Institute of Architecture of Application Systems

University of Stuttgart
Universitätsstraße 38
D–70569 Stuttgart

Master Thesis

IoT based monitoring and

control for energy management
system
Karim, Ahmad

Course of Study: INFOTECH

Examiner: Prof. Dr. Dr. h.c. Frank Leymann

Supervisor: Dipl.-Inf. Michael Hahn,

Dr. Dirk Pietruschka (Enisyst GmbH)

Commenced: October 19, 2017

Completed: April 19, 2018

Abstract

The world economy is growing rapidly, and global energy demands are predicted to increase
even more in the future. Energy is expected to get more expensive, in turn affecting the
economic development. Energy demand can be reduced by employing efficient Energy
Management Systems (EMS). The required infrastructure cost for EMS is often quite high,
making it an unviable solution. However, new Internet of Things (IoT) based solutions can
help to reduce the installation effort as well as cost significantly. Low-Power Wide-Area
Networks (LPWAN) hold the solution to this problem. The goal of an LPWAN is to provide
cheap nodes and communication distances of a few kilometers (5-10 km) with a low battery
consumption rate (5-10 years). The data transfer rate is low, but the range and power
consumption make it ideal for environmental data acquisition (samples per few minutes).
The current wireless technologies, such as ZigBee, Bluetooth, and Wi-Fi, that are being
used for wireless sensors are not suitable for industrial use where the number of connected
devices is significantly higher, and reliability is of dire need. Therefore, in this work, a
study of available LPWAN technologies is carried out which concludes in the selection
of LoRa technology for communication in an EMS. Furthermore, we evaluate possible
communication distances of the LoRa technology in an industrial area by conducting several
range tests which result in significantly higher communication distances as compared to
legacy WAN technologies. LoRa has proved to be a good start for adopting LPWAN,
especially in applications such as energy management systems. Based on that, we present a
system architecture for an EMS using LoRa as the underlying communication technology as
well as a prototypical implementation of it. This implementation is furthermore integrated
into the current EMS of Enisyst. LoRa-modulation combined with LoRaWAN communication
protocol proves to be a base for a reliable and scalable system.

3
Contents

1 Introduction 11
1.1 Motivation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
1.2 Document layout . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13

2 Background 15
2.1 Communication terms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
2.2 Network . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
2.3 Hardware Communication Protocols . . . . . . . . . . . . . . . . . . . . . . 24
2.4 LPWAN . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
2.5 History . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
2.6 State of the art in LPWAN . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26

3 Selection of an LPWAN 39
3.1 Selection Criteria . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39
3.2 Comparison . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40
3.3 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43

4 Concept 45
4.1 Energy Management System . . . . . . . . . . . . . . . . . . . . . . . . . . . 45

5 Prototypical Implementation of an LPWAN based EMS 47

5.1 Hardware Setup . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47
5.2 System Architecture . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51

6 Range Assessment 61
6.1 Range test . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61

7 Conclusion and Outlook 65

Bibliography 67

5
List of Figures

1.1 Over view of an EMS. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12

2.1 A basic wireless system. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16

2.2 A signal with an SNR of higher than one. . . . . . . . . . . . . . . . . . . . . 19
2.3 OSI layer model [46]. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
2.4 Star topology. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
2.5 Bus topology. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
2.6 Ring topology. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
2.7 Mesh topology. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
2.8 SPI Master-slave configuration. . . . . . . . . . . . . . . . . . . . . . . . . . 24
2.9 Classification of LPWAN. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
2.10 Sigfox band specification [35]. . . . . . . . . . . . . . . . . . . . . . . . . . 28
2.11 Random access [35]. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
2.12 Sigfox network architecture [35]. . . . . . . . . . . . . . . . . . . . . . . . . 29
2.13 LoRa and LoRaWAN stack [22]. . . . . . . . . . . . . . . . . . . . . . . . . . 30
2.14 LoRaWAN network topology [22]. . . . . . . . . . . . . . . . . . . . . . . . 31
2.15 LoRaWAN device classes [22]. . . . . . . . . . . . . . . . . . . . . . . . . . . 32
2.16 Excerpt of OSI layers based on [42]. . . . . . . . . . . . . . . . . . . . . . . 34

4.1 Block diagram of an EMS extracted from [13]. . . . . . . . . . . . . . . . . . 46

5.1 phyCore AM3352. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47

5.2 phyBOARD Regor. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48
5.3 RFM95W LoRa transceiver. . . . . . . . . . . . . . . . . . . . . . . . . . . . 48
5.4 RFM95W mounted on the break out board. . . . . . . . . . . . . . . . . . . 49
5.5 DS18B20 digital temperature sensor. . . . . . . . . . . . . . . . . . . . . . . 49
5.6 1611 extension module. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50
5.7 LoRa concentrator module ic880a. . . . . . . . . . . . . . . . . . . . . . . . 50
5.8 Top level view of the whole system. . . . . . . . . . . . . . . . . . . . . . . . 51
5.9 Detailed system architecture. . . . . . . . . . . . . . . . . . . . . . . . . . . 52
5.10 LoRa node. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53
5.11 Scheduler flow chart. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54
5.12 Structure of the first two bytes of a hex message. . . . . . . . . . . . . . . . 54
5.13 Class diagram. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55
5.14 Internal software modules at the gateway end. . . . . . . . . . . . . . . . . 56
5.15 Network configuration using LoRa Gateway bridge and packet forwarder. . . 58

6.1 Paths with SNR values. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62

7
 6.2 Area coverage. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63

8
List of Tables

2.1 Generic dBm values for transmitter and receiver. . . . . . . . . . . . . . . . 17

3.1 Comparison of various LPWAN technologies and standards extracted from

[10, 18, 22, 32, 35]. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42
3.2 LoRa versus Sigfox comparison (table taken from [25]). . . . . . . . . . . . 43

5.1 Structure of a HEX message of the 1611 node. . . . . . . . . . . . . . . . . . 53

5.2 Package dependencies. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56

9
1 Introduction

The world economy is growing rapidly, and global energy demands are predicted to get
higher in the future. Even though the energy industry is moving from conventional sources
(oil, gas, and coal) to clean and renewable sources, fossil fuels still dominate this industry,
which in turn is a cause of rising CO2 levels in the environment and also results in other
environmental problems. Energy is expected to get more expensive and thereby relevant
for the economic development. The increase in efficiency holds a very high potential to
reduce energy demand, decoupling economic growth from energy demand. This effect is
visible in Germany, where already substantial efforts have been taken to increase energy
efficiency in all sectors; however, the building sector in Germany holds even more potential
regarding energy efficiency. Apart from insulation measures on the buildings, intelligent
and efficient control has the potential to increase the energy efficiency of buildings by low
energy heating and cooling systems, this requires active energy management systems for
optimized control and observation to avoid inefficient energy use.

1.1 Motivation

Active energy management systems (EMS) comprise of a sensor network and control units
collecting a huge amount of data, which is then analyzed and used for optimization. On
the other hand, installation of sensors and the required cabling is expensive, and therefore
in some cases, it is not economically viable to install an EMS; however, new IoT based
solutions can help to reduce the installation effort as well as cost significantly. A wireless
sensor network (WSN) can be employed to tackle this issue. Yinbiao et al. [43] define WSN
as follows:

“A WSN is a network formed by a large number of sensor nodes where each

node is equipped with a sensor to detect physical phenomena such as light, heat,
pressure, etc. WSNs are regarded as a revolutionary information gathering
method to build the information and communication system which will greatly
improve the reliability and efficiency of infrastructure systems. Compared with
the wired solution, WSNs feature easier deployment and better flexibility of
devices.”

The data collected from these sensors can be directly transferred to a cloud server where
it can be stored and used by control systems for efficient control. Such solutions are
emerging, but they suffer from low range, low data transfer rate, or short battery cycles.
A wireless communication network known as Low-Power Wide-Area Network (LPWAN)

11
1 Introduction

holds the solution to this problem. The goal of an LPWAN is to provide cheap nodes and
communication distances of a few kilometers (5-10 km) with a low battery consumption rate
(5-10 years) [27]. Undeterred by the low data transfer its range and power consumption
makes it ideal for environmental data acquisition where data is gathered in samples per
few minutes. The current wireless technologies such as ZigBee [45], Bluetooth [4] and
Wi-Fi [14] that are being used for conventional WSN [37] are not suitable for industrial
use, where the number of connected devices is significantly higher and reliability is of dire
need. The major difference between a conventional WSN and an LPWAN is of the network
topology; WSN consists of a mesh or ad-hoc topology, whereas LPWAN employs a star
topology. The nodes are connected to one base station which is further connected to a wired
network, and this scheme enables LPWAN to extend the network coverage to the order of
dozens of kilometers [27]. Examples of such networks are “LoRaWAN” [23], “Sigfox” [34]
and “Weightless” [41]. State of the art LPWAN technologies will be thoroughly discussed in
Section 2.4.

Figure 1.1: Over view of an EMS.

EMS relies on the sensor network to acquire environmental data be it indoors or outdoors.
This data is crucial for efficient energy use. A building’s heating and cooling are managed
by reading the sensor data. These sensors are located at various points of interests in the
building. The EMS actively adjusts the temperature to maintain a pleasant atmosphere
during active hours, and it can be scheduled to turn off the heating system during idle
hours, e.g., night or weekends. Using an EMS is not just environment-friendly but also
has its economic benefits. These economic benefits are notably prominent on the high
scale projects, e.g., malls, multi-storied buildings, and commercial buildings, etc. In an

12
 1.2 Document layout

overview of the currently employed EMS shown in Figure 1.1, inputs from various sensors
and meters are read by the controller and corresponding decisions are made. Creating the
infrastructure for EMS is an integral part of a building’s construction work. The construction
plans are analyzed by project engineers months before construction begins. They provide
the construction firm with the locations of sensors. These sensors and their required cabling
are then installed during the construction of the building. There are major problems in this
scenario. The sensors are connected to the controller in a star topology, which increases the
installation effort; this scenario is difficult to implement for already constructed buildings
with residents in them. Apart from these problems, cable length can easily increase from a
few meters to several hundred meters in a multi-storied building, requiring repeaters and
signal amplifiers hence further increasing cost and effort.
Enisyst is an energy solution provider with the focus on smart energy management. They
offer an EMS based on industrial grade hardware assuring reliability and robustness. They
offer clients a 30% reduction in energy cost by using energy smartly. Enisyst’s current
solution is comprised of a wired system, as described above. Enisyst is also facing the
problem of increased effort and difficulties in deployment for already constructed buildings.
All these problems can be addressed by switching to wireless sensors.
In the area of wireless communication, significant advancements have been made in the
past few decades, providing consumers with various options. It is, therefore, an integral
part of this thesis to carry out a study on the wireless communication technologies and
selecting one for implementation an EMS.

