cost structure of iot connectivity services1326313/fulltext01.pdf · cases. the cost driver is...

INOM EXAMENSARBETE ELEKTROTEKNIK,AVANCERAD NIVÅ, 30 HP

, STOCKHOLM SVERIGE 2019

Cost Structure of IoT Connectivity Services

LU LIN

KTHSKOLAN FÖR ELEKTROTEKNIK OCH DATAVETENSKAP

Cost Structure of IoT Connectivity Services

Lu Lin

2019-06-08

Master’s Thesis

Examiner

Jan I Markendahl

Academic adviser

Mohammad Istiak Hossain

KTH ROYAL INSTITUTE OF TECHNOLOGY

I N F O R M A T IO N A N D C O M M U N I C A T I O N T E C H N O L O G Y

KTH Royal Institute of Technology

School of Electrical Engineering and Computer Science (EECS)

Department of Communication Systems

SE-100 44 Stockholm, Sweden

Abstract

This thesis is a techno-economic study focus on the IoT connectivity service field. It describes the specifications of LPWAN, LPLAN, and Cellular-IoT technologies. The analysis method of dimensioning and cost structure calculation of IoT network is enhanced from previous wireless network research, which is also a research gap in the telecom industry. By using this method, the dimensioning results and cost structure performances can be obtained when having the inputs of the scenario. The results are compared among SigFox, LoRaWAN, NB-IoT, LTE-M, EC-GSM-IoT, and WiFi-HaLow. Furthermore, we find out the cost driver of different technologies. When it comes to different actors, a new market entrant or an incumbent, the strategies taken by the actors are compared. They are building own sites or leasing sites from other actors, even reusing sites if possible. The contribution of this thesis is pointing out the dimensioning and cost structure calculation method of deploying IoT connectivity. Another contribution is comparing the cost performance and figuring out the cost driver. SigFox is cost-efficient for low capacity scenario, while LTE-M is of good cost performance for high capacity cases. The cost driver is operation and maintenance cost and site build cost. By leasing and reusing sites, the site build cost can be largely reduced.

Keywords

Network dimensioning, Cost Structure, IoT, LPWAN, Cellular, WiFi-HaLow

Sammanfattning

Denna avhandling har ett ekonomiskt och tekniskt perspektiv på anslutningstjänsten IoT. Den

beskriver specifikationerna för LPWAN, LPLAN och Cellular-IoT teknik. Analysmetoden för

dimensionering och beräkning av kostnadsstruktur av IoT-nätet har förbättrats jämfört med

tidigare trådlös nätverksforskning, vilket i sig är ett område som saknar tillräcklig forskning

inom telekombranschen. Genom att använda den här metoden kan dimensioneringens resultat

och prestandan av kostnadsstrukturen erhållas när indata finns tillgängligt för olika

fall/scenarion. Resultaten jämförs bland SigFox, LoRaWAN, NB-IoT, LTE-M, EC-GSM-IoT och

WiFi-HaLow. Dessutom hittar vi kostnadsdrivaren av olika teknologier. När det gäller olika

aktörer, en ny marknadsaktör eller en etablerad aktör, jämförs de olika strategier som

aktörerna använder. De bygger egna basstationer eller leasar basstationer från andra aktörer,

och i vissa fall även återanvändning av basstationer om så är möjligt. Denna avhandling bidrar

till att peka ut beräkningsmetoden för dimensioneringen och kostnadsstruktur av att bygga ut

IoT-anslutningar. En annan slutsats är jämförelsen av kostnadseffektiviteten mellan olika

teknologier och att komma underfund med kostnadsdrivaren. SigFox är kostnadseffektiv i

fallen där kapaciteten är låg, medan LTE-M har bra kostnadseffektivitet i fallen där kapaciteten

är hög. Kostnadsdrivaren är drift- och underhållskostnad samt kostnad för konstruktionen av

byggplatsen. Genom att leasa och återanvända basstationer kan kostanden för konstruktionen

av byggplatsen minskas till en stor del.

