Communication protocols evaluation based on energy models of an IoT application

(1)

Institut für Datentechnik und Kommunikationsnetze Institut für Datentechnik und Kommunikationsnetze

Master’s Thesis

Communication protocols evaluation based on energy

models of an IoT application

Simon Philipp

October 29, 2021

Institut für Datentechnik und Kommunikationsnetze Prof. Dr. techn. Admela Jukan

Supervisors:

Prof. Paavo Rasilo Dr. Jasenka Dizdarevi´c

Dr. David Blaževi´c

(2)

��

(3)

��

Foreword

This Master’s thesis entitled "Communication protocols evaluation based on energy models of an IoT application" was written at the Technical University of Braunschweig in Ger- many and the University of Tampere in Finland. The master’s thesis was carried out as a final project for my degree in Information Systems Engineering.

First of all, I would like to thank Professor Admela Jukan and Professor Paavo Rasilo for making it possible for me to do my Master’s thesis in Finland and for supporting me at all times. They helped me a lot not only with the subject matter but also with the organi- zation.

Furthermore, I would like to thank David Blaževi´c, who always supported me in Tampere in the laboratory and during the tests, which made my time in Tampere an unforgettable time. I would also like to thank Jasenka Dizdarevi´c, as she accompanied me during the writing of my thesis and supported me a lot during the writing process. I am very happy about the 6 months I spent in Finland and hope that at a later stage I will have the opportunity to come back for a longer period and work in animal welfare.

Last but not least, I would like to thank my family and my girlfriend for their support throughout my studies and especially during my master’s thesis.

(4)

Statement of Originality

This thesis has been performed independently with the support of my supervisor/s. To the best of the author’s knowledge, this thesis contains no material previously published or written by another person except where due reference is made in the text.

Braunschweig, October 29, 2021

(5)

Abstract

This master thesis describes the process of constructing an IoT device for cows. In this thesis, an energy harvester is developed with intent to transmit enough energy to send a signal. The main focus is on the creation of the energy harvester at the foot of the cow. This is designed with the help of a simulation and then independently designed and assembled.

This achieves2.4mW at a load of5000Wand a frequency of7Hzand an amplitude of2V.

In the course of this work, it will be possible to send and receive a BLE beacon signal with the harvested energy. The signal transmission is first tested in the laboratory and then in a field trial on a cow.

(6)

(7)

List of Figures

�.�. Happy cow Tampere . . . 1

�.�. Energy is stored, the system can use it when energy is needed . . . 4

�.�. Energy is not stored, the system can just boot if the EH harvest enough energy . . . 4

�.�. Operation Cycle [3] . . . 5

�.�. Operation modes [3] . . . 5

�.�. Network herd localization [18] . . . 7

�.�. Cow tracking triangulation . . . 8

�.�. Villard circuit . . . 10

�.�. Villard output . . . 10

�.�. Additive technology [21] . . . 11

�.��. Tampere Fablab EH printing process . . . 11

�.��. Mass Spring Damper . . . 11

�.��. Induction law . . . 12

�.�. Concept leg EH . . . 14

�.�. Result of leg EH . . . 14

�.�. Concept of a single DOF EH . . . 15

�.�. Cow leg acceleration from the simulation from Tampere University . . . . 16

�.�. FFT leg data . . . 17

�.�. Sensitivity coil height and width . . . 18

�.�. Sensitivity number of turns . . . 18

�.�. Sensitivity load resistance . . . 18

�.�. Sensitivity coil distance from equilibrium . . . 18

�.��. Sensitivity distance between magnet and coil . . . 18

�.��. Sensitivity spring magnet distance . . . 18

�.��. Sensitivity spring magnet radius . . . 18

�.��. Sensitivity spring magnet height . . . 18

�.��. Sensitivity of the output power to system parameters . . . 18

�.��. Simulated resultant force and displacement . . . 19

�.��. EH with Parameter . . . 20

�.��. Energy harvester: Tube . . . 22

�.��. Energy harvester: Ring . . . 22

�.��. Energy harvester: Ending part . . . 23

�.��. Energy harvester: Magnet holder . . . 23

�.��. Energy harvester . . . 24

(10)

� L�� F��

�.��. Printed EH . . . 24

�.��. Printing process . . . 24

�.��. Energy harvester spring magnet calibration . . . 25

�.��. Electrical elements [35] . . . 26

�.��. ConceptLTC3588. . . 27

�.��. Storage Capacitor [25] . . . 27

�.��. NRF52840dongle . . . 27

�.��.ESP8266andRaspberrypi4connection . . . 28

�.��. Simulation of electronic components . . . 28

�.��. Circuit diagram . . . 30

�.��. Software concept . . . 31

�.��. Cover inside view . . . 32

�.��. Cover outside view . . . 32

�.��. Software Concepts . . . 34

�.��. Acceleration sensor holder . . . 35

�.��. Circuit diagram . . . 35

�.��. Simulation voltage divider . . . 36

�.��. Simulation voltage divider output . . . 36

�.��. NRF52840holder . . . 37

�.��. Communication test circuit diagram . . . 38

�.��. Voltage output of the EH from human steps in blue and time span when voltage is applied to the communication module in orange . . . 39

�.�. Measurements setup . . . 41

�.�. Measurements setup shaker with energy harvester . . . 41

�.�. Frequency behavior of di�erent configuration. The configuration is described in section 4.2 . . . 43

�.�. Spring const diagram . . . 44

�.�. Mechanical damping diagram . . . 45

�.�. Voltage output . . . 46

�.�. Power output. Configuration is described in 4.3 . . . 46

�.�. MQTT message current consumption . . . 48

�.�. BLE beacon power consumption one execution time 18mW . . . 49

�.��. BLE beacon power consumption signal 0.038mW . . . 50

�.�. Pinja . . . 51

�.�. Leg with energy harvester . . . 51

�.�. Field test data. Acceleration in orange. Voltage output EH in blue . . . 52

�.�. Field test one step. Orange is the acceleration and blue is the voltage output of the energy harvester . . . 53

(11)

L�� F��

�.�. Cow herd . . . 54

�.�. EH . . . 54

�.�. Wireshark BLE Beacon recorded . . . 54

�.�. Screenshot BLE Beacon signal . . . 54

�.�. Wireless communication technology [22] . . . 56

�.��. MQTT Packet [15] . . . 57

(12)

(13)

�. Introduction

In this introduction, the first section 1.1 presents the motivation for the topic. The topic is briefly introduced and the reasons why this work is necessary. Then, in section 1.2, the problem and the objectives of this work are explained in more detail. Finally, section 1.3 describes the structure of this Master’s thesis.

