INOM
EXAMENSARBETE ELEKTRONIK OCH DATORTEKNIK, GRUNDNIVÅ, 15 HP
STOCKHOLM SVERIGE 2019,
Evaluation of Peer-to-Peer LoRa for Geolocation Usage in a Rural Environment
ELIAS CHAHINE WILLIAM LEWIN
KTH
SKOLAN FÖR ELEKTROTEKNIK OCH DATAVETENSKAP
Evaluation of Peer-to-Peer LoRa for Geolocation Usage in a Rural Environment
Elias Chahine William Lewin
2019-06-14
Bachelor’s Thesis
Examiner
Anders Västberg Academic adviser Mustafa Özger Industrial adviser Emma Larsson
KTH Royal Institute of Technology
School of Electrical Engineering and Computer Science (EECS) Department of Communication Systems
SE-100 44 Stockholm, Sweden
Abstract | i
Abstract
With the increased popularity of the Internet of Things (IoT) and vehicle tracking, long range communication alternatives become essential for connecting a large number of units in a network.
In areas where no coverage from existing technologies is available such as rural areas, the need for a solution presents itself. This thesis aims to examine the validity of using a Peer-to-Peer
network structure to send geolocation information among nodes in a network using the Long Range (LoRa) protocol as opposed to using Long Range Wide Area Network (LoRaWAN) structure. The research question that was used was the following: “Can LoRa be used for P2P networking to exchange geolocation messages in rural areas?”
By developing two modules using Raspberry Pi and Dragino LoRa/GPS shields, a prototype system could be created in order to test the performance of the network. The modules were tested in a rural area outside of Stockholm, Sweden.
Together with field test data from the prototype and existing network loss models the network performance was evaluated. Due to testing conditions a theoretical maximum communication range of around 12 km was established using this data. Therefore using Peer-to-Peer LoRa for geolocation purposes in rural areas was concluded viable.
Keywords
LoRa, Peer-to-Peer, Global Positioning System, Rural, Internet of Things
Sammanfattning | iii
Sammanfattning
Sakernas internet och fordonsspårning är två trender som ökar i popularitet i en rasande fart. Med detta ökar kraven på fler kommunikationsalternativ för att koppla upp ett stort antal enheter i ett nätverk med en lång räckvidd.
I områden såsom på landsbygden kan täckning av nuvarande teknologier kan vara bristfällande.
För att lösa detta problem presenterar detta arbete en föreslagen lösning. Målet med arbetet är att undersöka giltigheten i att använda en Peer-to-Peer nätverksstruktur för att skicka platsinformation inom ett nätverk med hjälp av Long Range (LoRa)-protokollet. I tidigare arbeten har detta
mestadels gjorts med hjälp av ett Long Range Wide Area Network (LoRaWAN). Frågan som
undersöktes i arbetet var följande: ”Kan LoRa användas för P2P nätverkande för att utbyta platsinformation i landsbygdsområden?”
För att undersöka giltigheten av denna idé utvecklades en prototyp i form av två moduler som använder sig av varsin Raspberry Pi och Dragino LoRa/GPS tillägg. Dessa system kunde senare användas för att testas i ett landsbygdsområde utanför Stockholm.
Tillsammans med data från fälttester med prototypen och modeller över fädning i ett nätverk utvärderades nätverkets prestanda. På grund av förhållanden i testmiljön bestämdes en teoretisk maximal räckvidd på runt 12 km. Med denna information dras slutsatsen att användandet av Peer- to-Peer LoRa för platsinformation i landsbygdsområden är giltigt.
Nyckelord
LoRa, Peer-to-Peer, Platsinformation, Landsbygd, Sakernas internet
Acknowledgments | v
Acknowledgments
We would like to thank our examiner Anders Västberg for continuous academic counseling throughout the project and sparking our interest in wireless systems.
We would like to thank Mustafa Özger for taking the time to answer our questions, refine the scope of the project and supporting our research process.
We would also like to express our gratitude towards Emma Larsson for her support and guidance as well as all of Cybercom for providing the opportunity to work with them.
Lastly, we would like to thank our friends and family for their encouragement and moral support during our years at KTH.
Stockholm, June 2019
Elias Chahine & William Lewin
Table of contents | vii
Table of contents
Abstract ... i
Keywords ... i
Sammanfattning ... iii
Nyckelord ... iii
Acknowledgments ... v
Table of contents ... vii
List of Figures ... ix
List of Tables ... xi
List of acronyms and abbreviations ... xiii
1 Introduction ... 1
1.1 Background ... 1
1.2 Problem ... 1
1.3 Purpose ... 2
1.4 Goals ... 2
1.5 Benefits, Ethics and Sustainability ... 2
1.6 Research Methodology ... 2
1.7 Delimitations ... 3
1.8 Structure of the thesis ... 3
2 Background ... 5
2.1 Wireless Interfaces ... 5
2.2 LoRa ... 6
2.2.1 Chirp Spread Spectrum ... 6
2.2.2 LoRaWAN ... 6
2.3 Signal Propagation ... 7
2.4 Peer-to-peer ... 7
2.4.1 LPWAN ... 8
2.5 Global Positioning System ... 8
2.6 Related works ... 8
2.6.1 Network Performance Evaluation of LoRa ... 8
2.6.2 LoRa Attenuation Between Nodes ... 9
2.6.3 Geo-location Tracking in Long-Range Wireless Sensor Networks ... 9
2.6.4 Summary of Related Works ... 10
3 Method & Methodologies ... 11
3.1 Project Methodology ... 11
3.2 Research Process ... 11
3.3 Prototype Development ... 11
3.3.1 System Design ... 11
3.3.2 Hardware ... 12
3.3.3 Software ... 12
3.4 Network Performance Evaluation ... 12
3.5 Testing ... 12
4 Implementation and Testing ... 15
4.1 Hardware Design ... 15
4.1.1 Raspberry Pi 3 Model B ... 15
viii | Table of contents
4.1.2 LoRa/GPS Shield ... 16
4.1.3 Assembly of the prototype ... 17
4.2 Testing Hardware ... 17
4.2.1 GPS Testing ... 18
4.2.2 LoRa Testing ... 18
4.3 Software process ... 19
4.4 Code process ... 21
4.4.1 GPS Filtration ... 21
4.4.2 Receiving/Transmitting LoRa Packages ... 22
4.5 Performance estimation ... 23
4.5.1 Hata Model Open Area ... 23
4.5.2 LEE Propagation model ... 24
4.6 Field Test ... 25
5 Results and Analysis ... 27
5.1 Final Prototype ... 27
5.1.1 Hardware ... 27
5.1.2 System ... 27
5.2 Performance evaluation ... 27
5.2.1 RSSI ... 27
5.2.2 SNR ... 28
5.2.3 PDR ... 29
5.3 Reliability of Result Analysis ... 30
5.4 Validity of Result Analysis ... 30
6 Discussion ... 31
6.1 System Prototype ... 31
6.2 Performance Evaluation ... 31
7 Conclusions and Future work ... 33
7.1 Conclusions ... 33
7.2 Limitations ... 33
7.3 Future work ... 33
References ... 35
Appendix A: Mathematica Code ... 37
Appendix B: Detailed System Flowchart ... 55
List of Figures | ix
List of Figures
Figure 2-1: Protocol comparison [7] ... 5
Figure 2-2: Chirp spread spectrum sweeping, varies in frequency [17 Ch. 1. p. 7] ... 6
