparent Power of a given phase and P F x
VA3 Voltage angle in Phase 3 is the Power Factor of this phase. The Total Real Power (ACA)
value is calculated by Equation 4:
x=3
TABLE III: Current Relative Values X
ACA = TPx (4)
Resource Description x=1
ALA Current average in Three Phases
ANE Neutral current where T P x is the Real Power value of a given phase.
AUN Unbalance of Current in Three Phases
A01 RMS current in Phase 1 The Equation 5 calculates the Reactive Power:
A02 RMS current in Phase 2 p
A03 RMS current in Phase 3 RP 1 = AP 12 − T P 12 (5)
PA1 Angle of Current in Phase 1
PA2 Angle of Current in Phase 2
Pa3 Angle of Current in Phase 3
where AP x and T P x are the Apparent Power value and the
Real Power of a given phase, respectively. The Total Reactive
Power Equation 6 is:
In short, each query provided by the Controller Module x=3
X
consists of a set of electrical quantities linked to a timestamp. RP A = RP x (6)
The set of quantities that composes one query is shown in x=1
the following tables. Tables II and III indicate the voltage and The equations to calculate the Power value are described in
current relative values respectively, and Table IV indicates the [17] and [18].
power values.
B. Interpreter Module
TABLE IV: Power Values (read and calculated) The Interpreter Module is responsible for interface with
the C++ code and the RESTful API developed in Java. When
Resource Unit Description
PFA W/VA Average of the Power Factor in Three Phases employing a Java Native Interface (JNI) technology, it is
ACA W Average of Real Power possible to initialize, control and receive information from the
RPA VAR Average of the Reactive Power Controller Module.
APA VA Average of the Apparent Power
PF1 W/VA Power Factor in Phase 1
PF2 W/VA Power Factor in Phase 2 C. RESTFul API Module - WSMeter
PF3 W/VA Power Factor in Phase 3
TP1 W Real Power in Phase 1 When the Interpreter (JNI) and the Controller Modules are
TP2 W Real Power in Phase 2
TP3 W Real Power in Phase 3 executed, a buffer of the readings, represented in JSON format,
RP1 VAR Reactive Power in Phase 1 called Completed Query Buffer (CQB), is generated. This
RP2 VAR Reactive Power in Phase 2 buffer waits for a concurrent process (WSMeter application)
RP3 VAR Reactive Power in Phase 3
AP1 VA Apparent Power in Phase 1 responsible for collecting and sending it to the Cloud.
AP2 VA Apparent Power in Phase 2
AP3 VA Apparent Power in Phase 3 WSMeter is a web application that aims to gather electrical
quantity measurements. This application was developed in Java
All values from the three tables are represented by a 32- and has three software components: (i) the Schedule Manager;
bits integer data type. In Table II, the values are represented in (ii) REST Client; and (iii) REST Service.
Volts (V) unit. In Table III, the values are represented in Amper The Schedule Manager is responsible for schedule the
(A) unit. Finally, in Table IV, the values are represented in the collected data from the buffer to the Cloud. This scheduling
following units: Watt (W), Volt-Amper (VA), Watt Relative happens when the WSMeter is initiated by the web container
Amper (VAR) and Watt/Volt-Amper (W/VA), as identified in and it is verified by any new measurements every 2sec in the
Table IV. buffer (CQB). In case of new measurements, they are added
The equations used to calculate Real, Reactive and Ap- to a queue.
parent Power values, using the same nomenclature presented Another concurrent process verifies the presence of any
in Tables II, III and IV are shown next. The Equation 1 to new data in this queue to be sent to the Cloud every 5sec. If
calculate the Apparent Power (AP) to a given specific phase there is new data, the process uses a REST Client module to
is: send it to the cloud. The reading can be verified at any moment
AP x = V x ∗ Ax (1) through endpoints available by the REST Service module.
 2017 IEEE 13th International Conference on Wireless and Mobile Computing, Networking and Communications (WiMob)