1.2 Document layout

A significant part of this document is dedicated to communications technology. To make

it easier for the readers to understand Chapter 2 will be focused on terminologies and
basic concepts of communication systems. Section 2.4 of Chapter 2 will present the
state of the art of LPWAN technologies. A comparison of the available technologies is
carried out in Chapter 3. Based on the results of the comparison, one of the introduced
LPWAN technologies is selected for implementing an EMS. In Chapter 4 a standard EMS
is discussed with its various components. In Chapter 5, hardware components and their
interconnections required for realizing an EMS with LPWAN technology will be explained.
Chapter 5 will also present the system architecture of the EMS as well as the detailed
working of the entire system and its internal modules. Results of the range tests will be
discussed in Chapter 6 with Chapter 7 concluding this thesis.

13
2 Background

In principle, any system that transmits information from one location to another location
without the use of wires is considered as a wireless system. Wireless communication is
achieved by using electromagnetic waves as a medium to transmit electrical signals instead
of wires. A wireless system is a vast topic, and a complete explanation of all the internal
workings are out of scope for this thesis, only the terms and concepts which are relevant
will be discussed in this chapter.
Wireless transmission uses waves to transmit information. In the realm of physics, waves
can be categorized by two kinds, mechanical waves, and electromagnetic waves. Mechanical
waves are those waves which need a medium to travel and cannot propagate through the
vacuum, for example, sound waves. Electromagnetic (EM) waves, on the other hand, do
not require a medium to propagate for instance visible light, radio waves, microwaves,
x-rays, etc. Communication systems use EM waves as a communication medium as their
ability to move through objects is far more than mechanical waves.
The characteristics and behavior of EM waves depend directly on their operating frequency,
i.e., wavelength. Low-frequency radio waves have a longer wavelength and can travel long
distances without attenuation, but they cannot carry much information. Contrarily, high-
frequency waves (shorter wavelength), like microwaves, are susceptible to interference
and prone to attenuation for long distances, but they can carry more information.
Digital signals cannot be directly transmitted wirelessly. They need to be converted
to a different form to be transmitted. As discussed earlier EM waves are used as a
communication medium for wireless systems, and conversion of information from digital
bits to EM waves is a process that is called modulation. There are several different
techniques for modulation. The operating frequency and modulation technique actively
govern the characteristics of any wireless technology. For example, Radio stations use
76 MHz-108 MHz as the base frequency for transmission and "Frequency-Modulation" as
modulation technique to transmit voice or songs over the air. Whereas cellular services
use microwaves (> 1 GHz) as the base frequency for transmission and shift keying as
their modulation technique. More information can be transmitted on the same operating
frequency by just changing the modulation technique, e.g., GSM, GPRS and EDGE. All are
cellular technologies that use same base frequency, but EDGE is capable of three times data
capacity as that of GPRS due to its more advanced modulation technique.
An ideal wireless system would be one that has the most extended range, consumes the least
amount of power, is resistant to interferences, and has the highest possible data throughput
while consuming the lowest possible bandwidth. All these parameters are interlinked,

15
2 Background

some inversely and some directly; high data throughput requires more bandwidth, more
bandwidth introduces more noise, and long range requires more power.

Figure 2.1: A basic wireless system.

Figure 2.1 shows a primary wireless communication system with two transceivers com-
municating over the air. A transceiver is a device that is capable of both transmitting and
receiving data. A basic transceiver has an antenna and a modulator. An antenna is an
electrical device that converts electric current to EM energy and vice versa; a modulator is
a device that converts the raw information into a form that can be transmitted through EM
waves.

2.1 Communication terms

This section will be focused on the physical properties of a communications system. The
subsections will include a brief description of relevant terminologies.

2.1.1 Channel

The medium through which signals travel is known as the channel. In wired communication,
it is the copper wire, in space communication, it is the vacuum of space, and in wireless
communication, the communication channel is the air. A channel is modeled by the
calculation of processes that cause changes to the signal. In a wireless communication
system, this can be achieved by calculating the signals strength before and after it reflects
off any surface or object.

16
 2.1 Communication terms

2.1.2 Decibel (dBm)

dBm is a logarithmic way of describing a signal and is the most common unit in communi-
cation systems. It is a ratio of the power of desired signal compared to 1 milliwatt of power.
The difference in decibels between two quantities is defined by Equation (2.1).

P1
[10 ∗ log ] (2.1)
P2

In communication systems, numerical values tend to get minute, and it is best to describe
them in terms of logarithms for more comfortable usability and readability. Table 2.1 shows
generic dBm values used in a communication system; not only does the dBm scale provide
ease of use but it is also the norm for expressing values in communication systems.

P (dBm) P (mW)
50 100000 strong transmitter
40 10000
30 1000
20 100
10 10
0 1
-10 0.1
-20 0.01
-30 0.001
-40 0.0001
-50 0.00001
-60 0.000001
-70 0.0000001
-80 0.00000001 sensitive receiver
-90 0.000000001

Table 2.1: Generic dBm values for transmitter and receiver.

2.1.3 Bandwidth

The range of frequencies that is used for transmission is known as bandwidth. Signal
bandwidth also refers to the frequency range in which a signal’s spectral density is nonzero.
Spectral density is the region of frequencies where the average power of a signal is
distributed.

17
2 Background

2.1.4 Antenna gain

Antenna gain is a performance parameter. It quantifies an antennas directivity and efficiency.

Directivity is a measure of the strength of radiation emitted in a particular direction.
Efficiency is a measure of conversion of electrical energy into radio waves and vice versa in
case of a receiving antenna.

2.1.5 Modulation

Modulation is a process of converting information to a form that can travel farther with
a lesser loss. A prime example of analog modulation is AM and FM radio. Sound waves
are converted into electrical signals and propagated through air. There are also digital
modulation techniques where discrete signals are converted to transmissible EM waves.
Some examples of digital modulation are PSK (phase-shift keying), FSK (frequency-shift
keying), ASK (amplitude-shift keying) and QAM (quadrature amplitude modulation).

2.1.6 Link budget

Transmitting information in terms of EM waves is subject to many losses. These losses

are at the transmission end, receiving end, in the channel, due to weather, etc. All these
losses when accounted together are known as link budget. Equation (2.2) depicts a simple
version of a link budget equation.

Received Power (dB) = Transmitted Power (dB) + Gains (dB) ∗ Losses (dB) (2.2)

Link budget can be referred to as a quality parameter for a communication system. It is

one value portrayed in dBm that gives an overview of the entire system.

2.1.7 Signal to Noise Ratio

Signal to noise ratio (SNR) is one of the most important parameters when it comes to a
communication system. SNR is the power level of the desired signal when compared with
the background noise. An SNR of less then one would mean that noise power is higher
than the desired signal hence making the signal hard to detect. SNR of a signal can be
calculated by Equation (2.3).

Psignal
SNRdb = 10 ∗ log( ) (2.3)
Pnoise

Figure 2.2 shows a signal in the frequency domain in comparison with the background
noise. It can be seen that background noise is present in all the frequencies whereas the

18
 2.2 Network

desired signal has a power level higher than the noise, showing an SNR of greater than
one.

Figure 2.2: A signal with an SNR of higher than one.

In general an SNR of greater then one is a measure of good signal strength. Lower SNRs,
notably less than one, may not always mean that signal is undetectable. There are special
techniques through which a signal whose power level is even below noise can be detected,
and communication can be carried out.

2.2 Network

In the previous section, physical properties of communication systems were discussed. This
section will cover networks and standardization in network technologies, communication
between devices on a higher abstraction level and different network topologies.

2.2.1 OSI Layer Model

Interconnected paths for exchanging information between heterogeneous devices is known

as a network. All the communication performed in a network is easily explained by the
Open System Interconnect (OSI) model [46]. The OSI model is a network framework that
provides an abstract view of the communication processes by dividing the various tasks into
seven layers. Each layer describes a small task in the longspun process of communication.
The lower layers of this model deal with the electrical signals, whereas the higher layers
deal with applications and network protocols. Figure 2.3 gives an overview of the OSI layer
model, which will be briefly described next.

19
2 Background

Figure 2.3: OSI layer model [46].

Application Layer

The application layer specifies shared interface methods and protocols. Hypertext Transfer
Protocol (HTTP), File Transfer Protocol (FTP) and Simple Mail Transfer Protocol (SMTP)
are a few examples of application layer protocols.

Presentation Layer

Operating systems such as (Windows, Linux and MacOS) reside on this layer and provide
basic services to the application layer.

Session Layer

This layer is responsible for creating communication between network devices, e.g., a
network session between an end user and a server.

Transport Layer

The transport layer enables communication between different applications. Each application
uses its own communication channel, making it possible to communicate with applications
running on different systems. The Transmission Control Protocol (TCP) is the common
transport layer protocol.

20
 2.2 Network

Network Layer

All the routing of data is performed at the network layer. The network protocol for the
internet is known as IP, which is responsible for providing addresses to the devices and
carrying data from one device to another.

Data Link Layer

The link layer is responsible for converting digital signals, such as bits into transmittable
signals. It also manages data encapsulation for reliable transmission.

Physical Layer

The physical layer is the bottom layer of the OSI model. It specifies the physical properties
of a communication channel. This layer is responsible for transmitting data from the data
link layer through a communication channel.
A layered network is scalable and flexible, but it is intricate and requires more memory
and processing power because every layer puts on additional information that causes
data overhead. A layer-less network design would be a proprietary application protocol,
communicating directly on the physical layer like a temperature sensor in a wireless
sensor network. This is a very efficient implementation, but it can only be used for simple
applications and will not be able to handle a complex network configuration.