Figure 2-3: General view of a P2P communication link ... 7
Figure 4-1: System diagram over module 1 and 2 ... 15
Figure 4-2: Raspberry Pi 3 Model B board ... 16
Figure 4-3: Dragino LoRa/GPS shield ... 17
Figure 4-4: Experimental prototype ... 17
Figure 4-5: Result of GPS test made with gpsmon and mapcoordinates.net [32, 33] ... 18
Figure 4-6: Receiving LoRa packages using example code ... 19
Figure 4-7: Loop for processing data ... 20
Figure 4-8: General overview of all loops activated ... 20
Figure 4-9: Function in class serial_rpi, written in Python [34] ... 21
Figure 4-10: Function in class serial_rpi, written in Python [34] ... 22
Figure 4-11: Function in class serial_rpi, written in Python [34] ... 22
Figure 4-12: Function in class serial_rpi, written in Python [34] ... 22
Figure 4-13: Calculated Received Signal Power using Hata Model for Open Area ... 23
Figure 4-14: Calculated Signal to Noise Ratio using Hata Model for Open Area ... 24
Figure 4-15: Received Signal Power using LEE Model ... 24
Figure 4-16: Calculated Signal to Noise Ratio using LEE Model ... 25
Figure 4-17: Field test location and surroundings ... 25
Figure 4-18: Blynk app visualization at 100 meters ... 26
Figure 5-1: RSSI plot showing field tested as well as estimated values ... 28
Figure 5-2: SNR values from field testing with approximated function ... 29
Figure 5-3: PDR values from field tests with approximated function ... 30
List of Tables | xi
List of Tables
Table 2.1: LPWAN protocol comparison [8, 9] ... 5 Table 3.1: Variables and performance metrics for network performance
evaluation ... 12 Table 5.1 Mean RSSI and confidence interval from field test values ... 28 Table 5.2 Mean SNR and confidence interval from field test values ... 29
List of acronyms and abbreviations | xiii
List of acronyms and abbreviations
CSS Chirp Spread Spectrum DSSS
GPS IoT LoRa LoRaWAN LOS LPWAN M2M NB-IoT P2P PDR RPI RSSI SF SNR TDD V2V V2X WSN
Direct Sequence Spread Spectrum Global Positioning System
Internet of Things Long Range
Long Range Wide Area Network Line of Sight
Low Power Wide Area Network Machine-to-Machine
Narrow Band – Internet of Things Peer-to-Peer
Packet Delivery Rate Raspberry Pi
Received Signal Strength Indicator Spreading Factor
Signal-to-Noise Ratio Test Driven Development Vehicle-to-Vehicle Vehicle-to-Everything Wireless Sensor Network
Introduction | 1
1 Introduction
This thesis project was conducted at KTH Royal Institute of Technology at the School of Electrical Engineering and Computer Science in collaboration with Cybercom Group in Stockholm, Sweden.
Cybercom group is an IT consulting company which specialize in helping companies and
organizations thrive in the connected world and its opportunities, providing them with innovative sustainable solutions [1].
This chapter describes background information regarding the project, the specific problem that this thesis addresses, the context of the problem, the goals of this thesis project, and outlines the structure of the thesis.
1.1 Background
In the current state of the Internet of Things (IoT), long range communication is a vital part of creating a network of connected sensors. The sensors can be set up as a network to provide
information regarding the physical state of technology such as a temperature or location, forming a Wireless Sensor Network (WSN). WSNs are today being used in precision agriculture, vehicle tracking and security and surveillance among other things [2]. Sending data within the network requires it to be based on a network structure, centralized or decentralized networks are two alternatives with each one having its own characteristics.
Centralized networks are based on multiple nodes connecting to a single point of processing unit called a gateway. The gateway processes and forwards the information that is being sent by the nodes in the network. Decentralized networks allow each node to process information. With this, the need for a gateway is eliminated [3]. Peer-to-Peer (P2P) networks are a type of decentralized network which allows every node to communicate with its adjacent nodes to transmit information.
Centralized networks are efficient at handling data but introduce a critical weak link in only having one processing unit. Decentralized and P2P networks allow for a more dynamic structure that provides improved scalability. This is due to an increase in processing power whenever a node is added in a P2P network. When a node is added in a centralized network, the processing unit instead becomes more congested [4p. 118].
One of the previously mentioned scenarios in which a WSN is implemented is vehicle tracking with Global Positioning System (GPS). In the future, when machines become autonomous, they need to be able to make informed decisions regarding routes, avoiding congestions and becoming more efficient. In this process, allowing for machine-to-machine direct communication would allow machines to be fully autonomous and make decisions based on the location of other machines.
In rural areas, P2P networks are essential where no gateways are previously established. Sensor networks do not require the use of high bandwidth to send information. In order to send the information, a wireless communication medium is needed. Many alternatives exist for sending low bandwidth information over long distances such as SigFox, NB-IoT and LoRa.
1.2 Problem
LoRa is in an evolving technology that is being researched in various use cases at the moment. A lot of research involves applying LoRa to a centralized network structure, the most common being Long Range Wide Area Network (LoRaWAN). Where no coverage is available from other wireless protocols such as in rural areas, a need for a dynamic network structure presents itself.
We would instead like to examine the possibility of implementing a LoRa P2P network structure to send packets containing GPS data from one machine to another. Therefore, our thesis aims to
2 | Introduction
answer the question: Can LoRa be used for P2P networking to exchange geolocation messages in rural areas?
In order answer the question, we propose building a prototype and performing experiments to determine the performance of the network in a rural area.
1.3 Purpose
With machine-to-machine (M2M) communication being developed, the demands for an efficient and stable communication network will grow. The purpose of this thesis project is to create a prototype which can send GPS coordinates using P2P LoRa in order to allow machines to communicate each others location directly. By the end of the thesis, an opportunity to view a conceptual model of a future in machine-to-machine communication will be available.