Nyckelord

Nätverksdimensionering, Kostnadsstruktur, IoT, LPWAN, Cellular, WiFi-HaLow

Acknowledgments

I would first like to thank my thesis advisor Mohammad Istiak Hossain of the School of Electrical Engineering and Computer Science (EECS) at KTH. The door to Istiak’s office was always open to me whenever I ran into a trouble spot or had a question about my research or writing. Through the discussion with him, I gradually find out the direction of this thesis and get good results in the end.

I would also like to thank my thesis examiner Jan I Markendahl of the School of Electrical Engineering and Computer Science (EECS) at KTH, who offer me the opportunity of doing this topic. His kindly comments also help me improve the quality and structure of my thesis a lot. Without their passionate participation and input, the thesis cannot be completed with such results.

I would also like to thank my thesis opponent Xinkai Xiong of the School of Electrical Engineering and Computer Science (EECS) at KTH, his opposition give me some ideas of improvements and future work.

Finally, I must express my very profound gratitude to my parents and to my boyfriend for providing me with unfailing support and continuous encouragement throughout my years of study and through the process of researching and writing this thesis. This accomplishment would not have been possible without them. Thank you.

Gothenburg, May 2019

Lu Lin

1

Table of Contents

Abstract .................................................................................................................................. 2 Keywords ........................................................................................................................................ 2

Sammanfattning .................................................................................................................. 3 Nyckelord ....................................................................................................................................... 3

Acknowledgments .............................................................................................................. 4

List of Abbreviations ......................................................................................................... 2

List of Figures ....................................................................................................................... 4

List of Tables ........................................................................................................................ 5

1 Introduction ...................................................................................................................... 1 1.1 Background............................................................................................................................. 2 1.2 Literature review .................................................................................................................. 4 1.3 Problem .................................................................................................................................... 6 1.4 Methodology ........................................................................................................................... 7 1.5 Outline ...................................................................................................................................... 8

2 Overview of Technologies ......................................................................................... 10 2.1 SigFox ...................................................................................................................................... 10 2.2 LoRaWAN ............................................................................................................................... 11 2.3 NB-IoT ..................................................................................................................................... 12 2.4 LTE-M ...................................................................................................................................... 12 2.5 EC-GSM-IoT ........................................................................................................................... 13 2.6 WiFi-HaLow .......................................................................................................................... 13

3 Calculation method...................................................................................................... 15 3.1 Scenario Assumption......................................................................................................... 15 3.2 Network Dimensioning .................................................................................................... 16 3.3 Cost structure Calculation ............................................................................................... 19

4 Analysis Results ............................................................................................................ 21 4.1 Scenarios ............................................................................................................................... 21 4.2 Dimensioning Results ....................................................................................................... 21 4.3 Cost Structure Results ....................................................................................................... 24 4.4 CAPEX/OPEX Comparison ............................................................................................... 26 4.5 Leasing/Building Comparison ....................................................................................... 28

5 Conclusions .................................................................................................................... 31

References .......................................................................................................................... 32

2

List of Abbreviations

2G 2nd Generation 3GPP 3rd Generation Partnership Project 5G 5th Generation AFA Adaptive Frequency Agility AP Access Point BPSK Binary Phase Shift Keying BW Bandwidth CAGR Compound Annual Growth Rate CAPEX Capital Expenditure CDMA Code Division Multiple Access DL Down Link EC-GSM-IoT Extended coverage GSM IoT E-CID Enhanced Cell ID EDGE Enhanced Data Rates for GSM Evolution EGPRS Enhanced General Packet Radio Services eMTC Enhanced Machine Type Communication FDD Feature-driven Development GPRS General Packet Radio Services GSM Global System for Mobile communications HD Half Duplex IEEE Institute of Electrical and Electronics Engineers ICT Information and communications technology IoT Internet of Things ISM band Industrial, Scientific and Medical Radio Bands ITU International Telecommunication Union LBT Listen Before Talk LoRaWAN Long Range Wide Area Network LPWAN Low Power Wide Area Network LPLAN Low Power Local Area Network LTE Long-Term Evolution LTE-M Long Term Evolution Category M1 NB-IoT Narrowband IoT MAC Media Access Control MCL Maximum Coupling Loss METIS Mobile and Wireless Communications Enablers for Twenty-twenty (2020) Information Society M2M Machine to Machine OPEX Operating Expense OSI Open Systems Interconnection OSS/BSS Operations Support System/Business Support System OTDOA Observed Time Difference of Arrival PBCH Physical Broadcast Channel