2.2.2 Network types

On a very basic level, networks are divided into the following three classes:

• Personal Area Network (PAN),

• Local Area Network (LAN),
• Wide Area Network (WAN).

Each of these networks differ in respect to range, data throughput and application.

Personal Area Network

This kind of network only covers a small area also referred to as personal space. Bluetooth
[4] is one of the prime examples of protocols that conform to PAN, other examples include
wired USB and Wi-Fi [14].

21
2 Background

Local Area Network

LAN is one of the most common types of networks. Modern-day LAN is based on Ethernet
and its coverage area includes homes and buildings.

Wide Area Network

WAN is a term used to describe networks that cover a wider area as compared to LAN, which
is a very generic description. A WAN could be a network connecting multiple buildings to
multiple continents. The Internet is the most prominent example of a WAN.

2.2.3 Network Topologies

Network components have to be connected with each other for them to communicate.
This arrangement of network components is known as a network topology. The network
topology is one of the core design elements in communication systems. It is one of the
factors that govern the primary features of a network including scalability, robustness, data
throughput, flexibility, and complexity.

Star Topology

One central server manages all the communication between all the network components
in a star topology. A star topology has low complexity is easy to setup and modify, and
requires less effort for upgrades. Figure 2.4 shows a star topology network.

Figure 2.4: Star topology.

The downside of the star topology is its complete dependency on the server; the perfor-
mance of the whole system is reliant on the server, i.e., the whole network fails if the server
crashes.

22
 2.2 Network

Bus Topology

In a bus topology all the network components are connected to one single cable making
it cost-effective, it also requires less infrastructure compared to other topologies. Bus
topologies are more suited to small networks with low traffic.

Figure 2.5: Bus topology.

The drawbacks of a bus topology are, slower speeds in case of high traffic and dependence
on one central link. Figure 2.5 depicts a bus topology network.

Ring Topology

In a ring topology, the network components are connected to each other in a ring fashion as
shown in Figure 2.6. Every single component in a ring topology has precisely two neighbors
eliminating the need for a router.

Figure 2.6: Ring topology.

Transmission is unidirectional, with the sequential transfer of data. It is cheap to install

and expand. All the components are interdependent, hence the failure of one component
causes the whole network to crash.

Mesh Topology

In a mesh topology, every network component is connected to all the other components.
There are two modes of operation for data transfer: routing and broadcast.

23
2 Background

Figure 2.7: Mesh topology.

In case of routing, devices are configured to send data to the required destination. This
operation requires some routing logic at target device, whereas the broadcast mechanism
transmits data to all the network components. The broadcast mode is easy to implement,
but adds overhead in the network traffic. Figure 2.7 depicts a mesh topology network.

2.3 Hardware Communication Protocols

The three most extensively used protocols for the implementation are RS232, 1-wire
protocol, and SPI (Serial Peripheral Interface). These protocols are briefly discussed
further.

2.3.1 Serial Peripheral Interface

The Serial Peripheral Interface bus (SPI) is a synchronous serial communication interface
used for short distance communication in embedded systems. SPI devices communicate in
full duplex mode using a master-slave architecture with a single master. The master device
originates the frame for reading and writing.

Figure 2.8: SPI Master-slave configuration.

24
 2.4 LPWAN

Multiple slave devices are supported through selection with individual slave select (SS) lines
as shown in Figure 2.8. SPI is used for communication between the 1611 and RFM95W
at the node end, and between the concentrator board and 1613 at the gateway end (see
Section 5.2).

2.3.2 1-Wire Protocol

As the name suggests, devices conforming to the 1-wire protocol use only one wire plus
a ground wire to transmit signals. This communication protocol is used by the DS18B20
temperature sensor (see Section 5.1.4).

2.3.3 RS232

RS232 is an asynchronous serial communication protocol that is used for communication

between 1613 and a PC (see Section 5.1). Though communication with the 1613 can be
established via an ssh client program, for initial board setup RS232 is used.

2.4 LPWAN

LPWAN is an acronym for "Low-Power Wide-Area Network" and is conceived to identify

particular kinds of networks, with unique operating capabilities. LPWANs tend to have a
long range and low power consumption as opposed to the conventional networks. This
advantage of long range and low power consumption comes at the expense of low data
rates. The sensors operating on this technology could run for 5 to 10 years without
replacing battery.
Beecham that is an analyst firm has forecasted LPWANs to be the backbone of IoT market.
It is predicted to hold a quarter of the total IoT connectivity market. Beecham stated that
by 2020 there would be approximately 345 million IoT connections. The use of unlicensed
frequency and open source technologies will end the mobile network operators monopoly
in IoT networks [39].
LPWAN is a new technology. The rapid growth in the IoT sector is the cause of the sudden
increase in the device count, creating a demand for LPWAN networks. The term LPWAN
was coined somewhere in the 2000s, but the idea behind it was first seen in the late 1980s
[29].

25
2 Background

2.5 History

Networks similar to LPWAN already existed in the end 1980s. One such network was
“AlarmNet” which was created by Alarm Device Manufacturing Company (ADEMCO), one
of the biggest producer of alarm panels and systems. ADEMCO built a 900 MHz network
for monitoring of the alarm panels, and as the data sent by the alarms was minimal, low
data rates were acceptable [29].
With the birth of 2G in the late 1990s, it was possible to transmit voice and data on the
same network, and it was no longer necessary to have a dedicated network for alarm
systems. Thus the need for such long-range and low data transfer rate networks died out
[29].
Another Motorola owned network known as ARDIS had a similar fate. ARDIS was also
a low-speed network designed for sales automation, fleet tracking, e-mail, transaction
processing and messaging [29].
This concept of long range and low power disappeared in the 80s because there were
never enough users and IoT was nonexistent. The use of this concept resurfaced with
the introduction of IoT and increasing number of devices to be connected. Today the
requirement for low cost and low data devices has increased even more.
SIGFOX [34], Ingenu (formerly OnRamp) [16], members of LoRa Alliance [23] etc. are
current active contributors to the LPWAN technology each with their own respective
approach, some targeting the open frequencies with open standards while others operating
in licensed frequencies and offer proprietary solutions.

2.6 State of the art in LPWAN

LPWAN technologies are classified regarding the frequency spectrum they use, i.e., two basic
subcategories: licensed spectrum and unlicensed spectrum. Figure 2.9 depicts graphically
how the categories are made and includes the subcategories.
Cellular service providers use technologies that fall into the category of licensed spectrum.
Services offered by the mobile network operators are expensive and it would be counter-
productive to use them for IoT. LTE is the prevalent cellular technology and has a high data
throughput, but it is expensive and LTE-based devices consume much power. NB-IoT is the
new proposed technology for IoT applications in the licensed spectrum. It is promised to
be cheap and power efficient. In the following, we will have a look at technologies from
both categories, starting with the ones operating on unlicensed frequencies.

26
 2.6 State of the art in LPWAN

Figure 2.9: Classification of LPWAN.

2.6.1 Sigfox

Sigfox is a French company that provides communication solutions for IoT. It is Europe’s
leading company that provides IoT connectivity solutions based on LPWAN. Sigfox follows
the business model of a cellular service provider despite the fact that it operates in the
unlicensed frequency spectrum.

Ultra Narrow Band (UNB)

Sigfox uses ultra-narrow band (UNB) and binary phase-shift keying (BPSK) to achieve
the extended range of an LPWAN. In UNB communication, the spectral density (see
Section 2.1.3) of signal is concentrated in a narrow band of frequencies, the energy density
of noise is distributed along the whole spectrum instead of being concentrated, which
makes the signal in any narrow portion of spectrum always above the noise floor. The
drawback is interference if noise peaks exist in the communication band.
The UNB is 192 KHz wide between 868 and 868.2 MHz in Europe, and between 902 and
928 MHz in the rest of the world. Each message on the air occupies 100 Hz as shown in
Figure 2.10.
Transmission between an end device and the network is unsynchronized, the device
transmits data on a random frequency and then sends two additional copies of the same
message at another point in time on another frequency. This feature is known as "random
access" depicted in Figure 2.11. On the other end, base stations scan the whole 192 kHz
searching for UNB signals for demodulation [36].

27
2 Background

Figure 2.10: Sigfox band specification [35].

Sigfox offers "cooperative reception", which is a resource sharing technique that increases
the robustness of the network by using minimum redundancies. The network element
is not attached to one permanent base station; instead, the message is broadcasted and
picked up by several stations and filtered at the back end. On average, Sigfox offers three
base stations per device.

Figure 2.11: Random access [35].

Figure 2.12 shows the standard network architecture of the Sigfox network. The end devices
send data to base stations over the air using the UNB and BPSK modulation technique. The
base stations communicate with the backend via DSL and 3G. The backend handles all the
message processing, redundant messages are discarded, and only the relevant messages
are stored in the database. The web interface allows customers to access their messages by
using the browser or a REST API [35].

28
 2.6 State of the art in LPWAN
Figure 2.12: Sigfox network architecture [35].
29
2 Background

2.6.2 LoRa

LoRa is a new modulation technique that offers long-range communication. This modula-
tion technique can be described as a "frequency modulated (FM) chirp" or "Chirp spread
spectrum" [22, 30, 33]. Chirp spread spectrum has already been in use by the military
and space communication for decades because of its significant long range and robustness
against interferences. However, LoRa is the first commercial use of chirp spread spectrum.

LoRaWAN Protocol

LoRaWAN is developed by the LoRa Alliance, which is an open, non-profit organization.

LoRa is a link layer protocol and does not specify the network architecture. LoRaWAN, on
the other hand, is a Media Access Control (MAC) layer protocol which specifies the network
architecture. The protocol and network architecture are amongst the key elements that
govern battery lifetime of a node, the network capacity, the quality of service, the security,
and the variety of applications served by the network [22].

Figure 2.13: LoRa and LoRaWAN stack [22].

Mesh topology variants offer extended network range by using network elements as routers.
Extension of the range is as simple as adding more devices to the network where every
device acts as a repeater. This range extension adds complexity and processing overhead to
the network and also reduces the battery life and network capacity [22]. Zigbee [45] is a
prime example of such topology.
LoRaWAN offers "cooperative reception" where end nodes are not specified to one gateway.
Instead, a node transmits the data packet to multiple gateways as shown in Figure 2.14.
Each gateway then pushes packets to the server via Ethernet, Wi-Fi or some other backhaul.
Message filtering and storage is done at the server side, reducing complexity from the end
device and gateway.