D. Cloud computing environment • Complete Query CPU (CQC): is the effective CPU run
7
time to build a CQ and it did not consider the waiting
The cloud environment was implemented in the Azure and interaction times between the programs.
platform by Microsoft. The environment is composed by a
Web application, called Smart API, which is composed of • Smart Meter CPU Usage (SMC): is the CPU usage of
RESTful services and it is responsible for receiving data from JVM in the BBB executing the Smart Meter.
the Smart Meter. An asynchronous approach using a service
• Smart Meter RAM Usage (SMM): is the RAM mem-
bus (Enterprise Bus), managed by Azure, was adopted to
ory usage of JVM in the BBB executing the Smart
receive this data. This service bus was defined a structure
Meter.
for queuing messages that are received and forwarded to
the autonomous processors specified by the Smart Services Regarding data accuracy, two experiment sets were exe-
application. Furthermore, to optimize all the queries during cuted to evaluate the precision of the electrical quantities. In
the monitoring process (and improve the performance) Redis, the first set, the meter device (DM6200) was used in simulation
a memory data caching service, was used. The available data mode where the electrical quantities were predetermined. In
employed by Redis come from a relational database (SQL the second set, the measurements taken by the Smart Meter
Server), managed by Azure. This database offers structured were compared against the measurements taken by another
data for all of the smart meter’s resources. This way, whenever device, legacy meter device (ER4000), while the fabric-cutting
a monitoring routine happens, it is possible to identify which machine was in use.
resource and which smart meter belongs to the monitored data.
The Smart Service application is responsible for defining A. Smart Meter Application Impact
the asynchronous processes of the platform that implements The first experiment set evaluated the impact on the BBB
the monitored data persistence. The database used was the when gathering data, building and providing complete queries
DocumentDB, a non-relational database managed by Azure. in the Controller Module, considering application wall-clock
DocumentDB is a document-oriented database, which uses time (CQT) and run time (CQC).
a data model schema such as JSON, named BSON, which
facilitates the data integration in some applications. Measuring the complete query were observed the average
time to: (i) read the set of electrical quantities, (ii) to calculate
Finally, the application to visualize the monitored data the power values and (iii) to parse to JSON format. The time
are executed in Azure infrastructure. This application was to do these was 472.76 ms in CQT, and 70.44 ms in CQC;
developed using a "light" Web architecture through HTML5, and the size of a CQ is 3.5 KB RAM. The Table V shows
CSS, and Javascript. The application reads the monitored data, the average, the standard deviation and the variance for this
sends and consumes REST service data in JSON format, which assays.
is defined by the platform. Therefore, the application performs
queries directly in the DocumentDB database.
TABLE V: Controller Module application impact, wall-clock
time and run time
V. A NALYSIS OF E XPERIMENTS
CPU Time (ms) Total Time (ms)
The goal of the experiments was to evaluate the solu- Average 70.439 472.757
tion in two different aspects: performance and data accuracy. Standard Deviation 0.974 27.190
The experiments involved an industrial fabric-cutting machine Variance 0.950 739.314
through a Three Phase electric power system, a Schneider
Electric meter device (DM6200) connected to the same power The second experiment set evaluated the impact of CPU
system, and the smart meter prototype embedded in a Beagle usage and RAM memory on BBB while the Smart Meter was
Bone Black (BBB) board on the production floor. The BBB in use. The Profiling JConsole tool was used to measure these
board has an AM335x 1GHz ARM R Cortex-A8 processor, values in preset time samples and is capable to monitor the java
512MB DDR3 RAM, and 4GB flash memory, and it was application remotely. Figure 4 shows the JConsole Interface
connected to the company’s Ethernet and to the DM6200 while monitoring Smart Meter Application.
through a RS485 link.
Table VI shows the averages and the standard deviations
In regards to performance evaluation, two sets of experi- of the experiment results. In there, we can see that average
ment were executed to analyze the CPU usage in the Smart CPU usage was 24% of the evaluated time and the average
Meter. The first experiment set considered only the Controller RAM usage was 30.4 MB. In this experiment, samples were
Module (CM) processes to build complete queries (CQ). The collected running Smart Meter application in the JVM (Java
second set considered the entire BBB Java Virtual Machine Virtual Machine) in ten groups with a refresh interval of 10
(JVM) processes when the Smart Meter was in use. The seconds for reading data.
evaluated metrics from both experiment sets are as follows:
B. Data Accuracy Results
• Complete Query Time (CQT): is the total time to build
a Complete Query (CQ) by the Controller Module in To evaluate the data accuracy between the monitored and
wall-clock time and it considers the CPU sharing time. the cloud persisted data, two experiments were executed. First,
the simulation model of meter device DM6200 was used;
7 https://azure.microsoft.com/pt-br/ and second, the Smart Meter data was compared against
 2017 IEEE 13th International Conference on Wireless and Mobile Computing, Networking and Communications (WiMob)

TABLE VII: Average - Voltage, Current and Power Factor

Average
Phase 1 Phase 2 Phase 3
Voltage 244,859 244,915 244,529
Current 52,000 51,941 52,139
Power Factor -0,503 -0,504 -0,504

measured data by the Embrasul ER4000. To do this, the Smart

Meter and the Embrasul ER4000 acquired data simultaneously.
Figure 6 compares the Voltage measurements, and Figure 7
compares the Current measurements between the Smart Meter
and the ER4000, respectively.