In the future, our conclusions can be used as a proof of concept for further development regarding peer-to-peer sensor networks. Since Cybercom Group are currently evaluating Vehicle-to- Vehicle (V2V) and Vehicle-to-Everything (V2X) [5] communication alternatives, we hope our product and research will contribute to their development process.
1.4 Goals
The main goal of this project is to create a prototype system that allows for machines to
communicate their position to nearby machines using LoRa in rural areas. The system is only meant as a proof of concept to establish P2P Lora as a valid communication for sending GPS coordinates and testing the networks performance. This system should be able to achieve above 95% Packet Delivery Rate (PDR) at a distance range of 0-1000 meters.
The prototype can be used as a base for future implementations of P2P Lora communication. In the case of future implementations, the performance evaluation can be used to determine whether the system is suitable for the intended work.
1.5 Benefits, Ethics and Sustainability
If one would implement this proof of concept, the usage target would be for many areas. With the built-in functionality to connect the prototype with a user interface in the form of a mobile application, monitoring of each prototype is simple.
Being a proof of concept, no encryption is implemented with the transmission. Location data is sensitive information and should be treated as such. In the current version of the build, the product should not be used to transmit or collect sensitive location information.
The Raspberry Pi (RPI) could be replaced with a product with lower power consumption in order to achieve better efficiency. A suitable battery could be hard to find and powering it with sustainable energy sources such as solar or wind can be difficult with products in the current state of the market.
1.6 Research Methodology
To investigate the viability of using P2P LoRa, a prototype of the system was developed using previous knowledge of embedded systems and following Applied research. By following Applied research [6], already existing research works will be used in order to build a practical application, in this case P2P LoRa system. This system was then field tested in a rural area using a number of variables and performance metrics. The collected performance metrics were then analyzed in comparison to estimations from signal propagation models to establish its viability.
Introduction | 3
3 1.7 Delimitations
This thesis will have constraints, mainly because of limited time but also due to resource limitations:
Only LoRa networks will be investigated, no interaction with WiFi or cellular for
communication among nodes will be examined. No practical interference testing will be done by us and will instead be estimated from previous research done on the topic.
1.8 Structure of the thesis
Chapter 2 presents relevant background information about different wireless interfaces and why the choice of LoRa was done. This follows up with some information about LoRa protocol and adjacent areas, Peer-to-Peer communication (P2P), Global Positioning System (GPS) and three different related studies. Chapter 3 presents the methodology and method used to solve the problem.
Chapter 4 describes the working process; from assembly to the working cycle of the system prototype. Chapter 5 contains results from tests that has been conducted, as well as analysis on reliability and validity of the result. Chapter 6 will be discussions about the project itself. Chapter 7 will be about conclusions and future work.
Background | 5
2 Background
This chapter provides basic background information about: Overview Wireless Protocols, LoRa protocol usage, communication applications and GPS link. Additionally, this chapter describes related work on how the network performance could look like using LoRa, what will happen when distances exceeds certain range and how could one design a geo-location tracking system in sensor network using long range communication.
2.1 Wireless Interfaces
There are a number of wireless interfaces that enable machine-to-machine communication (M2M).
Each of them is characterized in data rate and coverage as well as the allowed spectrum of each technology having some sort of regulation.
The main focus of the prototype is creating a long-range peer-to-peer system with a low cost.
High bandwidth is not a requirement for the system, meaning that the least and most important characteristics for the system are bandwidth and range respectively. To find a suitable interface for the project, these factors were compared in a number of related wireless interfaces, see Figure 2-1 [7].
Figure 2-1: Protocol comparison [7]
Further deducing can be done by comparing additional parameters. Below is a comparison of more contributing factors in the system, see Table 2.1 [8, 9].
Table 2.1: LPWAN protocol comparison [8, 9]
Interface Coverage Range (km)
Peak data rate
Deployment cost estimate
Spectrum cost SigFox 40-50 100 bps 4000 € per
base station
Free, unlicensed NB-IoT 1-10 200 kbps 15 000 € per
gateway
>500 M€ per MHz
LoRa 5-20 50 kbps 100 € per
node
Free, unlicensed
6 | Background
2.2 LoRa
Like many other wireless systems there is some sort of modulation technology for creating a communication link [10, 11]. Due to LoRa being patented by Semtech, the underlying modulation technology remains unknown. But together with analysis and information provided from Semtech, the LoRa modulation technology is derived from Chirp Spread Spectrum (CSS) [12, 13]. A long- range communication link is achievable due the properties from the CSS modulation scheme that it inherits. [10–13].
2.2.1 Chirp Spread Spectrum
Most older systems are functioning on modulation schemes that are based on frequency shift keying (FSK) because it could achieve low power. A similar modulation scheme that can achieve low power is Chirp spread spectrum (CSS). CSS has been used for decades by the military and space
communication but the difference between FSK is that the communication range is a lot longer and robustness for interference is high [10, 14–16]. CSS got released for commercial usage and LoRa was the first technology to implement the commercial version [10].
LoRa:s spread spectrum is associated with the traditional Direct Sequence Spread Spectrum (DSSS) systems and traditional spread-spectrum communications techniques. Combine these together and LoRa modulation will be low-cost, low-power, and robustness application [15]. The spread spectrum for LoRa is made of generated chirp signals with a fixed amplitude that in a continuous way varies in frequency, as shown in Figure 2-2 [13, 15, 16] [17 Ch. 1, pp. 6-7].
Figure 2-2: Chirp spread spectrum sweeping, varies in frequency [17 Ch. 1. p. 7]
The CSS is able to access the spectrum and broadcast information, this provides resistance to jammers and noise but with the cost of spectral efficiency. In some circumstances, CSS will be more resistant to multipath and Doppler effects than other modulations [16]. In LoRa, every bit in a data packet is spread by a chipping factor. These number of chips per bit is called spreading factor (SF) [16] [17 Ch. 1, p. 7]. Each chirp represents a symbol that has a certain number of encoded bits which varies with a depending SF [16]. SF determines if each chirp transmission will have less over-the-air time or the opposite, more over-the-air time. Each SF increase means that it will increase the chirp transmission which makes the message being sent having bigger robustness to noise or interference.
But with increased SF, the symbol number is being increased which means that frequency of symbol error being increased [16] [17 Ch. 1, p. 7].