30
 2.6 State of the art in LPWAN
Figure 2.14: LoRaWAN network topology [22].
31
2 Background

Cooperative reception is quite helpful in case of a mobile node; as the node does not need
to switch from gateway to gateway, it keeps on broadcasting the signal and its position can
be determined by monitoring the signal strength [22].

LoRaWAN Device Classes

LoRaWAN provides three device classes for various applications. These classes have trade-
offs between network downlink, communication latency, and battery life as shown in
Figure 2.15. Communication latency is crucial in a control application, where timely
control of specific actuators is required.
Bi-directional end-devices (Class A): Class A devices allow bi-directional communica-
tion. Up-link transmission of an end device is followed by two short downlink receive
windows. The Class A operation requires the lowest power as it requires downlink commu-
nication from the server only for a short interval after an uplink transmission. Downlink
communication will need to wait till the next up-link transmission in any other case.

Figure 2.15: LoRaWAN device classes [22].

Bi-directional end-devices with scheduled receive slots (Class B): Class B devices have
all the features of the class A devices, and they support the feature of extra receive windows
at scheduled times.
Bi-directional end-devices with maximal receive slots (Class C): End-devices of Class
C have almost continuously open receive windows, only closed when transmitting.

32
 2.6 State of the art in LPWAN

2.6.3 DASH7

DASH7 is an LPWAN communication standard. It offers extended range at low data rates
for monitoring applications. DASH7 is an open source standard, and it operates in the
sub-1 GHz frequency band (315 MHz and 915 MHz). The DASH7 Alliance is a nonprofit
organization that is responsible for maintaining the DASH7 protocol. The protocol, officially
known as the DASH7 Alliance Protocol (D7AP), has various versions, where version 1.0
is focused on wireless sensors and actuators. D7AP is a variant of the ISO/IEC 18000-7
standard [19] which was designed for 433 MHz band only, but D7AP extends it to all
sub-GHz Industrial Scientific and Medical (ISM) and Short Range Device (SRD) bands
(433, 868 and 915 MHz). These radio bands are reserved internationally for the use of
radio frequency (RF) energy for industrial, scientific and medical purposes other than
telecommunications
D7AP is an asynchronous network protocol with medium range when compared to other
LPWAN technologies, following a request-response style of communication. To decrease
complexity in the network, it uses a star topology, but to extend the range it allows a
maximum of one hop between the node and gateway. Nodes have the option to enforce the
acknowledgment from at least one gateway [11].
D7AP extends to all OSI-layers (see Section 2.2.1), from physical to application layer, as
shown in Figure 2.16. D7AP has its own file system with data elements and their properties,
such as permissions and storage classes. Some data elements define various parameters,
while others are for sensor data storage; these data elements can be managed via the
D7AP Application Layer Programming Interface (ALP). The ALP consists of commands that
perform actions such as read, write, execute, etc. These ALP commands are used by the
gateway or a sub-controller to manage and communicate with the end nodes [11].
The D7AP physical layer (PHY in Figure 2.16) supports 433, 868 and 915 MHz unlicensed
ISM/SRD bands with three channel classes, i.e., Lo-Rate, Normal and Hi-Rate, each with
a symbol rate of 9.6, 55.555 and 166.667 kbps respectively [10]. All the channels use
Gaussian frequency shift keying (GFSK) as the modulation technique, but with different
frequency deviations and channel spacings. Parameters, such as frequency band, data rate,
transmit power, coding parameters, and constraints, such as range, antenna size, cost, and
energy consumption, create a trade-off which depends on the application. For instance,
using the 868 MHz instead of the 433 MHz band will decrease the antenna size, but will
also affect negatively on the range as well [11].

33
34

2 Background
Figure 2.16: Excerpt of OSI layers based on [42].
 2.6 State of the art in LPWAN

D7AP Communication Schemes

For various application scenarios, D7AP provides three communication schemes. The
schemes are as follows:
Polling data using the D7AP Advertising Protocol: A gateway can query a node using
ALP commands, but the node is power constrained, it does not continuously scan for
requests. D7AP has a low-power wake-up mechanism called D7AP Advertising protocol
(D7AAdvP), which performs synchronization only when a request is queued [11].
Pushing data using the D7AP Action Protocol: By using the D7AP Action Protocol
(D7AActP), it is possible to push data instead of querying it, as pushing data is far more
efficient than querying. D7AActP allows the sensor to push data to the gateway, eliminating
the need for polling. This scheme is efficient for both, event-driven and periodic data
acquisition; it also provides an advantage in reduced power consumption and spectrum
usage [11].
Dormant Sessions: In use cases where nodes continuously push data to the gateway
(D7AActP), there might arise situations where data is to be transferred to the node, e.g.,
controlling an actuator. In a dormant session, the gateway queues the message with a
timeout instead of transmitting it directly. When the timeout is reached, the gateway
initiates a dialogue session. However, if the node initiates a dialogue with the gateway
before the timeout, then the ongoing dialogue will be extended making the gateway
requester and the endpoint responder. The concept of dormant sessions can be employed
to avoid using the relatively expensive D7AAdvP mechanism [11].

2.6.4 Weightless

Weightless [41] is an IoT LPWAN technology that operates in both licensed and unlicensed
spectrum. This technology is governed by Weightless Special Interest group (SIG), which
was formed in 2012 as a nonprofit organization to cater to all the requirements of the
weightless standard in the area of IoT. Weightless offers three different standards depending
on the application of use, i.e., Weightless-P, Weightless-N, and Weightless-W.
Weightless-P is two-way communications standard that uses Gaussian minimum shift keying
(GMSK) and quadrature phase shift keying (QPSK) modulation techniques. Weightless-P
is not suitable for wide area networks. The Weightless-W standard operates in the TV
white space (TVWS) spectrum. TVWS are the unused licensed frequencies that are made
available for unlicensed use, e.g., television broadcast. The regulations for TVWS change
with the region of operation making it unfit for an EMS. Weightless-N, like Sigfox, is an
ultra-narrow band system. It uses the same modulation technique as Sigfox, but instead of
being a closed system, Weightless-N is comprised of network of partners. [31].

35
2 Background

2.6.5 Ingenu RPMA

Ingenu is a wireless network provider, whose former name was On-Ramp wireless. Ingenu
provides machine to machine (M2M) communication solutions for IoT platforms. In the
LPWAN sector, Ingenu holds a proprietary technology which is called Random Phase
Multiple Access (RPMA).
RPMA is a cellular type technology that provides 300+ square miles per access point.
It achieves this significant coverage through increased transmission power and receiver
sensitivity. RPMA operates in 2.4 GHz unlicensed band with a bandwidth of 80 MHz. The
most prominent advantage of using 2.4 GHz is its availability in all the countries of the
world. RPMA uses the Viterbi algorithm for channel coding; this gives RPMA a Packet
Error Rate (PER) of 50%, in other words even if half of the packet containing the message
were lost the entire message would still be decoded [17]. Based on RPMA, Ingenu is
building a network known as the "Machine network" for M2M communication only. To
use the Machine Network users need to subscribe as with any other cellular network. This
technology looks very promising in the IoT domain, but the fact that it is proprietary creates
a dependency on Ingenue to deploy access points in the region of operation. Their network
only covers the U.S. for now. It is expected to take a few years till they extend their service
to European countries.

2.6.6 Licensed LPWAN

LPWAN technologies operating in the licensed spectrum are standardized and governed by
the 3rd Generation Partnership Project (3GPP). To capture the IoT market in the licensed
spectrum 3GPP has introduced three LPWAN standards, i.e., Extended Coverage GSM for
the Internet of Things (EC-GSM-IoT), Long Term Evolution Machine Type Communications
Category M1 (LTE MTC Cat M1, also referred to as LTE-M) and Narrowband IoT (NB-IoT).
3GPP represents 400 individual members; this broad support is a significant advantage
when it comes to standardization and interoperability between different mobile operators
and vendors [1].

2.6.7 Extended coverage GSM IoT (EC-GSM-IoT)

EC-GSM-IoT was designed to facilitate time to market by making it backward compatible;

it can work with the existing GSM networks. This backward compatibility allows an
easy resource sharing between EC-GSM-IoT and legacy network which enables a gradual
introduction of the technology, eliminating the need to reserve dedicated resources for IoT
[1].

36
 2.6 State of the art in LPWAN

2.6.8 Narrowband IoT

Narrowband IoT (NB-IoT) is a 3GPP release 13 feature. It uses the same principle and
building blocks as the LTE physical layer, this allows for rapid standardization and product
development. It has been designed inherently to have more coverage and consume less
power as compared to conventional GSM networks.

2.6.9 LTE-m

LTE MTC Cat M1 or more commonly known as the LTE-m is the successor of Cat-1 and
Cat-0. In 3GPP release 12, a Cat-0 UE (User equipment) was introduced with the aim to
reduce complexity as compared to GSM/GPRS mobile devices. The extended battery life
is achieved by use of power saving mode (PSM) and extended idle-mode Discontinuous
Reception (eDRX) [1].

37
3 Selection of an LPWAN

One of the tasks of this thesis is to select the most feasible LPWAN solution for the
application, i.e., an energy management system. Section 2.4 discusses the available LPWAN
technologies in detail. This chapter will focus on the evaluation of previously discussed
LPWAN technologies and selection of the most feasible technology for a prototypical
implementation of an EMS.

3.1 Selection Criteria

In order for the selection of an LPWAN technology, a selection criteria has to be defined.
The technologies will be evaluated on the bases of this criteria. The selection criteria is
tailored for our application and should not be used as a general rule for the evaluation of
an LPWAN technology. The following are the main selection criterion:

• Modulation technique,
• Data throughput,
• Operating frequency, and
• Range.

The modulation technique (see Section 2.1.5) determines the key characteristic of a
communication system. Techniques that provide inherent interference protection are
preferred. For instance, spread spectrum techniques are designed to be detected even
below the noise level and are preferred over the others. Data throughput should be enough
to transmit 100-500 bytes per 5 minutes. Operating frequency of the technology should be
in the unlicensed spectrum. Node power consumption should be as minimum as possible
with a range of a few kilometers. Low node power consumption is an inherent property of
an LPWAN technology therefore, it is not mentioned as a key criteria.
Apart from the aforementioned points, availability, cost, effort and support for the technol-
ogy are also important when it comes to implementation.