Fig. 4: JConsole Interface

TABLE VI: Smart Meter application impact experiments re-

sults
CPU Usage (%) RAM Usage (MB)
Average 24.000 30.400
Standard Deviation 29.048 4.400

the acquired data from Embrasul ER4000 device, both data

acquired in real industrial machine. Figure 5.A, Figure 5.B Fig. 6: Comparison between Voltage values acquired by Em-
and Figure 5.C shows the voltage measurements in simulation brasul equipment and by Smart Meter
mode with the fixed voltage, current and power factor values
in Three Phases respectively. Moreover, the Figure 5.D and
Figure 5.E illustrates the Voltage in Phase 1 and the Current
in Phase 2 stored in the cloud.
In this experiment, (i) the fixed values in simulation mode
were read and sent to the cloud. To evaluate the reading
data accuracy, three groups of measurements were evaluated
sending 100, 500 and 1000 messages to the cloud. Table VII
shows the average voltage, current and power factor values.
The cloud acquired values were consistent with the
DM6200 data without variance or standard deviation in first
three decimal digits.
The second experiment analyzed the consistency of ac- Fig. 7: Comparison between Current values acquired by Em-
quired data by the Smart Meter (BBB and DM6200) with the brasul equipment and by Smart Meter

Other comparison between Smart Meter and Embrasul

ER4000 were calculated the simple arithmetic average of the
40 acquired values of Voltage (Vx) and Current (Ix). Table
VIII shows the average, standard deviation and variance of
these values in each Phase.

The results of this acquisition demonstrate that the inter-

mediates values varied due to the sensitivity, gentle curves
and sharp curves in different cases. However, the two devices
read proportionate values over time (Figure 6 and Figure 7).
Regarding the average, standard deviation and variance (Table
Fig. 5: Fixed values in simulation mode in DM6200, and values VIII), the variation can be explained through equipment toler-
persisted in cloud ance, and it was considered acceptable by the ABC Enterprise
engineers.
 2017 IEEE 13th International Conference on Wireless and Mobile Computing, Networking and Communications (WiMob)

TABLE VIII: Comparison between Smart Meter and ER4000 to read Voltage and Current in each Phase
Smart Meter Embrasul ER4000
Va Vb Vc Ia Ib Ic Va Vb Vc Ia Ib Ic
Average 205.67 201.15 208.78 13.87 16.85 15.40 206.02 201.36 209.23 12.51 15.39 14.40
Deviation 2.58 2.59 2.61 15.48 16.73 16.06 2.90 2.83 2.83 16.70 18.35 17.52
Variance 6.69 6.73 2.85 239.87 280.12 258.23 8.41 8.04 8.08 279.03 337.04 307.03