�.�. Motivation

Animal products are very popular in the European Union. That is why, according to the European Commission, there were 21 million cows in the EU in 2018, each producing 7000 kg of milk [1]. To preserve the well-being of these cows, animal welfare concepts are needed. The concept of five freedoms is one of these concepts. It was published in 1979 by the British Farm Animal Welfare Council and identifies the following five freedoms [2].

Figure�.�.: Happy cow Tampere

Freedom from hunger and thirstThe animals have access to fresh water and healthy and nutritious food.

Freedom from discomfort caused by husbandryThe animals have suitable accom- modation such as a shelter in the pasture, dry runs and soft lying areas.

Freedom from pain, injury and diseaseThe health and integrity of the animals is also maintained by preventive measures, sick and injured animals are cared for by appropriate treatment, amputations are avoided or the animals are anesthetized during such procedures.

Freedom from fear and stressFear and stress are avoided through good handling of the animals and suitable housing conditions.

(14)

�

Freedom from normal behavioral patternsThe animals have the opportunity to ex- ercise their species-appropriate behavior (normal behavior), e.g. by having enough space, not being tethered and having contact with the outdoor climate.

In order to improve well-being, Internet of Things (IoT) devices are now being developed to measure the state of health and well-being and thus contribute to improving liv- ing conditions. The applications for IoT Devices in the field of animal welfare are almost limitless. A battery or accumulator is usually installed to supply these devices with power.

Since this makes the devices dependent on maintenance, as batteries have to be replaced and rechargeable batteries have to be charged, the solution is not very e�ective. Also because batteries are products that have to be disposed of once they have been discharged, which also pollutes the environment. Therefore, a promising solution would be to convert existing energy into electrical energy in order to become independent of batteries and accumulators. This process for obtaining small amounts of energy is called Energy Harvesting (EH). Therefore, in the course of this work, an IoT device will be constructed that can be used for cows.

�.�. Objective of the scientific work

In recent years and decades, IoT systems based on energy harvesting have been developed again and again. However, this has mostly been designed for humans, as this is the most obvious use case. Since we humans are better at expressing our emotions, it would only make sense to create such an IoT device for farm animals to monitor their health. So far, there is little research on energy harvesting with animals. Therefore, it is necessary to lay the foundations in this area and find out what is possible. In this thesis, an IoT device with energy harvesting will be created to show that it is also possible to operate energy harvesting with animals and to send data.

�.�. Outline

First, in chapter 2, the currentstate of the arton IoT devices, energy harvesting and communication is presented. This chapter also deals with the theoretical foundations. Next follows the chapter 3 on system designof the IoT device, in which the energy harvester, the electronic components, the software and the structure of the field tests are described.

Afterwards, the energy harvester is measured in the chapterTest bed4. In theperformance evaluationchapter 5, the two field tests are described. Finally, the results arediscussed and concludedin chapter 6.

(15)

�. State of the art

The following section describes the current state of the art. First, the basics of Internet of Things applications are in section 2.1. Then the current communication technologies and protocols are described in section 2.2. Since this is very important for this work, the basics of energy harvesting are in section 2.3. Last but not least, the theoretical foundations that are important for this master’s thesis are described in 2.4.

�.�. Internet of Things Applications

Internet of Things (IoT) is the networking of devices that send/receive data to other devices or servers and measure quantities in the environment into which they are placed and these quantities can the be used to improve or simplify everyday life. The devices usually have a wireless internet connection, but wired internet connections are also possible. The applications are almost limitless. Some examples of IoT applications are automation, health monitoring, smart cities and as used in this master thesis as a monitoring system for the agricultural sector. A general IoT process can be described as follows: the devices first collect data, which is then analyzed and with this analyzed data it is possible to change the environment. The Third Generation Partnership Project (3GPP) has defined 4 di�erent families of applications for the IoT.

Wearable devices such as remote health monitoring. Battery life here should be dependent on maintainability. For example, a smartwatch that can be charged every night does not need a large battery just to extend the runtime.

Environmental monitoring in real time for tracking industrial plants, for example.

High coverage high coverage for e.g. smart parking or control of industrial ma- chines.

Usually main poweredsuch as home appliances and smart city lighting which are stationary powered.

IoT devices usually consist of 4 components. One is the power management, this man- ages the energy from power sources such as batteries or energy harvesters. In a few cases, no energy storage is needed and the generated energy is directly forwarded. Secondly, there is the input/output subsystem for sensor data. In most cases, this subsystem deals with the collection of sensor data, such as temperature sensors. The data is then pro- cessed by the next component, the processor. Here, arithmetic operations take place that pre-process the data. These are then sent via the fourth component, the network module.

To illustrate this, the following diagrams were created to show the rough structure of IoT

(16)

�

applications in a simplified way. The signal flow and power flow were also shown here in a simplified way with arrows. The first image 2.1 shows an IoT device that temporarily stores the electrical energy generated in a capacitor. In this way, the electrical energy is consumed at a di�erent time than it is generated or the power management system waits until a certain minimum of energy is available. The second picture 2.2 describes a system where the generated electrical energy is directly consumed by the energy harvester. This would simplify the system considerably, as the power management unit and the capacitor are no longer required. However, this is very di�cult to implement, as the energy has to be generated at the moment it is needed [3].