39
3 Selection of an LPWAN

3.2 Comparison

NB-IoT is a desirable solution regarding cost and area coverage, but when it comes to cost
per byte and longevity, then it is not viable, especially in applications like EMS. In Germany,
only Vodafone has recently started offering an NB-IoT solution, however, due to its high
dependency and cost it is not a feasible solution for now. NB-IoT operates in the licensed
spectrum making it unfit for an EMS.
Section 3.2 summarizes our findings with some values extracted from the work of Raza,
Kulkarni, and Sooriyabandara [32]. A comparison of the technologies can be made with
respect to modulation technique, range, data throughput and operating frequency. The
majority of the LPWAN technologies use an operating frequency in the SUB-GHz band
except for Ingenu using the 2.4 GHz. As mentioned in Chapter 2, SUB-GHz frequencies are
less prone to interferences and offer a more significant area coverage.
Sigfox uses UNB, as mentioned in Section 2.6.1. Data transfer rates are in between 100-600
bits per second depending on the region. Sigfox is well suited for low-bandwidth (less
than 300 bits per second / up to 12 total bytes per payload), and low transmit frequency
(up to 140 messages per day) applications. Sigfox is effective for communications from
endpoints to base stations (uploads), but it is not particularly effective from base stations to
endpoints (downloads) [8]. Sigfox offers excellent coverage compared to all other LPWAN
technologies. It follows a cellular network like business model, forcing users to buy service
and hardware. It is not allowed for the users to create custom private networks.
Weightless-N is governed by Nwave, which is an IoT solution provider. The requirement
for special hardware, such as temperature compensated crystal oscillator (TCXO) and the
relatively unbalanced link budget, puts Nwave at a disadvantage over the LoRa based
systems. The Weightless-W standard operates in the TV white space (TVWS) spectrum.
TVWS are the unused licensed frequencies that are made available for unlicensed use,
e.g., television broadcast. It is an advantage to use the unused VHF and UHF bands,
but the regulations for TVWS differ from region to region besides nodes are designed
specifically to a certain frequency. Designing nodes for a wider range of frequencies would
put a huge constraint on antenna construction and other parameters. RF systems are not
flexible to adapt to multiple frequencies with the same antenna. Weightless-P is a two-way
communications standard that uses a 12 kHz channel, making it reasonably narrow band.
Weightless-P is not suitable for wide area networks because the receiver sensitivity of 12
kHz minimum shift keying channel will not be able to match a narrow band Binary phase
shift keying (BPSK). Weightless-P caters to all the cons of Weightless-W/N, but at the cost of
less range. Aside from these, there is not much support available for weightless conforming
devices; and weightless based sensors are hard to find in the market [31].
Ingenu’s RPMA is the most attractive technology that exploits the free ISM 2.4 GHz band
for its data throughput, although this frequency band is more susceptible to interference
because a massive number of other devices also use this band, such as Wi-Fi devices.
This technology requires that the area of operation is covered by Ingenu’s RPMA access
point, adding a dependency on a third party. RPMA is a proprietary technology, and

40
 3.2 Comparison

Ingenu’s business model is that of a Mobile network provider. Apart from these downsides,
Ingenu is offering its services only in the U.S. for now, making it unsuitable for application
requirements of this thesis.
DASH7 offers an excellent energy-per-bit aspect and is ideal for sensor/actuator data, but
DASH7 lacks the range required for the application of EMS. Higher ranges are obtainable
at the cost of more energy per bit.
LoRaWAN uses the star topology. To achieve long-range communication, a gateway must
be capable of receiving data from a high number of end nodes. LoRaWAN offers "Adaptive
data rate" and "Multi-channel multi-modem transceiver" on the gateway, which allows
the reception of multiple messages simultaneously on multiple channels. The key factors
that affect a gateways capacity are the number of concurrent channels, data rate (time
on air), the payload length, and how often nodes transmit. LoRa is a "spread spectrum"
based modulation technique, where signals on different spreading factors do not interfere
with each other, hence making it possible for the gateway to receive multiple data rates
on the same channel simultaneously. If a node is closer to the gateway, it does not always
have to use the lowest data rate and occupy the link for a longer time. Instead, it could
use the highest possible data rate and keep the link free for a long time for other nodes.
The adaptive data rate can optimize the battery life of a node. Adaptive data rate requires
symmetrical uplink and downlink with sufficient downlink capacity, causing LoRaWAN to
have a high capacity and scaling capability of 6 to 8 times. This scaling capacity is unique
to LoRaWAN as other LPWANs have asymmetrical up and downlink. LoRaWAN nodes
communicate asynchronously, i.e., they only transmit when the data is ready to send, or an
event is triggered, this kind of protocol is called Aloha protocol. In synchronous networks
such as cellular networks, the nodes have to frequently synchronize with the network to get
messages, this synchronization is power intensive and reduces the battery life significantly.
LoRaWAN has shown to have 3 to 5 times the advantage, compared to all other LPWAN
technologies in the area of power consumption [22].

41
42

3 Selection of an LPWAN
Technology/Standard Modultion Range Data throughput Operating Frequency Channels

SUB-GHz ISM:
EU (433 MHz,
5 km (Urban), 0.3-37.5 kbps (LoRa),
LoRaWAN CSS 868 MHz), 10 in EU
15 km (Rural) 50 kbps (FSK)
US (915 MHz),
Asia (433 MHz)
SUB-GHz ISM:
UNB DBPSK (UL), 10 km (Urban), 100bps (UL),
Sigfox EU (868 MHz), 360
GFSK (DL) 50 km (Rural) 600bps (DL)
US (902 MHz)
40 1-MHz channels,
RPMA-DSSS (UL), 78 kbps (UL),
Ingenu 15 km ISM 2.4 GHz up to 1200 signals
CDMA (DL) 19.5 kbps (DL)
per channel
16-QAM, BPSK, TV white space
WEIGHTLESS-W 5 km 1-10 kbps 16 or 24 channels (UL)
QPSK, DBPSK 470-790 MHz
GMSK, SUB-GHz ISM or multiple 12.5 kHz
WEIGHTLESS-P 2 km 200 bps - 100 kbps
offset-QPSK Licensed channels
SUB-GHz ISM:
multiple 200 kHz
WEIGHTLESS-N UNB DBPSK 3 km 30-100 kbps EU (868 MHz),
channels
US (915 MHz)
SUB-GHz ISM: 3- different channel
433 MHz, types number
DASH7 GFSK 0-5 km (Urban) 9.6, 5.6, 66.7 kbps
868 MHz, depends on type
915 MHz and region

Table 3.1: Comparison of various LPWAN technologies and standards extracted from [10, 18, 22, 32, 35].
 3.3 Conclusion

LoRaWAN based on LoRa modulation has the range, data rate, and an open source mac
layer protocol (LoRaWAN) making it an excellent candidate. LoRa and Sigfox were tested
in a real-world environment by Nolan, Guibene, and Kelly [25]. They carried out their
experiments on the eastern seaboard of Ireland. They were able to cover an area of 1380
square kilometers with 90% coverage using a single LoRa base station. Table 3.2 taken
from [25] shows useful field results of the two LPWAN technologies. It is clear that LoRa
modulation requires significantly less transmit power, has an antenna sensitivity that is
better by a factor of 8, and a bigger payload size while employing open standard. Petajajarvi
et al. [26] also carried out field experiments on land and water in Oulu, Finland. Their
experiment focused on the coverage of mobile nodes based on LoRa technology. They were
able to achieve a packet loss ratio of 15% in the range of 2-5 km on land and 31% in the
range of 5-15 km on water.

Features LoRa (Semtech SX1272) SigFox (AXSEM)

symmetrical Technology Y Y
Uplink Data + ACK Data
Payload size 19-250 bytes 12 bytes
Protocol Overhead 12 bytes 26 bytes
TX power 13 dBm 14 dBm
TX consumption 28 mA 45 mA
RX consumption 10.5 mA 10 mA
Encryption AES-128 E2E AES-128
Open Standard Y N
Technology CSS/FSK/OOK/GMSK GFSK/BPSK
Sensitivity (dBm) -137 -129
Table 3.2: LoRa versus Sigfox comparison (table taken from [25]).

LoRaWAN comes with one drawback, i.e., the dwell time limitation. The time a signal
requires to be in the air is known as dwell time, in other words, it is the duration for which
the device occupies the channel for active communication. In the EU 863-870 MHz ISM
band, LoRaWAN limits the duty cycle to 1% for data, to avoid network congestion.

3.3 Conclusion

Every technology has some perks that others lack. It is always a trade-off between range,
data throughput, power consumption, cost and device availability. The selection of technol-
ogy is highly dependent on the application at hand. For instance Wi-Fi is easily available
and has a high data throughput, but cannot be used for wireless sensors because of its

43
3 Selection of an LPWAN

limited range and high power consumption, same holds for the conventional technologies
like ZigBee, Bluetooth, Z-Wave, etc.
This thesis focuses on an energy management system, where data acquisition through
sensors is the essential task. Technologies that are region independent and use unlicensed
spectrum are preferred. Technologies that are strictly proprietary e.g., (Sigfox, Ingenu’s
RPMA) and compel users to buy hardware and services (network) also pose a limitation to
the customer base. Enisyst offers its smart energy solution to customers throughout Europe,
and cannot afford to deploy technologies that are regional dependent or infrastructure
dependent. Using such technologies will add unnecessary dependencies.
Keeping in view the selection criteria LoRaWAN fits this application best. Though it uses
a proprietary modulation technique (LoRa); only Semtech produces chips that perform
LoRa modulation. It still is in the same category of cost as other LPWAN end devices.
Manufacturers, such as Microchip [24], are to produce LoRa based modules in the future,
further reducing the cost and eliminating monopoly of the manufacturer.

44
4 Concept

In this chapter, a standard energy management system will be discussed, which will be
used as a template for the implementation.