VI. C ONCLUSION [6] A. Whitmore, A. Agarwal, and L. Da Xu, “The internet of

things—a survey of topics and trends,” Information Systems Frontiers,
The proposed solution for continuous monitoring of elec- vol. 17, no. 2, pp. 261–274, 2015. [Online]. Available: http:
trical quantities in smart industrial environments that make //dx.doi.org/10.1007/s10796-014-9489-2
use of three phase systems (Smart Meter) has proven feasible [7] P. Parwekar, “From internet of things towards cloud of things,” in Com-
puter and Communication Technology (ICCCT), 2011 2nd International
for the real case study of the ABC Enterprise. The proposed Conference on, Sept 2011, pp. 329–333.
solution follows an architecture based on smart gateways
[8] S. Karnouskos, D. Guinard, D. Savio, P. Spiess, O. Baecker, V. Trifa,
for controlling and monitoring of smart environments and L. Moreira, and S. Souza, “Towards the real-time enterprise: Service-
cloud computing. The use of the cloud allows a scalable based integration of heterogeneous soa-ready industrial devices with
infrastructure, which allows allocation of resources on demand enterprise applications,” 2009.
to the Web application and persistent data services. [9] D. Zeng, S. Guo, and Z. Cheng, “The web of things: A survey,” Journal
of Communications, vol. 6, no. 6, 2001.
The experimental results indicate that to acquire monitored [10] M. C. Domenech, L. P. Rauta, M. D. Lopes, P. H. D. Silva, R. C. D.
values, the sampling time was lower than 500ms, which is Silva, B. W. Mezger, and M. S. Wangham, “Providing a smart industrial
considered acceptable for reading and calculating different environment with the web of things and cloud computing,” in 2016
IEEE International Conference on Services Computing (SCC), June
electric quantities. Therefore, this solution provides continuous 2016, pp. 641–648.
monitoring of these quantities. According to the computational
[11] C. Selvam, K. Srinivas, G. Ayyappan, and M. Venkatachala Sarma,
impact, the experiments showed that the hardware used in the “Advanced metering infrastructure for smart grid applications,” in
proposal is capable of supporting the necessary processing. Recent Trends In Information Technology (ICRTIT), 2012 International
Conference on, April 2012, pp. 145–150.
As a proof of concept, a web application was developed to [12] D. Gallo, C. Landi, M. Luiso, G. Bucci, and E. Fiorucci, “Low cost
access the non-relational database and to provide stored data smart power metering,” in Instrumentation and Measurement Technol-
visualization through reports composed by graphs and tables, ogy Conference (I2MTC), 2013 IEEE International, May 2013, pp. 763–
with specific time interval options. The application also enables 767.
the display of power consumption in time intervals [13] C.-H. Chen, C.-Y. Chen, C.-H. Hsia, and G.-X. Wu, “Big data collection
gateway for vision-based smart meter reading network,” in Big Data
The Smart Meter Usage Case described in this article (BigData Congress), 2014 IEEE International Congress on, June 2014,
pp. 266–269.
generates a huge data load. How to extract and present to
[14] B. Bitzer and T. Kleesuwan, “Cloud-based smart grid monitoring and
the users and services relevant information is a critical issue controlling system,” in Power Engineering Conference (UPEC), 2015
that will need to be better addressed in future work. Moreover, 50th International Universities, Sept 2015, pp. 1–5.
as a future work, intelligent applications can be developed to [15] D. F. Sadok, L. L. Gomes, M. Eisenhauer, and J. Kelner, “A
provide remote proactive and reliable machine maintenance. middleware for industry,” Computers in Industry, vol. 71, pp. 58 –
Finally, we intend to use fog computing and data fusion 76, 2015. [Online]. Available: http://www.sciencedirect.com/science/
algorithms to develop a more efficient and cost-effective Smart article/pii/S0166361515000615
Meter solution. [16] N. Nurseitov, M. Paulson, R. Reynolds, and C. Izurieta, “Comparison
of JSON and XML Data Interchange Formats: A Case Study,” in
CAINE, D. Che and D. Che, Eds. ISCA, 2009, pp. 157–162. [Online].
Available: http://dblp.uni-trier.de/rec/bibtex/conf/caine/NurseitovPRI09
R EFERENCES [17] “Ieee standard definitions for the measurement of electric power
[1] PwC, “Industry 4.0: Building the digital enterprise,” 2016. quantities under sinusoidal, nonsinusoidal, balanced, or unbalanced
conditions,” IEEE Std 1459-2010 (Revision of IEEE Std 1459-2000),
[2] D. B. Ramamurthy, “Development of a low-cost gsm sms-based humid- pp. 1–50, March 2010.
ity remote monitoring and control system for industrial applications,”
2010. [18] A. Emanuel, “Summary of ieee standard 1459: definitions for the
measurement of electric power quantities under sinusoidal, nonsinu-
[3] A. Biral, M. Centenaro, A. Zanella, L. Vangelista, and M. Zorzi, “The soidal, balanced, or unbalanced conditions,” Industry Applications,
challenges of {M2M} massive access in wireless cellular networks,” IEEE Transactions on, vol. 40, no. 3, pp. 869–876, May 2004.
Digital Communications and Networks, vol. 1, no. 1, pp. 1 –
19, 2015. [Online]. Available: http://www.sciencedirect.com/science/
article/pii/S235286481500005X
[4] C. Antón-Haro and M. Dohler, “1 - introduction to machine-
to-machine (m2m) communications,” in Machine-to-machine (M2M)
Communications, C. A.-H. Dohler, Ed. Oxford: Woodhead Publishing,
2015, pp. 1 – 23. [Online]. Available: http://www.sciencedirect.com/
science/article/pii/B9781782421023000010
[5] M. R. Palattella, N. Accettura, L. A. Grieco, G. Boggia, M. Dohler,
and T. Engel, “On optimal scheduling in duty-cycled industrial iot
applications using ieee802.15.4e tsch,” IEEE Sensors Journal, vol. 13,
no. 10, pp. 3655–3666, Oct 2013.