Sensor

MCU Network Modul

Processor Network

Input Power

Power flow Signal flow

Input data Energy

Internet

EH PMU Super Capacitor

Figure�.�.: Energy is stored, the system can use it when energy is needed

Sensor

Energy Harvester

MCU Network Modul

Processor Network

Input Power

Power flow Signal flow

Input data

Energy Internet

Figure�.�.: Energy is not stored, the system can just boot if the EH harvest enough energy

The following guidelines for IoT devices can be derived from the paper [25]. If possible, the microcontroller should have extremely low power consumption in a range ofnAand, if possible, an ultra-low power sleep mode. In addition, the duty cycle should be as small as possible. Furthermore, no large computing operations should be carried out.

For energy production, it should be clear exactly how much energy is needed and how much energy can be harvested. The environmental influences play a very important role if one is to use energy harvesting and move away from batteries. In addition, a power management unit (PMU) should be designed so that it can convert the inputs and also output the voltage output appropriately for the downstream load.

For communication, it is very important to choose the right technology. This means it is important to clarify what range and data throughput is needed to reduce energy.

Operation cycle

The following image shows the operating cycle that describes the application running on an IoT device. There are two states of the application: the power-saving t_sleep mode and the power-intensivet_activemode. The Duty cycleDis the frequency of activity and is described by the following equation 2.1 [3].

D= ^t^active

t_active+t_sleep (2.1)

(17)

�.�. I�� T��A��

Wake up Sensor operations Network operations Actuator operations

Sleep

Figure�.�.: Operation Cycle [3]

Operation modes

The following figure 2.4 shows 4 di�erent operation modes.

0 1

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20

Time-triggered

0 1

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20

Event-triggered

0 1

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20

Always-on device

0 1

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20

Event-blocking

!!"#$%&

!'(&&)

" "

"

!!"#$%&

!'(&&)

Figure�.�.: Operation modes [3]

TheTime triggeredmode is a periodic wake up. The execution of operations is timed in this mode. This means that the operation takes place after the set time, so it is possible to calculate the exact energy demand of the IoT application.

Inevent triggeredmode, the device is activated in response to a specific external event in order to read out sensor data and perform an operation. Problem is if the event occurs to often, the IoT device may have not harvest enough energy to send the data and the data will be lost. An example is an earthquake detection device.

As the name suggests, theAlways onmode is always on and is not switched to sleep mode. A lot of energy is required for the always high power to continuously read sensor data, carry out operations and transfer data. Power consumption is constant but very high.

Event blockingmode is a special case. If an event occurs, the IoT device listens for a response in the active mode. Only when the response reaches the IoT device will it revert

(18)

�

to sleep mode. This operating mode has increased power consumption, since it is not clear how long the IoT device must be in active mode until a response is received [3].

�.�. Communication

This section deals with communication for IoT. First, the wireless technologies for IoT are described and compared in section 2.2.1 . The communication protocols are then described and compared in section 2.2.2. Finally, common communication strategies are explained in more detail in section 2.2.3 .

�.�.�. Wireless technologies

Zigbee

For agriculture, ZigBee protocol is well suited, as it allows the sensor nodes to communi- cate with a range of up to 100 meters. It is possible to cover a larger area with more router nodes. The power consumption is also very low. [22]

Bluetooth Low Energy (BLE)

Bluetooth Low Energy is particularly well suited for places where little energy is needed.

Since it has a limited range of about 10 meters, it is mostly used in mobile devices to transmit data such as temperature or weather data. [22]

WiFi

WiFi is one of the most widely used communication technologies. Pretty much every portable device is WiFi compatible. The communication technology has a range of 100m and is expandable with routers. In the agricultural sector, WiFi is used to transmit data on soil temperatures and soil moisture. WiFi consumes comparatively high power, has a long communication time and is suitable for transmitting large amounts of data. [22]

GPRS/3G/4G/5G Technology

For cell phones there is the General Packet Radio Service that is a packet oriented communication technology. The technology has a long range and high power consumption.

In agriculture, these technologies are already used to check sensor data for water quality, for example. [22]

Long Range Radio (LoRa)

LoRa was developed for long distances and consumes only comparatively small amounts of energy. Communication takes place via end devices, gateway and server. Thanks to

(19)

�.�. C��

its long range and low data rate, IoT is already being used for data transmission in kiwi production, for example. [22]

SigiFox

SigFox is a cellular communication technology with low data rates. In IoT, this has been used to locate animals in mountain pastures throughout the summer. Due to the low data transmission SigFox also consumes low energy. [22]

�.�.�. Protocol

Message Queue Telemetry Transport (MQTT)

The network architecture of MQTT is based on the components Publisher, Broker and Subscriber. Participating devices can subscribe to a topic or publish a message. The protocol is suitable for unreliable connections or connections with low bandwidth. MQTT is based on the TCP protocol. [19]

�.�.�. Communication strategies

A concept for stove localization is described below. It consists of a primary node and a secondary node. The figure 2.5 shows an overview of the network architecture.

Figure�.�.: Network herd localization [18]

The figure 2.5 shows the interaction of secondary nodes and primary nodes. These are both IoT devices that have been bound to animals. They are connected via a secondary connection. Since the communication of the secondary nodes does not have to reach a high range to send the data to the primary nodes, they can run with less energy than the primary nodes. It would therefore be possible to operate the secondary nodes using only kinetic energy, so that no battery is required. Data can be sent via beacon signals [18].

(20)

�

�.�.�. RFID Path tracking

To enable path tracking of the cow, so-called tags are needed to record the paths. These tags can be uniquely identified and their path can be detected and recorded by other electronics positioned in the field. With this method, the paths of individual cows can be very precisely recorded and evaluated. The disadvantage of this tracking method is that the cow must carry an RFID tag. This tag is attached to the cow’s leg or neck, for example.

Of course, the data can be falsified by attaching a tag, since this leads to a di�erent behavior of the cow. The second disadvantage is the infrastructure that must be placed in a field. One possibility to implement RFID path tracking would be to use BLE beacons.

The big advantage would be a significantly lower energy consumption since the position determination can be performed on the infrastructure and not on the RFID tag or mobile device.