4.1 Energy Management System

An EMS that is designed specifically for buildings was formerly known as Building Energy
Management System (BEMS), today it is referd to as EMS. An EMS is defined as a computer
based system that monitors and controls a buildings mechanical and electrical equipment
[15]. EMS consists of software and hardware. Its hardware includes a network of sensors
and actuators, whereas the software includes algorithms for smart control and buses for
intercommunication. The aim of an EMS is to minimize energy waste by using it efficiently
to achieve monetary and environmental benefits. Figure 4.1 shows a block diagram of a
wired EMS. A standard EMS comprises of the following components:

• Central controller,
• External controller,
• Actuators,
• Sensors, and a
• Graphical User Interface (GUI).

These components are describe as follows.

4.1.1 Central controller

The central controller is where all the algorithms for smart control are implemented. It polls
the data from the sensors and controls the actuators accordingly. The central controller can
be accessed via a Graphical User Interface (GUI) for setting up various parameters.

4.1.2 External controller

The external controllers serve the purpose of Inputs/Outputs (I/O) extension as well as
repeaters or signal amplifiers.

45
4 Concept

4.1.3 Actuators

An actuator is an electromechanical component that is used for moving and controlling a

system. An actuator needs a control signal and a source of energy. In an EMS the actuators
are generally mounted on heat pumps, CHP generator, water pumps and valves. The
algorithm running on central controller is responsible for controlling the actuators.

Figure 4.1: Block diagram of an EMS extracted from [13].

4.1.4 Sensors

A sensor is a component that converts non electrical energy into electrical energy so that
it can be measured. The sensors generally used in an EMS are: temperature sensors,
humidity sensors, smoke detectors, carbon mono-oxide detectors, light sensors, and flow
measurement sensors etc., also various types of meters including electricity, gas and water
meters.

4.1.5 Graphical User Interface (GUI)

The GUI is used to access the Central controller. An EMS is designed to be flexible and can
be configured for different seasons and operating conditions. The GUI of an EMS provides
a platform where such changes can be made.

46
5 Prototypical Implementation of an
LPWAN based EMS

In this chapter, a prototypical implementation of an Energy Management System (EMS)

will be carried out using LoRa technology. A standard EMS is defined and discussed in
Chapter 4. Based on this standard, a prototype will be implemented.

5.1 Hardware Setup

This section will focus on the hardware required to set up a LoRa based EMS. All the
hardware for its implementation has to be industrial grade because Enisyst provides
industrial grade solutions to its customers. The LoRa based EMS is integrated with the
current Enisyst system and to reduce extra hardware, all the implementation is done on
the standard hardware used by the eni.os, which is the Enisyst’s backend responsible
for monitoring and control. The internal workings of eni.os are confidential and are
out of scope of this thesis. The transfer of sensor data to the eni.os will be discussed in
Section 5.2.2.

5.1.1 Enisyst 1613 module

Enisyst 1613 module uses Phytec’s phyCORE-AM3352 [28] System On Module (SOM).
This SOM is equipped with Texas Instruments’s Sitara [40] processor which is based on the
ARM Cortex-A8 [2] processor as shown in Figure 5.1.

Figure 5.1: phyCore AM3352.

47
5 Prototypical Implementation of an LPWAN based EMS

This processor supports Controller Area Network (CAN), Serial Peripheral Interface (SPI),
RS232, USB, and Ethernet. It is entirely capable of running embedded Linux on it as well.
The distribution of Linux currently running on phyCORE is created by "The Yocto Project"
[44], which will be discussed in Section 5.2.2.

5.1.2 phyBOARD

The phyBOARD-Regor is an extension board on which the phyCORE-AM3352 module

is mounted. This extension board is developed for use in standard DIN rail housings
(e.g. Bopla CombiNorm-Connect). As shown in Figure 5.2 it contains all the ports for

Figure 5.2: phyBOARD Regor.

the common interfaces such as 2x Ethernet, 2x LAN, CAN, RS485 and RS232. Also, the
phyBOARD-Regor can be extended via a 60-pin expansion bar.

5.1.3 LoRa Node

RFM95W as shown in Figure 5.4 is used as LoRa transceiver. It supports SPI protocol for
communication with other modules.

Figure 5.3: RFM95W LoRa transceiver.

48
 5.1 Hardware Setup

It is mounted on the breakout board for ease of use. The RFM95W costs less than 1 C and
offers the following key features:

• 168 dB maximum link budget,

• +20 dBm - 100 mW constant RF output vs. V supply,
• +14 dBm high-efficiency PA,
• Programmable bit rate up to 300 kbps, and
• High sensitivity: down to -148 dBm.

Figure 5.4: RFM95W mounted on the break out board.

It is connected with the 1611 extension board via the SPI to create a LoRa node.

5.1.4 DS18B20

DS18B20 is the standard temperature sensor used by Enisyst systems. Unlike the famous
PT1000, it is a digital temperature sensor that uses a 1-wire protocol for transmitting
information.

Figure 5.5: DS18B20 digital temperature sensor.

49
5 Prototypical Implementation of an LPWAN based EMS

5.1.5 1611 extension module

Enisyst’s 1611 is an extension module which is used to extend the I/Os for the Enisyst 1613
module (phyBOARD-Regor). 1611 is equipped with an ATMEL’s AVR [3] microcontroller
to communicate with the 1613 module via the RS232 interface.

Figure 5.6: 1611 extension module.

Figure 5.6 depicts a 1611 module. For the application of LoRa based EMS, 1611 is used
as a node; multiple DS18B20 temperature sensors are connected to the 1611 module. It
takes the inputs from the temperature sensors via the 1-wire protocol and sends it to the
RFM950W LoRa transceiver via the SPI.

5.1.6 LoRa concentrator board

For setting up a LoRaWAN network, it is required to have a LoRaWAN gateway. This

gateway works like any other gateway and receives the LoRa modulated signals from the
end nodes. These received packets are then forwarded to the back-end server where the
messages are stored and analyzed.

Figure 5.7: LoRa concentrator module ic880a.

LoRa concentrator ic880a, as shown in Figure 5.7, is capable of receiving LoRa modulated
signals on eight different channels simultaneously. This module communicates with 1613
module via the SPI protocol, and together they act as a LoRaWAN gateway. There are a few
LoRaWAN gateways readily available on the market, but due to specific requirements, a
custom LoRaWAN gateway was designed using the 1613 module and ic880a concentrator.

50
 5.2 System Architecture

5.2 System Architecture

This section will focus on the whole system architecture of an IoT based EMS. Figure 5.8
shows a top-level view of the entire system. A LoRa node sends temperature sensor values
to the gateway via LoRa modulated signals over the air, the signals are received at the
gateway-end, and are published to a message broker at the server side.

Figure 5.8: Top level view of the whole system.

The end user application can also monitor the values by subscribing to the specific topic.
The received values at the gateway-end are decrypted and decoded by a JavaScript (Js)
script. The script sends these values to eni.os via Unix domain socket. The values are
analyzed by eni.os, and respective actuators are activated. A detailed overview of the whole
system architecture is depicted by Figure 5.9. All the internal software modules and their
interconnecting communication protocols are also highlighted.
Figure 5.9 can be compared to the standard architecture shown in Figure 4.1. Due to
LoRa’s dwell time limitation, the wireless part of the EMS will be used just for monitoring,
whereas for controlling the actuators standard cable connection is used. The 1613 module
in Figure 5.9 is the central controller in our EMS, which receives the temperature sensor
data wirelessly via the LoRa modulated signal. The extension board 1611 can be compared
to the external controller in the standard architecture. The end user can access the Enisyst
EMS to set various parameters, including a timing schedule for heating and cooling.

5.2.1 Node

The node consists of DS18B20 temperature sensors, 1611 module, and the LoRa RFM95W
transceiver. As shown in Figure 5.10, the values are read from the sensors and sent to the
gateway via the uplink. To comply with the 1% rule of the ISM band and to always send
the most up-to-date data, a "scheduler" programme was implemented.

51
5 Prototypical Implementation of an LPWAN based EMS
Figure 5.9: Detailed system architecture.
52
 5.2 System Architecture

Figure 5.10: LoRa node.

The flow chart shown in Figure 5.11 depicts the sequence flow of a single loop of the
scheduler programme. In every loop, the scheduler reads all sensors and inputs, "check-
Values()" performs this operation. After that, it checks if the timer for SYNC message has
expired. The SYNC message is a flag for the timer that is used for sending the sensor data
periodically. An expired SYNC means that the sensor values should be sent right now while
resetting the SYNC. This mechanism is useful in case of environmental data acquisition.
It is not efficient to store real-time environmental data. Measuring values every 1 to 6
minutes is good enough for efficient heating control. After SYNC, the scheduler checks
if LoRa’s 1% rule is met, if it is true, the next HEX string for the uplink is assembled. If
not, this step is skipped, and the loop ends. At any given time in the loop of the scheduler
programme, it is possible that it can be interrupted to set an error code. If this happens,
the next uplink is an error message instead of a data message.
Since LoRaWAN is subject to the 1% rule, care must be taken to minimize the amount
of data to be transmitted. Therefore it will not use an ASCII-based approach, but a
hexadecimal approach. This means, data to be transmitted is in hexadecimal format. For
this purpose, a separate format was created, which is optimal for this application. The data
format essentially consists of two parts. The first two bytes describe the data type, whether
it is a error message or a data message. The bytes after represent the data. This structure is
shown in Table 5.1

2 Byte X Byte ...

assignment field data ...

Table 5.1: Structure of a HEX message of the 1611 node.

The first two bytes of each message is referred to as "assignment field." Here bits 1-15 are
used to define the type of data that is present in the message. Bit 0 denotes the mode bit,
i.e., when it is set, the message is an error message. For each sensor or input on the 1611
module, its respective bit shows if data is present in the message. For instance, if the bit 0 is
set and bit 1 is also set, this means an error message from the temperature sensor T1, and
if bit 0 is not set and T1 is set it means data message from T1. An error message is defined
to have a length of one byte; which means up to 255 different errors can be transmitted.
If the message is a data message, bit 0 is not set. A temperature value has the length of

53
5 Prototypical Implementation of an LPWAN based EMS

Figure 5.11: Scheduler flow chart.

2 bytes and is transmitted as a 16-bit signed integer. An analog input has the length of 2
bytes and is sent as 16-bit signed integer.