2.2.2 LoRaWAN
Long Range Wide Area Network or LoRaWAN is a definition of communication protocol and system architecture for a network. Here, LoRa is acting like the physical layer which makes the long-range communication link possible for the network. The protocol and network architecture combined will determine several factors for an end-node. Security, network capacity, quality of service, and battery lifetime is some factors that is being affected.
Already deployed networks are typically utilizing a mesh network architecture, meaning information from nodes is forwarded by surrounding nodes to nearby gateways. This is done to increase the communication range. This complex structure of the network will decrease the factors that was previously stated.
Background | 7
7 In LoRaWAN, nodes are not forwarding information through other nodes instead they are sending information directly to a nearby gateway. Data that is being received by the nearest gateway is forward to a cloud-based network server through some sort of backhaul (Wi-Fi, cellular, satellite, Ethernet). All packets in the LoRaWAN network is being transmitted to the cloud where an
automated manager is performing different operations to determine whether to store or disregard the packet [10].
2.3 Signal Propagation
In a LPWAN network, a signal loses power from the sender to the receiver. The degree to which the signal deteriorates depends on a great number of variables. These variables have mostly been abstracted over time to be determined by pre-determined models to estimate loss depending on environment. In most cases, models are either adjusted to things such as proximity to cities, surrounding population and surrounding obstructions. In short, this means that a model is more or less suited to its surroundings according to its creator.
For our working conditions, there are models which correspond well for estimating loss. The Okumura-Hata model is a well-known model based on empirical data from various areas in Tokyo.
In this model, a carrier frequency along with the receiver and transmitter antenna heights are used to estimate path loss for environments. These estimations are based on if the area is considered a highly populated urban environment, small urban environment, rural area or open area [18p. 122, 19].
The Lee model has also been proven to act as a valid method to conduct further research in the 868 MHz-band [20, 21]. The Lee model can be used to estimate loss in point-to-point networks as well as area-to-area. In order to receive the best results, knowledge about the area is required to re- tune the model according to measured path loss from the area [21, 22p. 154].
2.4 Peer-to-peer
Some networks are configured to send data through a central server. For example, if there is a LoRaWAN network containing two devices and one gateway that is compatible for downlink. One of these devices need data from the other one. In order for one node to receive data from another, the data needs to be transmitted first to the gateway and transmitted from the gateway to reach the end node [10]. Such information that is transmitted and received by end nodes can be done without a central server, these implementations are called Peer-to-peer communication.
In order for peer-to-peer communication to work, a link between nodes needs to be established.
How this link is established depends on the technology that is implemented, Figure 2-3 shows a general view of a P2P communication link [23].
Figure 2-3: General view of a P2P communication link
8 | Background
2.4.1 LPWAN
LPWAN or Low Power Wide Area Network, is designed for applications among IoT (Internet of Things) devices. LPWAN is suitable for sensors and different applications that can manage to send few times per hour, small amount data for long distances [10].
2.5 Global Positioning System
Global Positioning System (GPS) is an American state-owned program which uses a network of 24 satellites to determine a precise position based on a signal sent from the satellite to the user. This network of satellites is designed to enable a receiver to be in view of at least four satellites regardless of the receivers location. To determine to position of the user, a distinct signal is sent from each of the four satellites in view at the same time. The signal contains an ultra-accurate measurement of when the time was sent and the location of the satellite. A receiver then receives the signal and compares the time when the signal was sent to the time when the signal was received to pinpoint the distance between the sender and the receiver [24].
Why four satellites? This is needed in order to achieve an accurate three dimensional
positioning according to the trilateration method; meaning using the spherical intersection of the distance to the satellites as a system of equations that is solved to output a location.
⎩⎪
⎨
⎪⎧(𝑥'− 𝑋*),+ (𝑦'− 𝑌*),+ (𝑧'− 𝑍*),= 𝑑(Δ𝑡*, 𝜀), (𝑥'− 𝑋,),+ (𝑦'− 𝑌,),+ (𝑧'− 𝑍,),= 𝑑(Δ𝑡,, 𝜀), (𝑥'− 𝑋9),+ (𝑦'− 𝑌9),+ (𝑧'− 𝑍9),= 𝑑(Δ𝑡9, 𝜀), (𝑥'− 𝑋:),+ (𝑦'− 𝑌:),+ (𝑧'− 𝑍:),= 𝑑(Δ𝑡:, 𝜀),
(4) [25]
𝑋; , 𝑌;, 𝑍;: The location of the satellites 𝑥; , 𝑦;, 𝑧; : The location of the receiver
This outputs a function of the sent and received time difference Δ𝑡𝒊 and receiver time-error 𝜀 that can be converted into coordinates [25]. All the calculations are done by the receiver, a computationally simple task for modern GPS systems.
2.6 Related works
Under this section three different related work will be presented. There will be a study about how the network performance looks like when using LoRa technology. A study about what kind of coverage there will be between base station and a node when the distance exceeds 10 km. Lastly, how a geo-location tracking system using long range communication in a sensor network would be designed and evaluated.
2.6.1 Network Performance Evaluation of LoRa
A previous study has been conducted on measuring a LoRa network by Alexander Liljegren and Robin Franksson at Blekinge Institute of Technology with various experiments. The network was set up with a SX1272 LoRa shield for the node and a Wimod LoRa Lite gateway attached to a Raspberry Pi acting as a gateway to the Internet.
This study includes a number of case studies which were analyzed according to some limitations. In testing the signal quality in terms of Received Signal Strength Indicator, a large decrease in urban and dense forest areas was noted. This problem could be largely bypassed with elevating nodes to reach a line-of-sight signal. Large scale implementations of LoRa networks are
Background | 9
9 also noted to reach a transmission distance of around 15 km, as long as the antennas are aligned [26].
Packet loss was noted to be a major concern and an important factor to consider when designing a LoRa network, especially since the usage of bandwidth is limited [26].
2.6.2 LoRa Attenuation Between Nodes
A research project was carried out in the Finnish city Oulu by Centre of Wireless Communications DCE section at the University of Oulu. Three typical challenges were presented but only one of them was decided to be experimented with, how will the coverage be between base station and a node when distance exceeds 10 km?
They used Kerlink’s LoRa IoT station with an biconical shaped antenna, D100-1000 from Aerial. This station was acting like a base station and was located at the University Oulu antenna tower. The antenna that was used provided 2 dBi gain on the band range 100 MHz to 1 GHz at the height 24 m from sea-level. LoRaMote devices was used for the nodes, all these devices were equipped with external transceiver antennas SX1272 that was Semtech-made.