Mobile station (X,Y)

Base station (X2,Y2) Base station

(X3,Y3)

Base station (X1,Y1)

r1

r2 r3

Figure�.�.: Cow tracking triangulation

�.�. Energy harvester

Energy harvesting is one of the most promising concepts when it comes to reducing battery waste. Therefore, energy harvesting has become more and more interesting for research and industry in the last decades. The principle is very simple: an attempt is made to convert potential and kinetic ambient energy mostly into small scale electrical energy (uW - mw) su�cent for powering of IoT technologies. Probably the best known example is the conversion of motion energy to set the hands of a wristwatch in motion. The energy is stored mechanically source.

The following table 2.1 presents a good overview of energy harvesting principles and how much electrical energy can be harvested with them. This table refers to humans and not to farm animals. However, these can be used to get a first overview and a first classification of what is possible with energy harvesting.

(21)

�.�. T��

Energy source Power Density Advantages Disadvantages

Solar energy (PV Panel)

15-100 mW/cm²

Available during daytime high output voltage

Not available in night e�ciency is low during cloudy days Wind energy

(Anemometer)

1200

mWh/day Available in open areas Not available in closed areas Finger Motion

(Piezoelectric) 2.1mW Available whenever needed Energy is harvested only when finger is moved Footfalls

(Piezoelectric) 5W Available whenever needed Highly variable output Thermal Energy

(Thermocouple) 50mW/cm² Long life Low energy conversion

e�ciency Motion

(Piezoelectric) 200µW/cm² Light weight Highly variable output Breathing

(Ratchet-flywheel) 0.42W Available all the time Di�cult to install Radio frequency

(Rectennas) 1µW/cm² Allow mobility Low power density

Table�.�.: Energy harvester concepts [20]

Positve sensing

There are several sensors that generate their own voltage, such as the thermocouple for measuring temperatures. When these sensors generate more electrical energy than the system consumes, this is called positive sensing. For sensing, the signal must be input via an Analog digital converter (ADC). This process requires the signal acquisition power.

The acquisition performance ratio (APR) is defined with the following formula APR= ^P^harvester

Pacquisition (2.2)

The formula can be used to describe di�erent sensing categories. If the APR value is less than 1, more energy is consumed than produced by the sensor. It is negative sensing.

Neutral sensing is when the APR value is exactly 1. If the energy harvester generates more energy than is needed, the APR is greater than 1. The principle is described in more detail in the paper [28].

�.�. Theoretical foundations

In this section, the theoretical foundations relevant to the thesis are explained. The Villard circuit is described first 2.4.2. Secondly, the additive technologies 2.4.2 are described. Then Mass spring damper system 2.4.3 is described and finally the inductive law 2.4.4.

(22)

��

�.�.�. Villard circuit

The Villard circuit 2.7 is a voltage doubler and represents a basic circuit and which consists of a capacitor C and a diode D, as shown in the adjacent circuit diagram 2.7. This works with an AC sourceUE. The Villard circuit represents a clamping circuit with peak clamping: after a few periods, the capacitor is charged to the peak value of the AC voltage supplied by the transformer on the secondary side. The voltage supplied by the transformer is shown in the voltage diagram 2.10 in blue V(n001). After a few periods, the output voltageU_Aoscillates between0Vand twice the value of the AC voltage on the secondary side as shown in greenV(002)in the time diagram 2.8 [4] . With several Villard

U_E D U_A

C

Figure�.�.: Villard circuit

0.0s 0.5s 1.0s 1.5s 2.0s 2.5s 3.0s 3.5s 4.0s 4.5s 5.0s V

2 1 -

V 9 -

V 6 -

V 3 -

V 0

V 3

V 6

V 9

V 2 1

V 5 1

V 8 1

V 1

2 V(n002) V(n001)

Figure�.�.: Villard output

circuits in a row and diodes turned in the other direction, the output voltage is smoothed to a DC voltage that increases with each pulse. The maximum DC voltage is then the mul- tiple of the number of Villard circuits. This voltage doubler was chosen for this thesis because it is particularly easy to implement and understand. It is needed as soon as the output voltage of the energy harvester is too low to work with.

�.�.�. Additive technology

Since during this thesis a large part of the parts will be produced with the additive process, it is important to describe the process in the following. In order to understand the additive manufacturing process, it is useful to first describe the additive manufacturing process chain. The first step in the process chain is to generate CAD data. This is the modeled component. The data is then prepared in the slicer before the machine (3D printer) is prepared in the third step. The slicer converts the surface structure of a model (stere- olithography file) into machine sequence code for the 3D printer. In the fourth step, the actual building process or additive manufacturing takes place. In the fifth step of the component chain, the removal and post-processing takes place. Before the component is finished, the finishing process takes place, e.g. painting or filing the component. At the end of this chain is the finished component.

(23)

�.�. T��

Y-Axis

X-Axis

FDM-Head

Extruder

Support Extruder

Baseplate Tabletop

Z-Axis Filament

roll

Figure�.�.: Additive technology [21] Figure�.��.: Tampere Fablab EH printing process

The picture 2.9 shows the construction of a 3D printer for the fused deposition mod- eling process. The picture 2.10 shows the printer printing a component.In this work, the materials polylactic acid (PLA) and polyethylene terephthalate (PETG) are used, whose properties are briefly summarized here.

PLAis made from corn starch and is industrially biodegradable. It is food safe, but has moderate temperature and weather resistance.

PETG is glycol-modified polyethylene terephthalate. Transparency possible, also food safe and more stable than PLA.

�.�.�. Mass spring damper system

The following figure 2.11 shows a mass damper system.mdefines the mass,dstands for the damping constant andk defines the spring constant. It can be seen that this is a 1 Degree of Freedom (DOF) system, since the mass can only move inxdirection.

d

k

y(t) x(t)

z(t) m

Figure�.��.: Mass Spring Damper

The picture 2.11 of the system shows that there is a displacement of the frame of the system which is given byy(t). Since the magnet in the centre is connected to the frame

(24)

��

via a spring and a damper, the displacement over the time of the magnet is given asx(t). To indicate a displacement of the two points relative to each other at a specific time, the relative displacementz(t)is used. The following formula 2.3 is used to calculatez(t). [23]

z(t) =x(t) y(t) (2.3)

�.�.�. Induction law

When a permanent magnet is moved through a coil at a speedv, a voltageU_indis induced.