Figure 5.12: Structure of the first two bytes of a hex message.

An electricity meter value has a length of 3 bytes. The meter gives two values; the pulse
counter and the time between two pulses. The pulse counter value has a length of 1 byte
and is transmitted as an 8-bit unsigned integer, whereas the time between two pulses has
the length of 2 bytes and is a 16-bit unsigned integer. The detailed structure is shown in

54
 5.2 System Architecture

Figure 5.12. The identifier M stands for the mode of the message, data or error message.
T1 to T5 stands for the temperature values, AE1 and AE2 for the analog inputs and DI1 to
DI8 for the digital inputs.

Figure 5.13: Class diagram.

The implementation of the data layout was done on 1611 as a class; the class diagram is
shown in Figure 5.13. One complete message can have a maximum length of 51 bytes, and
the maximum transferable data at the highest spreading factor is also 51 bytes.

5.2.2 Gateway

A block diagram of the gateway is shown in Figure 5.14. The gateway is formed by several
software modules working together. Each of these modules is communicating with the
others via communication protocols such UDP, MQTT, and SPI. These software modules are
running on a custom distribution of embedded Linux version 4.4. The software modules,
i.e., Mosquitto server, nodejs script and LoRaWAN gateway can be run separately on
different hardware systems as well.

Enisyst 1613 software setup

The hardware specifications of the 1613 module were mentioned in Section 5.1.1. In this
section, the base software setup of 1613 will be discussed. The 1613 module supports
embedded Linux. The Linux distribution required for the implementation is custom created
via "The Yocto Project."
The Yocto Project is an open source collaboration project that creates custom Linux-based
systems for embedded products, regardless of the hardware architecture. The project
provides a flexible set of tools and a space where embedded developers worldwide can

55
5 Prototypical Implementation of an LPWAN based EMS

Figure 5.14: Internal software modules at the gateway end.

share technologies, software stacks, configurations and best practices which can be used to
create tailored Linux images for embedded devices [44].

Package name Required for

mc, nano, autoamake, git Debuging, on board compiling
openvpn, apache2 eni.os
stunnel, tunctl, tcpdump Network managment and security
spitools 1611, 1613, ic880a
nodejs, nodejs-npm Running Js scripts
mosquitto Message broker
erlang LoRa Gateway Bridge
python (pandas, matplotlib) Data analysis and viusalization

Table 5.2: Package dependencies.

The Yocto Project uses a build system known as "BitBake." It is a scheduler and execution
engine which parses instructions (recipes) and configuration data. It then creates a
dependency tree to order the compilation, schedules the compilation of the included code,
and finally, executes the building of the specified, custom Linux image (distribution).
BitBake is a make-like build tool. BitBake recipes define how a particular package is

56
 5.2 System Architecture

built. They include all the package dependencies, source code locations, configuration,
compilation, build, install and remove instructions. Recipes also store the metadata for the
package in standard variables. Related recipes are consolidated into a layer. During the
build process, dependencies are tracked and native or cross-compilation of the package is
performed [44]. To run all the software modules at the gateway end, a Linux distribution
with support for the packages shown in Table 5.2 was created.

Packet forwarder

The packet forwarder [9] is an open source program provided by Semtech. It is running on
the 1613 module, and it communicates with the concentrator board via the SPI protocol.
All the received RF packets are forwarded to LoRa Gateway Bridge via UDP. It uses the
following helper programmes:

• util_sink: It is a simplistic program listening on a UDP port for datagrams and displays
a notification message when a packet is received. The content of the datagram is not
displayed.
• util_ack: This program sends merely the acknowledgment of messages.
• util_tx_test: The network packet sender is a simple helper program used to send
packets through the gateway-to-server downlink route.

LoRa Gateway bridge

LoRa Gateway Bridge [5] is an open source service which abstracts the "packet forwarder"
UDP protocol into JSON over MQTT. It enables the use of MQTT for receiving data from
and sending data to the gateways. LoRa Gateway Bridge is part of the LoRa Server [6].
LoRa Server is an open-source LoRaWAN network-server, part of the LoRa Server Project
[7]. LoRa Gateway Bridge makes use of MQTT for publishing and receiving application
payloads. It is one of the requirements of LoRa Gateway Bridge to have an MQTT broker to
work, for this purpose Mosquitto is used.
As shown in Figure 5.15 taken from [5], the LoRa Gateway Bridge can run on the same
host as the packet forwarder or a separate host as well. Keeping in view the Enisyst system
the best possible scenario is to have all the software modules running on the same host.
Running all the modules on the same host will reduce the hardware cost and effort. The
current hardware used by the Enisyst system is fully capable of running all of the modules.
Using the modular approach makes the system scalable in case any of the modules need to
be deployed on different hardware because of resource management or any other reason.
All the modules communicate with each other via standard Internet protocols. The system
architecture will remain the same if they are deployed on different hosts.

57
5 Prototypical Implementation of an LPWAN based EMS

Figure 5.15: Network configuration using LoRa Gateway bridge and packet forwarder.

Mosquitto server

Mosquitto [21] is an open source message broker that uses the Message Queue Telemetry
Transport (MQTT) protocol versions 3.1 and 3.1.1. A message broker is a middleware that
acts as a mediator between applications. There are a few other open source MQTT message
brokers available today such as EMQ (Erlang MQTT Broker) [20], VerneMQ [12] and
RabbitMQ [38]. Amongst all these message brokers Mosquitto is the lightest. As described
before, the Enisyst system is running on an embedded computer, which makes Mosquitto
an ideal candidate for this application.
The publish/subscribe scheme used for machine to machine (m2m) communication has be-
come the defacto in the realm of IoT. The MQTT is a lightweight method of communication
between devices that do not have the computing resources to use complex communication
protocol.

58
 5.2 System Architecture

Nodejs script (Decoder)

The nodejs script creates a connection to the Mosquitto server. It does not make a difference
if the Mosquitto server is running on different hardware. In this case, the Mosquitto server
is running on the same hardware, so a connection to the local-host is made. In some cases,
the Enisyst system is deployed in complete isolation and no or limited internet connectivity
is available. For such cases, it is more suitable to run all the software modules on one
host. After establishing the connection, the script subscribes to the specific topics. The
topics have the same Unique ID as 1611 module. The received messages are decrypted,
and the 51-byte long hex string is decoded. The data is extracted from the string and sent
to eni.os via Unix domain socket. For the purpose of debugging, the script also establishes
a connection with a remote Mosquito server via the internet. It publishes the extracted
data to the server, which can be seen by any user who has subscribed to the correct topic.

eni.os

It is a proprietary software module owned by Enisyst. It is responsible for monitoring

and control. The internal workings of this module are strictly confidential and can not be
discussed in this document. For this thesis, this module will be considered as a black-box
that takes sensor values as inputs and transmits signals to actuators.

5.2.3 End user

The end user represents an interface where the data can be visualized. The visualization
tool is written in JavaScript and runs in a browser. It uses D3 (Data-Driven Documents) to
implement the line chart, bar graphs and scatter plot. D3 is a low-level JavaScript library.
Details of this tool are out of scope for this thesis.

59
6 Range Assessment

This chapter will be focused on our findings after implementing LoRa as a baseline technol-
ogy for communication between an end node and the central controller. The range tests
were performed in the industrial area of Pliezhausen, Baden Würtemburg, Germany. The
area is densely populated with various industry with more interference than residential
areas.

6.1 Range test

Figure 6.1 shows the paths over which the SNR values were recorded. SNR (see Sec-
tion 2.1.7) is a measure of the signal quality, the negative SNR shows that LoRa modulated
signals can be detected even if they are below the noise level. The orange lines in the
map denote the path over which SNR was recorded, every blue mark on the path is the
exact point where the value was recorded. The gateway was placed inside the office on the
1st floor like any other Wi-Fi router. The yellow lines in the map show the displacement
between the gateway and the peak points on the map.
The reason for the difference between distances amongst different paths is the uneven
terrain of the area. Figure 6.2 shows the total area that can be covered by one gateway.
The signal reception can be optimized even further by creating line of sight communication.
During the range tests, it was observed that, if there was no obstruction between the
gateway and the node the signal could reach even further. The empty pocket in Figure 6.2
is because a large metallic structure is obscuring the signals.

61
62

6 Range Assessment
Figure 6.1: Paths with SNR values.
 6.1 Range test
Figure 6.2: Area coverage.
63
7 Conclusion and Outlook

The ever-growing energy demand of toady’s industry and household can be reduced by
employing the efficient use of energy. Energy use can be optimized by employing energy
management systems. Enisyst offers its customers a reduction of 30% in energy use by
employing their EMS. Enisyst uses a wired sensor network for monitoring and a wired
network for controlling the actuators. This wired network requires a lot of effort and
time to deploy and in most cases, it can only be employed during the construction of the
building.
To tackle this problem it was suggested to use a wireless technology. By using a wireless
technology, the effort can be drastically reduced and the deployment of the EMS will no
longer be dependent on the construction. For this purpose, several technologies were
studied and it was decided to use an LPWAN technology for this task. The core concept
behind an LPWAN technology is to provide a significant long range with low power
consumption at the cost of lower data transfer rates. There are a lot of LPWAN technologies
available in the market right now, each offering various advantages and disadvantage. In
order to use the right LPWAN technology, a study about the current LPWAN technologies
was carried out. The focus of this study was to find an LPWAN technology that is suitable
for the task of implementing an EMS.
The study concluded in the selection of LoRa as the LPWAN technology to implement the
EMS. The dwell time limitation of LPWAN restricted us to use it only for the collection
of sensor data, and not for the controlling of actuators, therefore it was decided to limit
the LPWAN use for monitoring only. A prototypical implementation of the LPWAN based
EMS was carried out. In the prototype, a customized LoRaWAN gateway and a node
were implemented. The node transmitted LoRa modulated signals over the air which
were received by the gateway. At the gateway end, there are several software modules
running on an embedded computer. An open source message broker is used as a backend
for handling packets between software modules. Having a message broker provides the
flexibility and scalability required for an EMS. The software modules running at the gateway
end, decrypted and decoded the sensor values and passed them to Enisyst’s control software.
This software controlled the actuators correspondingly.
After the implementation of the system, several range tests were performed in an industrial
area which is densely populated with buildings. The gateway was placed inside a building
on the first floor. The maximum range at which the signal was transmitted was 360 meters.
It was observed that if the line of sight was maintained even more range could be attainable.
Due to the uneven mountainous terrain, it was not possible to achieve more in our case.