The goal as described was to research on the maximum communication range, and for that did the nodes got configured to have a spread factor of 12 [27]. Like previously stated, the higher spread factor that is configured one can have more over-the-air time but lower data rates [17p. 6]. The data rate was noted to be 293 bps. This configuration with a settled bandwidth at 125 kHz, improved the base station sensitivity to reach -137 dBm. Each node had a capability to have 20 dBm (100 mW) transmit power but that restricted the devices to only use one channel, instead 14 dBm (25 mW) was configured for using all six channels.
Since there were multiple devices, made it possible to mount on different objects. They had one mounted on a boat and one on a moving car. Tables were presented over the measurement on each site with different ranges to the base station. Packet loss ratio for the car between 0-5 km was between 12-15% but increased significantly to almost 33% when distances were between 5-10 km.
The highest packet loss ratio that could be achieved were in 10-15 km and reached almost 74 %. For the boat, ranges 5-15 km gave 32 % and between 15-30 km 38% packet loss ratio [27].
2.6.3 Geo-location Tracking in Long-Range Wireless Sensor Networks
At Keio University in Yokohama Japan, a study was carried out on how one would design and evaluate a geo-location tracking in long-range wireless sensor networks project. The proposed solution was to use GPS to collect geo-location data and LoRa modules for sending the data inside the network.
The proposed system design consists of multiple sensor nodes and a base station. Each node in the network was assumed to be independent as well as each node sending data with no regard for other nodes in the network. For having a low-cost infrastructure, easy deployment and
maintenance, they used a long-range star i.e., centralized network structure.
The evaluation of the system was made based on observing when one to four nodes moves away from the base station at an experimental distance on 0 to 225 meters. By adding more sensors that moves in the same path, they observed that the quality signal reception was degrading. Also, the signal that the base station was receiving attenuated more when there were multiple nodes. They concluded that degradation of the RSSI value was due to interference from neighboring nodes. But they noted that their LoRa module had a large RSSI range between 0 dBm to -127 dBm which meant that the communication could achieve a greater distance [28].
10 | Background
2.6.4 Summary of Related Works
Three works were studied using LoRa capabilities, despite having different working areas and different setup configurations, all the studies above came to a common conclusion. Using LoRa, communication distances up to 5-30 km can be achieved.
The repeated successful results with Semtech transceivers in two of the above mentioned studies leads to us being confident that we will reach our goals. Geo-location using a GPS and LoRa has also been proven to yield successful results.
One important difference between this work and these mentioned above is that P2P
communication is not studied, instead all works use a centralized network structure. This study will therefore examine how well LoRa P2P communication performs when used in rural areas.
Method & Methodologies | 11
3 Method & Methodologies
The purpose of this chapter is to provide an overview of the research methods used in this thesis.
This includes the structure of the project itself as well as the implementation of the prototype and the network performance evaluation.
3.1 Project Methodology
The development was conducted using a development framework for agile methods called Scrum in combination with Test Driven Development (TDD). This meant iteratively solving problems based on dividing the work into smaller parts and solving them individually. An implementation of a part of the solution was written and tested, if the test were to fail, the development cycle would continue until the test was passed.
3.2 Research Process
In order to build a functioning prototype and test the network performance, Applied Research is used. Due to the project presenting a practical problem for a certain situation, Applied Research was an appropriate method for this project [6]. Building a functioning prototype and testing the network performance requires research to be done, therefore some research regarding different method and technologies was conducted. Initially, some background information regarding each area of the prototypes features was conducted and summarized, highlighting key points from multiple sources in chapter 2.
A comparison of different wireless communication alternatives was conducted to facilitate the choice of wireless protocol to be examined further and eventually be implemented in the final artefact. Similar projects were then investigated in order to gain more insight regarding the actual implementation and how the stability of the network can be evaluated, including what variables and performance metrics to compare. This research was applied to the prototype development and extended according to what was needed during development.
3.3 Prototype Development
This section will outline the methods used to develop the system design, hardware and software respectively.
3.3.1 System Design
The goal of the system is to be able to act as a P2P network and send GPS coordinates with a PDR of over 95% over at least 1 km independent of the connection in the surrounding area. In order to do this, the most critical part of the design is the chosen link technology. In order to satisfy the requirements of the prototype, a microcontroller with purpose-specific modules is used for portability, increased cost-efficiency and scalability.
There are some advantages to establishing a P2P network as opposed to a standard network with a centralized structure. The main feature of P2P is the ability to achieve uplink and downlink capabilities with cheap nodes. In previous network configurations for LoRa such as LoRaWAN, an expensive gateway is needed to have uplink and downlink capabilities, fully allowing M2M
communication in remote areas. The P2P structure is also self-reliant and does not need external connections such as Wi-Fi to function properly.
12 | Method & Methodologies
3.3.2 Hardware
A Raspberry Pi model 3B was chosen as a processing unit due to its similarity to an actual processing unit in a machine, not just acting as a system on a chip but a system in itself. The Dragino LoRa/GPS shield was selected as an extension to the Raspberry Pi. This was due to the fact that the shield has a GPS receiver and a LoRa transceiver built in and being native to the Pi, requires no soldering to attach.
3.3.3 Software
The software was mainly based on Python but because of functionality and performance issues, parts were written in C and Bash. Because of time constraints, example code written in C from manufacturer were used for LoRa communication with some added functionalities for bridging Python and C for implemented calculations in Python [29]. Bash was mainly written for automations on Raspberry Pi board.
3.4 Network Performance Evaluation
Evaluation of the network performance will be made by presenting data from chosen performance metrics. The performance metrics is depending on different variables input, see Table 3.1 for the different variables and performance metrics.
Table 3.1: Variables and performance metrics for network performance evaluation
Performance Metrics Variables
Received Signal Strength Indicator (RSSI) Communication Range
Signal-to-noise Ratio (SNR) Spreading Factor
Packet Delivery Rate (PDR)
Due to limited resources, two prototype nodes of the system will be tested. Before any
conclusions is made further, testing on more than two nodes needs to be made since this project is only seen as a proof-of-concept.
In order to estimate the performance of the network in terms of range, field tests in a rural area around Stockholm, Sweden will be conducted based on estimations according to multiple
propagation loss models in order to compare the estimations with empirical data. In the field tests values are analyzed using a purpose-built application that displayed the individual unit locations and received performance metrics from each module in 100-meter increments. If the field tests show results where no maximum communication range has been found, the theoretical loss models will be used to estimate a maximum range for the network according to trends shown in field test data.
3.5 Testing
Separate testing of the system nodes needed to be done at first to ensure that each part of the system is working as intended and making it easier to detect possible hardware faults. Testing the whole node is then being done, this to ensure no errors when the different parts is working together. Both the nodes create a system that needs to be tested, therefore testing on the whole system is made.