Uind(t)

v(t)

Figure�.��.: Induction law

U_ind=v⇤^B⇤^l ^(2.4)

If there is more than one turn, N equal voltages are induced in N parallel, closely spaced conductors of a coil.

U_ind =N⇤^v⇤^B⇤^l ^(2.5)

U_ind= N⇤^df_dt ^(2.6)

(25)

�. System design

In this chapter the system design is explained. Since the energy harvester is the funda- mentals of the IoT device, it is described in detail here in section 3.1. The second section 3.1 is the system design of the electronics. A developed software concept is explained in section 3.3. Last but not least in section 3.4, the structure for the two field tests is described.

�.�. Energy harvester

This section explains the approach to designing the energy harvester. In the first subsection 3.1.1, the system is analyzed and then a concept is developed that serves as the basis for the energy harvester. With this basis, the energy harvester is first simulated in subsection 3.1.2 and then designed in subsection 3.1.3.

�.�.�. Concepts

In order to find a concept for the energy harvester, which is described in this subsection, the system at hand must first be briefly analyzed. The decisive factors that describe the system are briefly listed below.

External influances: Time the cow is outdoor and weather conditions Cow behaviour:Movement profile and cow noise level

It is di�cult to find values for the individual factors, as they vary greatly. For example, cows on di�erent farms are let out into the fields at di�erent times and for di�erent periods of time. Nevertheless, 4 di�erent concepts for energy harvesting can be derived from the factors, which are listed in the following table.

EH Concepts Description Output Weight

Solar Photovoltaic e�ect Outdoors:15mW/cm² 0, 54g/cm²[5]

Thermal Seebeck e�ect Human:100µW/cm³ no data Acoustic Helmoltz e�ect 1.436mW/cm²at123dB no data Movement EM induction Elephant:1.03mW[35] 745g[35]

Table�.�.: Energy harvester concepts [27]

Since the cow is probably outside for a long time, a lot of solar energy can also be used by converting solar energy or artificial light into electricity. In addition to the high energy

(26)

��

production, the modular construction is also an advantage in order to be able to scale up or down. The low weight is also an advantage here. On the other hand, there is the high dependence on external influences such as weather conditions or temperature in the stable. This is also a major disadvantage of energy generation through the Seebeck e�ect.

Here, a voltage is generated by a small gradient between the environment temperature and the body temperature. The guiding formula is the higher the di�erence, the higher the energy output. Cows have a body temperature of38to39degrees [6], which means that there is only a small di�erence with the normal outside temperature and therefore only a small energy output. It would also be possible to gain energy from the cow’s sounds, with louder sounds gaining more energy. The cow has a sound level of52to79dB[34] and would therefore have a very low energy output. It would be possible to increase this by using a cowbell. Furthermore, kinetic energy could be converted into electrical energy. There are two types of energy harvesting that are most commonly used for this. One is to use a piezoelectric ceramics that is deformed by the movement and thus causes a displacement of charge within the crystal. This system would be small, robust and easy to configure.

Although the use of piezo ceramics is well established, they are impossible repair and must oscillate at resonant frequencies for maximum output. This frequency range is mostly in high frequencies. The second frequently employed method of converting kinetic energy into electrical energy is electromagnetic induction, by moving a magnet through a coil or a coil through a magnetic field. As this is the most mature process of energy conversion and the also easier to build and implement, an electromagnetic energy harvester will be developed in the course of the master thesis.

The idea is to find a part of the cow that moves the most and disturbs the cow the least.

The following body parts would be suitable for an energy harvester that converts kinetic energy.

Tail:high acceleration, di�cult to attach

Ear:moves a lot, indicator for cow health [6], di�cult to attach Neck:Moves a lot, easy to attach, low displacement

Leg:High acceleration, easy to attach

Figure�.�.: Concept leg EH Figure�.�.: Result of leg EH

(27)

�.�. E��

According to the formula 2.5 from the chapter Induction law 2.4.4, there is a relation between speed and induced voltage, so a high acceleration would be good. Due to the high acceleration, the leg would be best suited for an energy harvester. The concept drawing 3.1 should make this clear.

Since there is one main direction of the leg, up and down, as on the concept drawing, an electromagnetic single degree of freedom harvester will be built in the process of this work. The basic idea is to induce a voltage in a coil with the help of a moving magnet. So, as shown in the picture 3.3, there is a coil, a magnet and magnetic springs that hold the magnet at zero point when there is no kinetic energy. The following picture 3.3 shows the first concept drawing.

Figure�.�.: Concept of a single DOF EH

P Description

z Displacement to the equilibrium point

h Coil height

w Coil width

N Number of windings

R_L Load resistance

a Spring magnet height

b Spring magnet width

d Distance spring magnet and magnet y Acceleration a�ecting the system

c Coil distance

Table�.�.: Parameter list of the concept

The system concept consists of a tube that serves as a guide for a round movable magnet.

This movable magnet should be kept in the centre of the tube as far as possible. This should be done by two magnets at the beginning and end of the tube. In order to indicate

(28)

��

a voltage by moving the movable magnet, two coils are wound around the tube. As a result, incoming vibrations at the tube cause the moving magnet inside to oscillate. The theoretical principles of this spring-mass-damper system have already been described in section 2.4.3 and are applied here. In addition, the theoretical principles of the induction 2.4.4 law are required here.

�.�.�. Simulation

The use case was concretized through the concept development. It should be an energy harvester that can be worn on the legs of cows. The first parameters that play a role in the development were already presented in the concept development. Besides these, it is also important that the design has no negative e�ects on the cow due to its size and weight. In order to take the parameters into account, the EH is first simulated in this subsection and then all parameters are interpreted. The simulation used was developed and provided by the University of Tampere. In the following the thesis deals with the acceleration data of the leg, as this has the highest acceleration. The data was provided by the University of Tampere. Figure 3.4 shows the acceleration over the time. It can be clearly seen that after a short phase between200msand800msof lifting the foot, the foot is lowered with a high negativ acceleration at800ms. The acceleration ranges from positive18m/s²to negative 18m/s².