65
7 Conclusion and Outlook

LoRa modulation is not limited by the distance but the obstructions in the path of the signal.
A range of more than 10 km has been achieved using the line of sight communication.
The findings from this thesis will be used to create the next generation Enisyst EMS. Further
research is required in order to use LPWAN for the controlling of the actuators as well.
Multiple technologies can be used in combination to further optimize the process. As of
today, there are not a lot of ready-made LPWAN solutions. We had to create our own
sensor node from scratch. Various providers claim to have LPWAN sensor nodes but the
availability of such products are said to be at the end of 2018.

66
Bibliography

[1] 3GPP. “White Paper: 3GPP Low Power Wide Area Technologies.” Tech. rep. 3GPP,
2016 (cit. on pp. 36, 37).
[2] ARM. “ARM Cortex A8.” 2018. URL: https://developer.arm.com/products/processor
s/cortex-a/cortex-a8 (cit. on p. 47).
[3] Atmel. “AVR atmega32.” 2018. URL: https://www.microchip.com/wwwproducts/
en/ATmega32 (cit. on p. 50).
[4] Bluetooth Special Interest Group. “Bluetooth.” 2017. URL: https://www.bluetooth.
com/ (cit. on pp. 12, 21).
[5] O. Brocaar. “LoRa gateway bridge.” 2018. URL: https://github.com/brocaar/lora-
gateway-bridge (cit. on p. 57).
[6] O. Brocaar. “LoRa Server.” 2018. URL: https://github.com/brocaar/loraserver
(cit. on p. 57).
[7] O. Brocaar. “lora server porject.” URL: https://www.loraserver.io/ (cit. on p. 57).
[8] I. Brown. “A Detailed Breakdown of LPWAN Technologies and Providers.” URL: http:
//web.luxresearchinc.com/hubfs/Insight_Breakdown_of_LPWAN_Technologies.
pdf (cit. on p. 40).
[9] M. Coracin. “packet forwarder.” 2017. URL: https://github.com/Lora-net/packet_
forwarder (cit. on p. 57).
[10] DASH7 Alliance. “DASH7 Alliance Wireless Sensor and Actuator Network Protocol
v1.0.” 2015. URL: http://www.dash7-alliance.org (cit. on pp. 33, 42).
[11] G. Ergeerts, M. Nikodem, D. Subotic, T. Surmacz, B. Wojciechowski, P. De Meu-
lenaere, M. Weyn. “DASH7 alliance protocol in monitoring applications.” In: P2P,
Parallel, Grid, Cloud and Internet Computing (3PGCIC), 2015 10th International
Conference on. IEEE. 2015, pp. 623–628 (cit. on pp. 33, 35).
[12] Erlio GmbH. “VerneMQ.” URL: https://vernemq.com/ (cit. on p. 58).
[13] D. Estates. “Building energy management systems.” In: Design and Maintenance
Guide 22 (2001) (cit. on p. 46).
[14] Wi-Fi Alliance. “Wi-Fi.” 2017. URL: https://www.wi-fi.org/ (cit. on pp. 12, 21).
[15] D. A. Al-Hammad. “BUILDING MANAGEMENT SYSTEM (BMS).” URL: http://faculty.
kfupm.edu.sa/ARE/amhammad/ARE- 457- course- web/Building- Management-
System.pdf (cit. on p. 45).
[16] Ingenu. “Ingenu.” 2018. URL: https://www.ingenu.com/ (cit. on p. 26).

67
Bibliography

[17] Ingenu. “White Paper: How RPMA works.” Tech. rep. ingenu, 2018 (cit. on p. 36).
[18] Ingenu. “White Paper: RPMA Technology for the Internet of Things.” Tech. rep.
ingenu, 2018 (cit. on p. 42).
[19] “Information technology – Radio frequency identification for item management –
Part 7: Parameters for active air interface communications at 433 MHz.” Standard.
Geneva, CH: International Organization for Standardization, Sept. 2014 (cit. on
p. 33).
[20] F. Lee. “EMQ.” 2002. URL: http://emqtt.io/ (cit. on p. 58).
[21] R. A. Light. “Mosquitto: server and client implementation of the MQTT protocol.”
In: The Journal of Open Source Software 2.13 (May 2017), p. 265. ISSN: 2475-9066.
DOI : https://doi.org/10.21105/joss.00265 (cit. on p. 58).

[22] LoRa Alliance. “A technical overview of LoRa and LoRaWAN.” In: White paper, Nov
(2015) (cit. on pp. 30–32, 41, 42).
[23] LoRa Alliance. “LoRa Alliance.” 2017. URL: https://www.lora-alliance.org/ (cit. on
pp. 12, 26).
[24] Microchip. “Microchip.” 2018. URL: http://www.microchip.com/ (cit. on p. 44).
[25] K. E. Nolan, W. Guibene, M. Y. Kelly. “An evaluation of low power wide area network
technologies for the Internet of Things.” In: 2016 International Wireless Communi-
cations and Mobile Computing Conference (IWCMC). Sept. 2016, pp. 439–444. DOI:
10.1109/IWCMC.2016.7577098 (cit. on p. 43).
[26] J. Petajajarvi, K. Mikhaylov, A. Roivainen, T. Hanninen, M. Pettissalo. “On the
coverage of LPWANs: range evaluation and channel attenuation model for LoRa
technology.” In: 2015 14th International Conference on ITS Telecommunications
(ITST). Dec. 2015, pp. 55–59. DOI: 10.1109/ITST.2015.7377400 (cit. on p. 43).
[27] J. Petajajarvi, K. Mikhaylov, A. Roivainen, T. Hanninen, M. Pettissalo. “On the
coverage of LPWANs: range evaluation and channel attenuation model for LoRa
technology.” In: ITS Telecommunications (ITST), 2015 14th International Conference
on. IEEE. 2015, pp. 55–59 (cit. on p. 12).
[28] Phytec. “phyCore AM335x overview.” URL: http://www.phytec.de/produkt/system-
on-modules/phycoream335x/#uebersicht (cit. on p. 47).
[29] B. Ray. “The Past, Present, and Future of LPWAN.” 2017. URL: https://www.link-
labs.com/blog/past-present-future-lpwan (cit. on pp. 25, 26).
[30] B. Ray. “What is LoRa?” 2015. URL: https://www.link-labs.com/blog/what-is-lora
(cit. on p. 30).
[31] B. Ray. “What is Weightless?” 2015. URL: https://www.link-labs.com/blog/what-is-
weightless (cit. on pp. 35, 40).
[32] U. Raza, P. Kulkarni, M. Sooriyabandara. “Low Power Wide Area Networks: An
Overview.” In: IEEE Communications Surveys Tutorials 19.2 (Secondquarter 2017),
pp. 855–873. DOI: 10.1109/COMST.2017.2652320 (cit. on pp. 40, 42).

68
[33] F. Sforza. “Communications system.” US Patent 8,406,275. Mar. 2013. URL: https:
//www.google.com/patents/US8406275 (cit. on p. 30).
[34] Sigfox. “Sigfox.” 2018. URL: https://www.sigfox.com/en (cit. on pp. 12, 26).
[35] Sigfox. “Sigfox Technical Overview.” 2017. URL: https://www.disk91.com/wp-
content/uploads/2017/05/4967675830228422064.pdf (cit. on pp. 28, 29, 42).
[36] Sigfox. “Sigfox Technology Overview.” URL: https://www.sigfox.com/en/sigfox-iot-
technology-overview (cit. on p. 27).
[37] J. Sliva. “Technologies used in wireless sensor networks.” In: Systems, Signals and
Image Processing, 2008. IWSSIP 2008. 15th International Conference on. IEEE. 2008,
pp. 77–80 (cit. on p. 12).
[38] P. Software. “RabbitMQ.” 2007. URL: https://www.rabbitmq.com/ (cit. on p. 58).
[39] TEOCO Corporation. “A Short History of LPWAN, and Why the IoT Can’t Live Without
It.” 2017. URL: http://www.teoco.com/latest/news/blog/short-history-lpwan/
(cit. on p. 25).
[40] Texas Instrument. “TI Sitara Overview.” 2017. URL: http://www.ti.com/processors/
sitara/arm-cortex-a8/am335x/overview.html (cit. on p. 47).
[41] Weightless SIG. “Weightless.” 2015. URL: http://www.weightless.org/ (cit. on pp. 12,
35).
[42] M. Weyn, G. Ergeerts, L. Wante, C. Vercauteren, P. Hellinckx. “Survey of the DASH7
alliance protocol for 433 MHz wireless sensor communication.” In: International
Journal of Distributed Sensor Networks 2013 (2013) (cit. on p. 34).
[43] S. Yinbiao, K. Lee, P. Lanctot, F. Jianbin, H. Hao, B. Chow, J.-P. Desbenoit. “Internet
of things: wireless sensor networks.” In: White Paper, International Electrotechnical
Commission, http://www. iec. ch (2014) (cit. on p. 11).
[44] Yocto Project. “The yocto project.” 2018. URL: https : / / www . yoctoproject . org/
(cit. on pp. 48, 56, 57).
[45] Zigbee Alliance. “Zigbee.” 2017. URL: http://www.zigbee.org/ (cit. on pp. 12, 30).
[46] H. Zimmermann. “OSI Reference Model - The ISO Model of Architecture for Open
Systems Interconnection.” In: IEEE Transactions on Communications 28.4 (Apr. 1980),
pp. 425–432. ISSN: 0090-6778. DOI: 10.1109/TCOM.1980.1094702 (cit. on pp. 19,
20).

All links were last followed on April 17, 2018.

Declaration

I hereby declare that the work presented in this thesis is

entirely my own and that I did not use any other sources and
references than the listed ones. I have marked all direct or
indirect statements from other sources contained therein as
quotations. Neither this work nor significant parts of it were
part of another examination procedure. I have not published
this work in whole or in part before. The electronic copy is
consistent with all submitted copies.

place, date, signature