Testing is done on the whole system to detect possible error before field testing; this is detected by monitoring the systems functionality.
Method & Methodologies | 13
13
Implementation and Testing | 15
4 Implementation and Testing
In this chapter, the implementation of the experimental prototype will be explained. Firstly, an overview on what kind of hardware that was in need, followed up with the chosen hardware information and as well as how the assembly was conducted. An overview of the chosen software will be followed, this will show what kind of software and libraries were chosen in order for making it possible to send LoRa packages containing coordinates. Important code snippets of how the collecting of coordinates was done but also how the package transmission was done will be
explained. Lastly, estimated and actual values of the systems performance metrics will be presented.
4.1 Hardware Design
The chosen hardware was in need to meet the set requirements that were established at the project start and the main requirement was to use LoRa as a radio channel, LoRa was chosen due to the low node cost and free spectrum, as is described in Chapter 2.
In order for the project to prove that P2P communication using LoRa in rural areas can achieve good performance, performance metrics needed to be measured. With all the requirements set, goals for finding the most optimal prototype for this purpose could be listed:
• Send GPS coordinates
• Measure RSSI (Received Signal Strength) of the received signal on each node
• Measure SNR (Signal to Noise Ratio) for the received signal on each node
• Measure the PDR (Packet Delivery Rate) on each node
The system that was made consisted of two modules, each having a transceiver, two antennas and a microcontroller for different purposes, see Figure 4-1 for system diagram over module one and two. The two modules was set up using Raspberry Pi board with a Dragino LoRa/GPS shield.
This shield enables LoRa/ GPS connection on each module.
Figure 4-1: System diagram over module 1 and 2
A Raspberry Pi was used in order to achieve mobility and simulate the processing unit of an autonomous machine [29, 30]. Any microcontroller with corresponding modules could hypothetically be used to achieve the same results, these are not covered in this report.
4.1.1 Raspberry Pi 3 Model B
The Raspberry Pi 3 Model B board is used by both modules for processing and packaging data. The board is running on Broadcom BCM2837 64bit CPU with the help of 1GB RAM and Micro SD port
16 | Implementation and Testing
for loading the desired OS (Operating System) and storing data. It is powered with Micro USB power source and could handle up to 2.5A. The board did not only contain all of the above, it featured Ethernet, USB, HDMI, CSI (Camera), DSI (Display), 4 Pole Stereo and composite video ports. Figure 4-2 shows the Raspberry Pi 3 Model B board [30].
Figure 4-2: Raspberry Pi 3 Model B board
4.1.2 LoRa/GPS Shield
The LoRa/GPS shield is made by Dragino and is intended to be used as an LoRaWAN solution for Raspberry Pi. The shield sends LoRa packets using Semtech transceivers SX1276. These transceivers can achieve 168 dB in maximum link budget and +20 dBm – 100 mW RF output. The shield comes with an included glue stick antenna with an estimated gain of 3dBi. As described before LoRa has a good immunity and these transceivers offers blocking immunity.
This shield features a GPS module which is based on MTK MT3339. It uses NMEA 0183 and PMTK protocols for GPS information. It has different setup times for GPS fix depending on the climate and area. The antenna it uses is a built-in internal patch antenna, but it could be used with an external as well. No configuration is needed for the external antenna, the module has automatic switching. Figure 4-3 shows the Dragino LoRa/GPS shield [29].
Implementation and Testing | 17
17
Figure 4-3: Dragino LoRa/GPS shield
4.1.3 Assembly of the prototype
In order to create the desired prototype all parts need to be connected. Since Raspberry Pi is being used all the components are already installed, the same goes for the Dragino shield. This makes the prototype assembly easy. Raspberry Pi board is mounted with GPIO pins which makes it possible to mount different kind of shields. The Dragino shield is designed for this model only, therefore it is easy to mount it on the board. Each shield is then connected with an antenna for the LoRa transceivers and to differentiate the modules from each other both the antennas are marked with either one or two. Figure 4-4 shows the experimental prototype.
Figure 4-4: Experimental prototype
4.2 Testing Hardware
To establish hardware functionality, simple testing was done. The Raspberry Pi was proven to work since an OS could be installed and one could access the board using HDMI for visual analysis.
18 | Implementation and Testing
The initial hardware tests were conducted at the Cybercom offices at Arne Beurlings Torg 9 in Kista, Sweden. In these tests, the GPS and LoRa components of the prototype were tested separately with instructions from the manufacturer [29] in order to establish their functionality.
4.2.1 GPS Testing
In order to establish GPS functionality, tests were conducted to figure out how its data is captured on the serial port on the Raspberry Pi. Testing were made by following instructions from the
manufacturer for the chosen board. A step-by-step guide on this was done by the manufacturer [31].
Some of the testing instructions were obsolete due to the release of an updated operating system since the instructions were written. When reading from the serial port of the Raspberry Pi a coordinate can be extracted directly from the GPS module. This coordinate was compared to the actual coordinates of the testing location. Since these two coordinates matched, the functionality of the GPS was confirmed. Figure 4-5 shows results from the coordinate extraction and corresponding coordinates of the testing location in real time on a map.
Figure 4-5: Result of GPS test made with gpsmon and mapcoordinates.net [32, 33]
4.2.2 LoRa Testing
Testing of the LoRa transceivers need to be done as well. By following example code [29]that was provided by the manufacturer, tests to confirm that the hardware worked could be done, see Figure 4-6. The code that was provided for transmitting and receiving using LoRa was the origin for the project code.
Implementation and Testing | 19
19
Figure 4-6: Receiving LoRa packages using example code
4.3 Software process
Software for both the modules were written for sending GPS coordinates between each other. In this, open source code was used; this code provides an ability to send LoRa packages with the same shield between nodes. It does not only send the package, it also calculates the desired measurements Packet-RSSI, RSSI and SNR when receiving the packages.
Both modules are running a simple loop that processes data, they are configured to be running inverted from each other. Figure 4-7 explains the loop in flowchart diagram, in Appendix B more specific flowchart diagram is illustrated. Parallel to this loop other jobs are activated at the same time for other purposes. Figure 4-8 explains in generally how all loops are activated, in Appendix B more specific flowchart diagram is illustrated. The loops are described in section 4-5 more in detail.
20 | Implementation and Testing
Figure 4-7: Loop for processing data
Figure 4-8: General overview of all loops activated
Implementation and Testing | 21
21 4.4 Code process
Under this section the code loops are being described. The loops will be described using code snippets and flowcharts. All functions that are mentioned below can be found on the projects GitHub [34].