0 100 200 300 400 500 600 700 800

Time [ms]

-20 -15 -10 -5 0 5 10 15 20

acceleration [m/s2]

Figure�.�.: Cow leg acceleration from the simulation from Tampere University

A Fast Fourier Transformation (FFT) is used to filter out the frequency components that are in the signal. Figure 3.5 shows The FFT output.

The main frequencies in the cow footstep are with flatting intensity1.28205Hz,3.84616Hz and7.69231Hz. These parameters are necessary for the design of the energy harvester.

The main conclusion that can be drawn from the graph is that a step has many small frequency components. Therefore, the parameters must be designed so that the energy

(29)

�.�. E��

0 5 10 15 20 25 30 35 40 45 50

frequency [Hz]

0 1 2 3 4 5 6 7 8

signal magnitude

X 1.28205 Y 7.04345

X 3.84616 Y 3.85966

X 7.69231 Y 2.32103

Figure�.�.: FFT leg data

harvester has the most power output at low frequencies. To generate as much energy as possible, it makes sense to let the energy harvester oscillate in the resonance frequency.

It is possible to exchange several parameters and to display the energy output. In the following an energy harvester is designed and the results are discussed and checked for plausibility. From the results, a suitable energy harvester can be derived. The parameters used for this first analysis are shown in the table 3.1.2. The following table 3.1.2 shows the

P Value Unit Description

N 6600 - Number of turns of wire in a coil

d 10 mm Coil height and width

c 30 mm Coil distance from equilibrium g 2 mm Distance between magnet and coil

d 1 mm Spring magnet distance

a 6 mm Spring magnet height

b 2 mm Spring magnet radius

R 1000 Ohm Load resistance

Table�.�.: Sensitivity system parameters

sensitivity of output voltage to system parameter. For this purpose, afor loopwas imple- mented in the simulation to change one parameter at a time. The limits of thisfor loop were set so that they are as realistic as possible.

The result of changing the coil height and width in the simulation to show how much energy is in relation to it is shown in picture 3.1.2. When changing the coil height and width, it can be seen that the optimum point for the parameters is9mm. Before9mm, the internal resistance of the coil is increased. Since the magnet only moves within a specific

(30)

��

0.005 0.006 0.007 0.008 0.009 0.01 0.011 0.012 0.013 0.014 0.015

Coil height and width 1.2

1.3 1.4 1.5 1.6 1.7 1.8 1.9 2 2.1

Power (mW)

Figure�.�.: Sensitivity coil height and width

1000 2000 3000 4000 5000 6000 7000 8000 9000 10000

Number of turns per coil 0

0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2

Power (mW)

Figure�.�.: Sensitivity number of turns

200 400 600 800 1000 1200 1400 1600 1800 2000

Load resistance 1

1.1 1.2 1.3 1.4 1.5 1.6 1.7 1.8 1.9 2

Power (mW)

Figure�.�.: Sensitivity load resistance

0 0.01 0.02 0.03 0.04 0.05 0.06

Coil distance from equilibrium 0

0.5 1 1.5 2 2.5

Power (mW)

Figure�.�.: Sensitivity coil distance from equilibrium

1 1.5 2 2.5 3 3.5 4 4.5 5

Distance between magnet and coil (m) 10^-3

1 1.2 1.4 1.6 1.8 2 2.2 2.4 2.6

Power (mW)

Figure�.��.: Sensitivity distance between magnet and coil

0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08 0.09 0.1

Spring magnet distance (m) 0.5

1 1.5 2 2.5 3

Power (mW)

Figure�.��.: Sensitivity spring magnet distance

0 0.002 0.004 0.006 0.008 0.01 0.012 0.014 0.016 0.018 0.02

Spring magnet radius (m) 0.5

1 1.5 2 2.5 3

Power (mW)

Figure�.��.: Sensitivity spring magnet radius

0 0.002 0.004 0.006 0.008 0.01 0.012 0.014 0.016 0.018 0.02

Spring magnet height (m) 0.5

1 1.5 2 2.5 3

Power (mW)

Figure�.��.: Sensitivity spring magnet height Figure�.��.: Sensitivity of the output power to system parameters

(m)

(m) Ohm

(31)

�.�. E��

range, it will no longer pass through some of the windings once the coil reaches a certain size. This parameter is therefore strongly dependent on the displacement that the magnet can achieve.

The picture 3.1.2 shows that the coils have their maximum power point at6600turns in total. This means that the more windings, the higher the energy output, but only up to a certain point. If the internal resistance of the coils becomes too high, the power output also decreases.

With the parameters, the picture 3.1.2 shows the load resistor reaches its maximum energy output at1000W. At the beginning, this rises quickly and then decreases slightly and constantly from1000Wonwards.

Picture 3.1.2 shows the displacement of the coil changes related to the energy output The diagram 3.1.2 shows that the gap between the coil and the magnet and the energy output are almost linear to each other. This means that the wall of the energy harvester must be as thin as possible in order to have the highest possible energy output.

The last three diagrams 3.1.2, 3.1.2 and 3.1.2 do not change. This means that the parameters are either irrelevant to the energy output or have not yet been taken into account in the simulation. The second is assumed to be the case.

The lesson learned from the analysis is that coil height and width, number of turns per coil and load resistance have a maximum that must be determined. Furthermore, the coil equilibrium can be defined as coil height divided by two. In addition, it can be stated that the distance between the coil and the moving magnet is one of the most important factors to increase the energy output.

Since the simulation does not support the spring magnets, the following section describes a method of dimensioning the magnetic springs by hand. In order to nevertheless calculate the value of the spring constant k with the simulation, the force to displacement at each point of the tube is calculated with the simulation. In this case, the values from the table 3.4 were taken. The result of the force needed to move the movable magnet is shown in the following figure 3.15.

Figure�.��.: Simulated resultant force and displacement

(32)

��

The figure shows that the relation between force to move the magnet and position in the tube is nonlinear. Also the picture 3.15 shows that the closer the moving magnet ap- proaches the ends of the tube and thus the spring magnets, the higher the force required to hold the magnet in position. In the diagram there is a range from 10mmto 5mm which is approximately linear. The slope is equal to the spring constantkand is445N/m in the simulation.