3

PDCCH Physical Downlink Control Channel PDSCH Physical Downlink Shared Channel PHY Physical Layer PSM Phase Shift Modulation RAT Radio Access Technology SC-PTM Single Cell Point to Multipoint SFBC Space Frequency Block Coding STA Station TDD Test-driven Development TWT Target Wake Time UL Up Link VoLTE Voice over Long-Term Evolution WLAN Wireless Local Area Network WAN Wireless Area Network

4

List of Figures

Figure 1 IoT Structure ......................................................................................... 3 Figure 2 IoT Technologies Grouped by Range[7] .............................................. 4 Figure 3 Methodology Flow Chart ...................................................................... 8 Figure 4 Analysis Method Overview .................................................................. 15 Figure 5 Scenario Partition ................................................................................ 16 Figure 6 Three types of sites (Omni directional, 3-RAT null-sector, Three-sector)[38] ......................................................................................................... 17 Figure 7 End-Device Cost ................................................................................. 20 Figure 8 Dimensioning of Scenario 1 ............................................................... 22 Figure 9 Dimensioning of Scenario 2 ............................................................... 22 Figure 10 Dimensioning of Scenario 3 ............................................................. 23 Figure 11 Dimensioning of Scenario 4 .............................................................. 23 Figure 12 Cost of Scenario 1 .............................................................................. 24 Figure 13 Cost of Scenario 2 ............................................................................. 25 Figure 14 Cost of Scenario 3 ............................................................................. 25 Figure 15 Cost of Scenario 4 ............................................................................. 26 Figure 16 CAPEX and OPEX Comparison for Scenario 1 ................................ 26 Figure 17 CAPEX and OPEX Comparison for Scenario 2 ................................ 27 Figure 18 CAPEX and OPEX Comparison for Scenario 3 ................................ 27 Figure 19 CAPEX and OPEX Comparison for Scenario 4 ................................ 28 Figure 20 Site Leasing Cost Compared to Site Building Cost of Scenario 1 .... 28 Figure 21 Site Leasing Cost Compared to Site Building Cost of Scenario 2 .... 29 Figure 22 Site Leasing Cost Compared to Site Building Cost of Scenario 3 .... 29 Figure 23 Site Leasing Cost Compared to Site Building Cost of Scenario 4 .... 30

5

List of Tables

Table 1 Specifications of Technologies ............................................................. 10 Table 2 Coverage range [39] .............................................................................. 17 Table 3 Cost Assumptions ................................................................................. 19 Table 4 Scenario List.......................................................................................... 21 Table 5 Device Penetration Rate ....................................................................... 21 Table 6 Reusable Site Number ......................................................................... 30