4.4.1 GPS Filtration
The main goal is to have a M2M (Machine to Machine) communication therefore sending GPS coordinates is the most valid information to have it sending between nodes. The reading of GPS information was done in three stages.
• Capture stage: where raw GPS information was captured.
• Filtering stage: where raw GPS information were being filtered for the chosen data.
• Extraction stage: where the chosen data were disassembled and left with coordinates.
4.4.1.1 Capture Stage
In the capture stage, the function def capture_gps_data(self) is called, see Figure 4-9. By importing serial library it is possible to capture serial data from the serial port. The variable serial_gps is assigned to be reading the output on /dev/ttyS0 with baud rate set to 9600, this reading duration is set to 8 seconds. After 8 seconds, the read values is being queued to variable serial_read, which is being read by variable gpsBuffer. This function will return raw GPS data.
Figure 4-9: Function in class serial_rpi, written in Python [34]
4.4.1.2 Filter Stage
To filter out GPS information, the function def filter_gps_data(self) is called, see Figure 4-10. This function collects raw GPS data from capture_gps_data(self), see section 4.5.1.1, and stores it in a variable called gpsBuffer. This variable is then used in a while loop, with if statements in order to filter out unwanted information. This function returns rows only containing “$GPRMC” which contains coordinates.
22 | Implementation and Testing
Figure 4-10: Function in class serial_rpi, written in Python [34]
4.4.1.3 Extraction Stage
In the extraction stage, the function def filter_coordinates(self) is called, see Figure 4-11. This function takes the filtered data from filter_gps_data(), disassembles the data and extracts
longtitude and latitude. The result of this function is coordinated that can be formatted into Decimal Degrees-format.
Function called def write_to_file(self) is called in order to bridge Python with C, see Figure 4- 12. The latitude and longtitude that was extracted in function filter_coordinates() is being stored in a file called “coordinates”
Figure 4-11: Function in class serial_rpi, written in Python [34]
Figure 4-12: Function in class serial_rpi, written in Python [34]
4.4.2 Receiving/Transmitting LoRa Packages
GPS coordinates are now ready for processing and packaging in order to send it using LoRa.
Because the module uses program written in C (programming language) for sending LoRa a bridge is therefore created in form of storing data in a separate file. To make this bridge, a function in Python def write_to_file(self) is called. Next stage is to package the data for transmitting via LoRa,
Implementation and Testing | 23
23 this is done by first calling function void readFromFile(), this function will read the data that was written to a specific file and then sending it using void txlora(byte *frame, byte datalen).
The receiving side will listen on the same frequency band until it senses that data is being transmitted, this is done calling function void recievePacket(). When a detection is made it will capture the package that was sent and store it in a specific file, this is done with function void writeToFile(). This data that is being written to a file on the receiving side will be taken for examination. During the examination, function def readFromFile(self) will read the coordinates, generate own coordinates and compare them. After the comparison, calculations will be made to look over the distance between these modules using function def calculate_dist_gps(self).
4.5 Performance estimation
Under this section estimated values of RSSI and SNR is presented. Calculations are made based on chosen propagation models that reflect the project scenario. All graphs are made using Mathematica version 11.3.0.0 [35]. All the Mathematica code including calculations can be found in Appendix A.
4.5.1 Hata Model Open Area
The Hata model for open areas is a model that assumes that no obstacles are in the way of the transmission, which is appropriate for testing in a rural area. The model is based on empiric data, making it appropriate for a practical coverage test.
Figure 4-13 shows maximum received signal power with both the minimum and maximum
spreading factor irrespective uneven terrain, trees, stones etc. [19]. Figure 4-14 shows an estimation of expected SNR value depending on what range the nodes have between each other but also on which spreading factor is enabled [36].
Figure 4-13: Calculated Received Signal Power using Hata Model for Open Area
24 | Implementation and Testing
Figure 4-14: Calculated Signal to Noise Ratio using Hata Model for Open Area
4.5.2 LEE Propagation model
LEE propagation model gives an estimation on how losses will occur in flat terrain. Figure 4-15 shows maximum received signal power that can be achieved in range with minimum and maximum spreading factor [21]. Figure 4-16 shows estimated SNR values depending on range but also which spreading factor that is in use [36].
Figure 4-15: Received Signal Power using LEE Model
Implementation and Testing | 25
25
Figure 4-16: Calculated Signal to Noise Ratio using LEE Model
4.6 Field Test
In field testing, the system parameters were set according to how the system is meant to function.
This meaning that the SF being set to 7, a packet size of 19 byte containing 9 bytes for latitude and longitude respectively and a newline-character. The center frequency was set at 868.1 MHz.
The field testing was done using our actual system, meaning that the sent packages contained a latitude-longitude coordinate from one node to another. The performance metrics were monitored using the output of the receiving node and our implementation of the Blynk app. The testing was done in a rural area just north of Stockholm, Sweden where a Line-of-sight path could be achieved at around 0-700 meters and the rest up to 1100 meters with surroundings of trees and houses, see Figure 4-17.
Figure 4-17: Field test location and surroundings
Sampling of performance metrics were extracted in 100 meters increments for the available line-of-sight distance. An example of this is shown in Figure 4-18, where the sender is placed around 100 meters from the receiver.
26 | Implementation and Testing
Figure 4-18: Blynk app visualization at 100 meters
To measure PDR, 100 packages were sent at each increment and analyzed from the receiver output to determine whether a package was intact or not.
Results and Analysis | 27
5 Results and Analysis
In this chapter, final version of the system prototype, results from the established tests together with reliability and validity analysis will be presented.
5.1 Final Prototype
Under this section the prototype nodes will be presented. What they are built around and what is running in the software code.
5.1.1 Hardware
The final node prototype is built around a Raspberry Pi 3 board together with the Dragino LoRa/GPS shield, see Figure 4-4 for both the nodes that complete the system. The Dragino
LoRa/GPS shield is being used to acquire GPS information in order to be transmitted using LoRa. It is also being used to receive the transmitted GPS information.
5.1.2 System
At startup, different jobs are being started for making the system to work. After startup procedure is completed, at every whole minute different jobs will be started and be active a certain time.
Approximately 40 seconds after startup, if the nodes have a clear view to the sky, GPS fix signal should be activated, and GPS information should be transmitted using LoRa. Without the GPS information, LoRa packages will be sent but containing without GPS information. When each node is the receiving part, performance metrics is being presented by reading from different
predetermined pins.