The system resonates at a natural frequency f₀is given by following equation 3.1.

f₀= ¹ 2p

r k

m (3.1)

By transforming the equation and using the mass values from table 3.4 and the f₀ = 7.69231Hzthe spring constant must be56N/m. To reduce the spring constant parameter it is necessary to use smaller magnets or reduce it with a permeability layer on the spring magnet.

From this and from the theoretical principles of electromagnetic, partial parameters for the energy harvester can now be derived. Since the simulation with the spring magnets did not show expected results, these are designed in the prototype. Therefore, variable mounts are needed for the spring magnets.

SMh

SMd

T^h

MMd

MMh

Cd

G Td

Figure�.��.: EH with Parameter

The simulation of the derived parameters in table 3.4 for one step resulted in an av- erage power of0.94167mWwith a calculated weight of131.0924g. Therefore, the energy harvester is to be built with the simulated parameters in the following.

(33)

�.�. E��

P Value Unit Description

T_h 40 mm Tube length

N 3000 - Number of turns in total C_d 10 mm Coil height and width

c 5 mm Coil distance equilibrium

d 0.1 mm Coil wire diameter

G 0.8 mm Gab size

d 20 mm Spring magnet distance

SM_h 2 mm Spring magnet height

SM_d 6 mm Spring magnet diameter

R 1000 Ohm Load resistance

MM_h 10 mm Moving magnet height

MM_d 10 mm Moving magnet diameter

m 24 g Inertial mass

w N42 - Magnets grade

Table�.�.: EH Parameter

�.�.�. Constructing

The development environment that was used to construct the energy harvester is Fusion 360 from Autodesk. After a short test phase, the environment has proven to be particularly clear and easy to learn

This section deals with the energy harvester to be modeled. The individual components are described first. The assembly consists of a total of 4 di�erent components. At the end the finished model is presented. All components can be found in the project folder as a common format and therefore all exact dimensions can be extracted from it. These are not dealt with specifically here.

Tube

The tube is used to keep the moving magnet on the Z axis. For this it is important that the tube has a low coe�cient of friction. To enable this using3Dprinting, it is necessary to divide the pipe along the z-axis. This makes it possible to rework the material inside the tube. One of these 2 halves is shown in the following picture 3.17. The inner radius is10.30mm with a thickness of0.50mm. That means that there is a gap of air between the tube and the moving magnet from0.30mm. There are 4 enlarged contact areas and 16 points to attach and align the rings and end pieces. The height is40.00mm.

(34)

��

Figure�.��.: Energy harvester: Tube

Ring

The ring has two functions. On the one hand, it serves to hold the two tubes together. On the other hand, it is used to delimit the coils. At least 3 rings are required for an energy harvester The following picture 3.18 shows the ring.

The inner radius of the ring is10.80mm. That means that the gap between moving magnet

Figure�.��.: Energy harvester: Ring

and coil will not be larger than0.80mm. In order to have the highest possible conversion of kinetic energy into electrical energy, this gap must be as small as possible, as can be seen in table 3.1.2. Due to manufacturing tolerances in 3D printing, this is the smallest possible distance. In addition, extra recesses were built into the ring to guide the wires of the coil.

Ending part

The end piece shown in the following picture 3.19 has three functions.

The first function is to hold the tube together. The inner radius is 10.80mm and is 2.00mmdeep. The second function cannot be seen in the picture, there is an adapter on the back to connect the energy harvester to a shaker for experimental tests. The third function is to hold the magnet holder. A cylinder is sunk in the middle. The inner radius of the cylinder is6.10mmwith a depth of2.00mm. There are also three holes in the bottom of the cylinder that are intended to remove the magnet holder from the end piece. There is a short and a long version of the end part. The di�erence is5mmand is intended for

(35)

�.�. E��

Figure�.��.: Energy harvester: Ending part

changing the magnets.

Spring magnet holder

The magnet holder is used to hold the magnet in place. The height and the inner cylinder are variable. The inner cylinder radius needs a0.05mmgab between the magnet and the holder to get the magnet in the holder and fixe it there without gluing. The following picture 3.20 shows the holder for a magnet with3.00mmradius and2.00mmhight. The hole in the middle is designed to get the magnet out of the holder.

Figure�.��.: Energy harvester: Magnet holder

Composite assembly

The following picture shows the energy harvester with magnets and without coils.

It can be seen on the picture 3.21, the component consists of 10 parts to be printed and 3 magnets. In the middle there are two gaps for winding the coil.

Printed Components

The printed result is presented in this section and in picture 3.22. The printer and the printing process were described in section 2.4.2. Tolerances were placed on the components to facilitate the movement of the magnet, but not too large tolerances to prevent

(36)

��

Figure�.��.: Energy harvester Figure�.��.: Printed EH

the magnet from spinning or jamming. To keep the friction as low as possible, the tube is printed as shown in the picture 3.23.

Printed surface low friction because the

direction of movement has the same direction as the layers (z-axis)

high friction because the direction of movement would cross the layers

Figure�.��.: Printing process

After the printing process, the components looked like figure 3.22. The parts were re- worked after printing. First, the tube where sanded down very finely in order to achieve significantly less friction. This was important because, despite the fine printing process, small edges were created with each layer height. The tube were then coated with Teflon.

This is used in many sectors where a low coe�cient of friction is required.

Coils

The coils are placed at the point with the strongest flow, according to the state of the art.

With the values in the table, these are the points +5mm and -5mm from the equilibrium point. Through the simulation we assume 1500 turns for one coil. The winding of the coil was done with the FZ-180 from the company QIPANG [7]. The internal resistance of the two coils totals515W. With the following formula it is now possible to determine the length of the line.

r= ^R⇤^q

l (3.2)

The specific resistancerof copper is assumed, which is0.01786W⇤^mm²^/mat 20 degrees.

The coil wire diameter is0.1mm, therefore the wire cross-sectionqis0.00784mm². This

(37)

�.�. E��

results in the calculated wire length of226.47m.