1

1 Introduction

As defined by International Telecommunication Union (ITU), to enable advanced services for the information society, Internet of things is to build a global infrastructure, which connects the physical and virtual things based on existing and developing ICT industry[1]. IoT is building the network connection between physical world and information world. Physical things can be sensed, actuated and connected. Information obtained from the physical things can be stored, processed and accessed. The communication is no longer between human to human, but also between human to things and things to things. So, the development of IoT is not about upgrading the data rate, capacity or quality but applying the wireless network to more devices. With the approaching of 5G, IoT is one of the technology trends nowadays. The development of this technology could kind of help us solving top global challenges resulting from population explosion, energy crisis, resource depletion and environmental pollution[2]. Both traditional fields and new areas, such as ubiquitous wireless communication, real-time analytics, machine learning, commodity sensors, and embedded systems, can encourage the growth of this IoT industry. According to Ericsson’s estimation, there will be 3.5 billion cellular IoT connections by 2023, increasing with an annual growth rate of 30 percent. This will impact industries and businesses across many different markets largely[3]. There will be a large amount of IoT use cases covering diverse areas, which can be categorized to massive IoT and critical IoT. Massive IoT is of the characteristics of low cost, low energy, small data volumes and massive numbers, while critical IoT are ultra-reliable and with very low latency and very high availability[4]. For massive IoT, it is facing some challenges to fulfill the market, to reduce the device cost, to extend the battery life and coverage and to enable larger scalability and diversity. Besides substantial use cases, all kinds of technologies are emerging to enable the use cases, both using licensed and unlicensed frequency, from LPWAN, LPLAN to Cellular-IoT. For actors who want to share the IoT market, it is hard to figure out the dimensioning and cost structure calculation method and find out cost-efficient technologies for a certain scenario. Most of researches in IoT area work on mechanisms and technical performance of IoT technologies. There is a gap when it comes to dimensioning and cost structure analysis. Dimensioning is always the first step of network planning. Only when having the results of the estimated number of elements from the network dimensioning, the cost structure can be calculated. The method of calculating is stated in this thesis as it is based on the characteristics of IoT network. Cost structure is one significant thing from the economic aspect. Analyzing cost structure for an industry can be helpful to maximize the profit

2

or control the cost. What’s more, it is valuable when making deployment strategies. So, this thesis focusses on the technology-economic aspect to analyze the cost structure of IoT connectivity to fill in the gap. The method of calculating cost structure of IoT connectivity is brought out as it is different from LTE network, which is worth analyzing. The cost performance of six technologies and different strategies will be compared under different use cases to give references for actors in the market to make strategies and apply them. In this section, the background, literature review, problem statement and outline of the thesis are introduced.

1.1 Background

To give a better understanding of how IoT is developed to build the network and create the connection, the architecture of IoT industries are described below. The system is divided to four layers which can support the environment with high flexibility and reliability. Figure 1 shows the structure of IoT. 1. Devices: The end devices consist of sensors and actuators. They can get

information and perform tasks but are with limited capability of computing, data storage, and transmission. They are connected to gateways for data aggregation or other devices directly for information forwarding[5].

2. Communications: This layer provide data communication network to IoT devices. IoT technologies are applied in this layer to support the wireless network.

3. Platforms: An IoT platform is about to enable a variety of important building blocks: connectivity & normalization, device management, database, processing & action management, analytics, visualization, additional tools, and external interfaces[6].

4. Applications: The software layer provides services for users to access functions of IoT services[5].

This thesis focusses on the communication layer, of which the cost structure of building and operating a network is analyzed. The physical layer, which is about the end devices, the sensors and actuators, are briefly discussed. The platform is another important layer to be filled for techno-economic aspect research. As for the application layer, it is software- oriented.

3

Figure 1 IoT Structure

With the aim of deploying the communicating/networking layer, diverse technologies are released. They can be grouped by coverage range, as seen in Figure 2 IoT Technologies Grouped by Range [9] from KeySight. Some technologies support indoor areas, while others can operate over long distance. Some technologies are well developed, such as Bluetooth, WLAN and cellular, they are already widely used. Others, such as ZigBee and Thread, are emerging in specific market niches. In this thesis, the focus is Low Power Wide Area Networks (LPWANs). LPWAN technologies are designed for low power and low data rate IoT devices, distinguishing from wireless WAN that used to connect users or businesses, and requires larger data capacity. In addition, it is cheap compared to other network technology, with low chipset cost and device operating cost. Currently, there are many LPWAN technologies merging and competing against each other in both unlicensed and licensed frequency bandwidth. Among them, SigFox, LoRaWAN, NB-IoT and LTE-M are the well-known technologies nowadays. 3GPP also released EC-GSM-IoT besides NB-IoT and LTE-M. WiFi-Alliance has introduced WiFi-HaLow to enable new power- efficient IoT use cases, such as smart home, smart city as well as connected cars and so on. Diverse technologies aim at the rapidly-developing the IoT industry. A more detailed description of all these technologies will be stated in the following thesis.