For monitoring purposes, this is being transmitted to a self-made Blynk application but in order for monitoring to work, both the nodes need to have some sort of internet connection. The software code that the system acquires can be found on the projects GitHub [34].
5.2 Performance evaluation
Under this section, performance metrics from the fields test will be presented. These performance metrics are Received Signal Strength Index (RSSI) followed with Signal to Noise Ratio (SNR) and Packet Delivery Rate (PDR) from the field test.
5.2.1 RSSI
With distance restrictions in the area, tests that was conducted was being limited to 1.1 km. Both the models that was chosen is being assumed that it is Line of Sight (LOS). The field tests LOS was possible for 0-700 meters, 700-1100 meters included some trees and houses.
Using spreading factor 7 and bandwidth 125 kHz, estimation on possible RSSI values for the two models were made in chapter 4.6.1 and 4.6.2. With estimating RSSI values, maximum distance could be established. Following Hata model and LEE model, maximum distance was approximately 3,6 km and 2,5 km respectively.
A plot showing field tests together with both models and spreading factor limits can be seen in Figure 5-1. With the field test only allowing for a 1.1 km estimation, the maximum distance needs to be established. If the LEE model is used to estimate, a theoretical maximum distance will be below 2,5 km when using spreading factor 7 and bandwidth 125 kHz. If a spreading factor of 12 is applied,
28 | Results and Analysis
a theoretical maximum distance would be below 12.6 km according to the LEE model. The
approximated linear function from the field test indicates an estimated maximum communication range of around 5 km using spreading factor 7. The associated confidence interval from the field test data is also listed in Table 5.1.
Figure 5-1: RSSI plot showing field tested as well as estimated values
Table 5.1 Mean RSSI and confidence interval from field test values
5.2.2 SNR
Figure 5-2 shows SNR values from the field tests with an approximated function. SNR values can be positive and negative, positive means stronger signal and negative means weaker signal. Spreading factor tells where the noise floor is, if the signal falls into the noise floor it got higher chance of being unable to retrieve [37].
According to the approximated function, 0-650 meters gives positive SNR values. 650 – 1100 meters SNR values are negative. SNR value around -7.4 dB can be achieved at 1000 meters, just above that the signal enters the noise floor for spreading factor 7. The associated confidence interval from the field test data is also listed in Table 5.2.
Results and Analysis | 29
29
Figure 5-2: SNR values from field testing with approximated function
Table 5.2 Mean SNR and confidence interval from field test values
5.2.3 PDR
In Figure 5-3 PDR values are being displayed with an approximated function. PDR values is around 95% - 100% for ranges between 500 meters. 600-800 meters PDR values is around 93%-95%. As the distance increases, the PDR value will decrease. Below 900 meters the PDR values are being decrease rapidly and this is being showed with the downward trend of the approximated function.
This is due to the SNR approaching the noise floor, as seen in the previous graphs.
30 | Results and Analysis
Figure 5-3: PDR values from field tests with approximated function
5.3 Reliability of Result Analysis
The result was impacted by several factors that could limit the testing abilities. Transmission distance, testing area, and node height are some examples of factors that impacted the result reliability. With carefully selected propagation models, an estimation on RSSI and comparison to the field tests values could be done. The propagation models provided estimated distances up to several kilometers, and both propagation models were modeled for LOS areas which was hard to acquire in reality. Both models also required the receiver node to be placed at a certain height, way above the tested height. This had an impact on the ability to use apply propagation models to the field test results. Since they showed similar trends however, they were deemed appropriate to estimate a maximum transmission range.
5.4 Validity of Result Analysis
For this project, despite the factors that made the result seem unreliable the likeness to the
investigated models is clear. To further solidify the viability of our results, further testing in a larger scale is encouraged.
Discussion | 31
6 Discussion
In this chapter, a discussion and evaluation regarding the problems and goals from Chapter 1 will be presented.
The purpose of this system was to present a proof of concept using P2P LoRa communication to send GPS coordinates and evaluate the performance. The system prototype that was built had fulfilled a P2P network structure and by fulfilling a requirement made it also possible to evaluate the performance of the system. The prototype met the set expectations, it could communicate using software with a similar machine and this communication could be evaluated using set performance metrics.
6.1 System Prototype
The use of GPS data for transmission testing and real-world implementation functioned well as an prototype for a future system. The Dragino LoRa/GPS shield functioned well under the right circumstances but took a long time to set up due to the lack of existing code for extracting data from the serial port and forwarding it to the LoRa module. The GPS receiver was sensitive to which environment it is calibrated in. This created some trouble when testing the hardware due to the initial test occurring indoors. However, in the field test this proved to not be an issue. In Chapter 5, Figure 5-1 shows the final version of the system prototype.
When developing the system, the use of LoRaWAN was considered to enable M2M
communication but discarded in favor of using P2P LoRa due to not needing a previously set up gateway to send packages in rural areas.
6.2 Performance Evaluation
Using our built prototype system made it possible for field testing in order to extract performance values. Performance values were captured at ranges between 0 – 1.1 km with 100 meters
increments.
In section 5.2.1 Figure 5-2, shows approximated field test function together with two estimated RSSI evaluations based on two different propagation model and RSSI limits for two different spreading factors. Since the propagation models assumes a starting height of about 20 meters for the receiver and with no obstacles such as trees and houses, the initial RSSI value would be better.
The field tests showed that it had similarities of these two different RSSI calculations, but it followed the LEE Model more. With this in mind estimations using LEE Model would be more ideal for this project.
In section 5.2.2 Figure 5-3, shows SNR values from the field test with approximated function together with two different spreading factor limits. The two spreading factors limit acts like a noise floor, which indicates that the signal has higher chance of being corrupted. The plot shows a stable downtrend with the incrementing distance. With the signal closing in on spreading factor 7 noise floor, it becomes more obvious that packets are being corrupted. In section 5.2.3 Figure 5-4, shows PDR values from the field test with approximated function. With Figure 5-3 and 5-4 in mind, we can see the signal being more corrupted as it descends in the noise floor.
With these three performance metrics, an evaluation of the system can be made. For this thesis project this system performs well in LOS, where in our area was the distance 0-700 meters. As explained above, due to similarities choosing LEE Model as a base ground for calculations would be the best option. LEE Model theoretical range goes up to 2,5 km, if the LEE model is used to estimate
32 | Discussion
a maximum range, the distance would reach 2,5 km. If the linear approximation of the field test is used, a maximum range of around 5 km is achievable.
During the system performance process, the possibility of investigating the network performance under interference testing proved to be time-consuming. It was found to be done extensively in the past however, leaving us confident that the system would perform well even during active interference [36, 38, 39].