Summary

The energy harvester has a size of30mmx30mmx56mmand a weight with the magnets and windings of54.22g. As the rings stabilize the energy harvester very well, only 3 rings are needed, so that one ring in the middle is eliminated. The material price is listed in the following table 3.5.

Description Quantity Price Price

Moving magnet 1 0, 35EUR 0, 35EUR

Spring magnet 1 2, 80EUR 2, 80EUR

Tube 18g 20, 56EUR/1000g 0, 37EUR Wire 226, 47m 6, 88EUR/856m 1, 82EUR

- - - 5, 34EUR

Table�.�.: Energy harvester material price

The first measurements showed that the magnet does not levitate in the middle as shown in the picture and therefore does not generate the maximum energy when oscil- lating. This was adjusted by adjusting the bottom magnet. This was achieved by adjusting both coils as shown in the following picture 3.24 . This is how the correct spacing and size

Figure�.��.: Energy harvester spring magnet calibration of the spring magnet was designed.

(38)

��

�.�. Electronic Parts

The following describes the electronic parts that are required. The energy harvester created in the previous picture can be represented with the following equivalent circuit 3.25.

R_iis the internal resistance,Lis the inductance of the coil,R_Lis the load resistance,irep- resents the current andadescribes the electromechanical coupling coe�cient. In the first

L

R_i R_L

⍺ẏ i

Figure�.��.: Electrical elements [35]

subsection 3.2.1, a concept for sending a signal is created, then the concept is simulated 3.2.2 and a circuit diagram is derived from it in 3.2.3.

�.�.�. Concepts

A power management unit is required for the energy supply. This should rectify, store and output the energy in a regulated manner. A special feature is that the input voltage is an alternating voltage. The module must therefore work with an alternating voltage. One of these modules that supports this is theLTC3588. The Output Voltage is selectable of 1.8V,2.5V,3.3V,3.6V. Since most electronic components are operated at3.3V, it is used below3.3V. The following picture 3.26 shows the conceptual structure of the electronics. TheLTC3588first smoothes the AC voltage. The smoothed voltage is then applied to the storage capacitor, charging it. The load resistor simulates the communication module. Furthermore, the storage capacitor which has to be designed to store enough energy.

The graphic 3.27 was designed to show how the storage capacitor behaves over time. For execution of a transmission, it is assumed that10mWis needed.

The stored energy available at execution time can be calculated with the following formula 3.3. The valuesV_ON=5.05VandV_OFF=3.67Vare taken from the data sheet [8].

E= ¹

2 ⇤^C⇤(V_ON² V_OFF² ) (3.3)

With a capacity of2200µF , an energy of 13.23696mJ is calculated and thus a power of 13.23696mW for one second with equation 3.4.

P= ^E

t(s) ^(3.4)

(39)

�.�. E��P��

PZ1 PZ2

CAP

Vin SW

Vout

Vin2

D1 D0

Pgood

GND U1

LTC3588-1 C1 1µ

C4 47µ L1 10µ EnergyHarvester

LoadRes

C5 1µ StorageCap

Figure�.��.: ConceptLTC3588

VON

VOFF

Time

Storage Capacitor Voltage

Execution Time

Figure�.��.: Storage Capacitor [25]

NRF52840

TheNRF52840dongle [9] was selected as the communication module for Bluetooth Low Energy. The power supply range is1, 7Vto5, 5V from USB or external. The USB inter- face was used for programming. To connect theNRF52840as LoadRes, this must be done via the external power supply VDD OUT. Please note that SB2 has to be cut on the back and SB1 has to be soldered together. Otherwise, theNRF52840dongle may break. The microcontroller used is the energy-saving Arm® CortexTM-M4, which is programmable via NRF software. The IDE SEGGER Embedded Studio was used for the software development.

Figure�.��.:NRF52840dongle

ESP8266

To analyze Wifi technology and MQTT protocol, and whether it is possible, anESP8266 was chosen. This is easy to program. The NodeMCU V3 was used with the connection of a Raspberry Pi 4. The following electronics was used to create the measurement setup shown in picutre 3.29 for a MQTT message via Wifi.

(40)

��

ESP8266

Publisher Raspberry pi 4

Broker publish

message

Figure�.��.:ESP8266andRaspberrypi4connection

�.�.�. Simulation

The following picture shows the simulation of the electrical components with LTspice [16]. The most important components here are the energy harvester, the load resistor and the storage capacitor with voltage protection and voltage divider. To simulate the energy harvester, an AC voltage source with a frequency of7.69Hzand an amplitude of2V was used. In addition, an internal resistance of the voltage was set to1000Win the simulation to limit the current.

PZ1 PZ2

CAP

Vin SW

Vout

Vin2

D1 D0

Pgood

GND U1

LTC3588-1 C1

1µ

C2 47µ L1 10µ EnergyHarvester

SINE(0 2 7.69 0 0 0 0)

C3 4.7µ C4 2.2m

StorageCap 2.2m

D1 D

D2 D D3

D

D4 D C6 2.2m

C7 2.2m

C8 2.2m

Load 200 R2

4500000

R3 4500000 D5

BZX84B6V2L

D6 D D7

D C9 2.2m

C10 2.2m

ADC .tran 1000

Figure�.��.: Simulation of electronic components

If a voltage below5.05Vis assume for the energy harvester and thus at the inputsPZ1 andPZ2, the storage capacitor will not be charged enough. Therefore, a 3-fold Villard circuit was simulated here. This can be recognized by the Schottky diodes and the capacitors, which are rated at2200µF.

The storage capacitor is also rated at2200µF. Since the size of the capacitor is relevant and this is related to the maximum voltage of the capacitor, it is rated at6.3V. To protect it from voltage peaks, a Zener diode is added parallel to the capacitor, which has a breakdown voltage of6.2V. In addition, a voltage divider is added in parallel to the storage capacitor, which can then be used to interpret the voltage via an analogue digital converter.

Communication protocols evaluation based on energy models of an IoT application

Institut für Datentechnik und Kommunikationsnetze Institut für Datentechnik und Kommunikationsnetze

Master’s Thesis