4

Figure 2 IoT Technologies Grouped by Range[7]

eMTC shall gradually become a key point in the telecom market with the development of 5G. In the past, many economic analyses of this industry have been delivered, from 2G to LTE, from base stations to mobile APP. One important concept within the techno-economic area is the cost structure. With the willing to succeed in one business, it is necessary to know what kinds of expenses spent in this business and to keep track of them possibly. In a business , the list of the types of expenses will cost, or has, is called cost structure[8]. If someone ran a cupcake shop, he or she needs a lot of ingredients to make the cupcakes and lots of boxes to package the cupcakes. That’s the cost structure of cupcake business. However, each business is unique, in other words, each business has its own cost structure. So, it is also meaningful to figure out the cost structure of IoT network. In this thesis, we group the costs into CAPEX and OPEX.

1.2 Literature review

At the beginning of doing the thesis, the references regarding IoT technologies and ICT techno-economic analysis are reviewed. A number of papers have been published on IoT technologies, their specifications have been introduced, detailed comparisons among the technologies are given. In order to analyze the cost structure, it is necessary to understand the IoT technologies to calculate the network capacity. In [9], Mekki et al focus on SigFox, LoRaWAN and NB-IoT to deliver a comparative study, the technical details are stated, comparison on quality of service, battery life, latency, scalability, payload length, network coverage, range, deployment model, and cost are compared. The paper also figures out some use cases to find out the technology that fits best. Finnegan

5

and Brown provide a definition of LPWA paradigm, present an approach to give suitable use cases, and deliver comparison of current LPWA standards, covering technologies, upcoming cellular options, and remaining proprietary solutions in [10]. 802.11ah and its challenge are introduced by Qutab-ud-din in [11], devices in the Sub- 1 GHz ISM bands must comply with the maximum duty cycle limit of 2.8% in Europe. LTE-M is another important candidate to support M2M communication, whose performance is deeply analyzed in [12]. There are a lot of other papers introduce on the IoT specifications. The parameters that matters will be listed in the following. Some paper also stated the current status of M2M communications. But most of the papers are focus on the technical aspect, only a few researches are delivered from the economic aspect. Operators and companies may have a hard time choosing a technology when deploy an IoT network under a certain scenario. Nowadays, more and more IoT applications are merging, a scientific research and comparison considering cost structure of IoT services are necessary. From the past analysis, Jens Zander delivered some simple cost models taking wireless network parameters into consideration. He believed the cost of infrastructure (building and maintaining) and the cost of spectrum (licensing and bandwidth auctions) are also important parts in the cost structure in some cases besides the base station cost (including base station sites, antennas, towers etc.) and the fixed wired communications network connecting the base stations[13]. However, there is no existing IoT cost models’ assumptions, but which can contribute to the development of this industry. Cost drivers and deployment scenarios for broadband wireless networks are introduced by Tim Giles et al. In this paper, cost factors are presented in percentile, operator costs for wireless network are express as CAPEX and OPEX. In order to achieve low cost infrastructure of different scenarios, four strategies were given[14]. We think such analysis towards IoT network is helpful for its future development. Some more detailed cost analysis of wireless network has been offered. Bogdan Timus compared infrastructure costs of a hybrid cellular-multi-hop and a traditional single-hop cellular system using a linear cost model[15]. Holger Claussen et al also deliver financial analysis of a pico-cellular home network deployment[16]. A financial comparative study for microcellular and femtocell networks is given by Jan Markendahl and Östen Mäkitalo[17]6/8/2019 1:08:00 PM. From this paper, we can know which wireless network is more cost-efficient. A combined CAPEX and OPEX cost model are also applied to LTE network in Thomas Martin Knoll’s analysis. As for the sensor network, it is known that it distinguishes from mobile wireless network in some ways. Lifetime and energy efficiency are important parameters, to satisfy these demands, Zhao Cheng applied different cost models and

6

provided corresponding strategies[18]. This is also an essential influencing factor when delivering IoT services. A survey of data collection and wireless communication in IoT using economic analysis and pricing models has been given by Nguyen Cong Luong et al. They conclude a general architecture of an IoT system, various pricing models with their general objectives[19]. Actor roles of IoT communication has also be analyzed by Mohammad Istiak Hossain and Jan Markendahl in[20]. But cost structure analysis of IoT is still a blank area. This thesis is with the goal of filling the blank area. A similar techono-economic cost analysis is taken on heterogeneous wireless network by Usman Rauf Kamboh et al. They compared the cost and capacity performance of Micro, Pico, WLAN, Femto network with the motivation of finding the most cost-effective one[21]. In this thesis, cost performances of both licensed and unlicensed IoT network are calculated by CAPEX and OPEX models under diverse scenarios of different capacity need and coverage range. The methodology and tools stated in [22] can be a good reference when it comes to the dimensioning of LTE network.

1.3 Problem

As we can see from the related work, there is still a research gap here, no existing dimensioning and cost structure analysis towards IoT connectivity. However, for actors who want to enter this new market, knowing the method of calculating the cost structure can be helpful for them to plan their network deployment and budget, knowing the performance of existing technologies can be beneficial for them to make choices, knowing the comparison of different strategies is referable for them. By filling the gap, we provide another important perspective which can benefit the development of this IoT industry. With the popularization of IoT devices and applications, the economics of IoT is playing a more and more important part, it can have a great impact to the success of the development of IoT industry[23]. No matter what kind of technology it is, if we don’t think it from economics of the network, it cannot be well applied and used well in the market, which indicts the necessity of this analysis. Since the dimensioning of IoT network is different from LTE network because of its own characteristics, it is compulsory to bring out the method of it. Although the cost segments that contributes to the IoT network is similar to other wireless networks, the cost structure is still worth analyzing. The distribution and driver of the cost are the keys for network planning. Using the data assumptions as inputs, we are able to do the dimensioning and make a cost structure comparison analysis of cellular wide area IoT technologies. The quantity analysis of taking different strategies gives the hints of building IoT network based on current network. So, the problems we want to solve in this thesis are:

7

1. What is the method of dimensioning and cost structure calculation for IoT network? 2. What are the cost structure performances of different technologies and strategies? To answer of the first question, we use excel as a tool and the inductive approach. The calculation method of each steps is inducted from references of previous networks. The answer to this question is stated in chapter 3 Calculation method. Based on the calculation method brought out, the cost structure performances of SigFox, LoRaWAN, NB-IoT, LTE-M, and WiFi-HaLow are calculated.They are shown in chapter 4 analysis results. Besides, the strategies that can be taken by the actors of reusing and renting sites are compared.

1.4 Methodology

This thesis is using quantitative research method mostly, the method supports calculation to measure variables. And by using analytical research method, with assumptions and information that is already collected, we analyze the material to make a critical evaluation[24]. The research method aids in decision making in this IoT area. The way we used for data analysis is computational mathematics, which is used for calculating numerical methods. These are the methods used in this thesis. However, very few researches have focused on the dimensioning and cost structure of IoT network. That means there is no much references that can be followed directly to finish the analysis. The method proposed in Klas Johansson and Jan Markendahl’s methodology paper[25]6/8/2019 1:08:00 PM was adapted and enhanced in this thesis for the analysis. Also, based on the existing 3GPP standardization documents, our methodology was brought out to achieve the goal of this thesis which is shown in the Figure 3 Methodology Flow Chart.

8

Figure 3 Methodology Flow Chart

The pre-study part is a significant part while doing the thesis. The concept, principle of IoT connectivity need to be clarified. After understanding the characteristics for each technology, I reviewed some similar works for other wireless network technology. The specifications or parameters that play a role in the analysis are listed and understood. The above technology background is either mentioned in the introduction or technology chapter. Based on the assumed scenarios, the dimensioning of coverage and capacity can be calculated. Followed by the cost structure, it can be compared among the technologies and strategies. The calculation method that I summarized and deduced is explained in detail in the method chapter. And the results I get by using this method is stated and analyzed in the result chapter. This is the whole procedure of the thesis. Excel is the tool we choose for managing the data and doing the calculation. The reason of choosing Excel as the tool is because it is easy to change and maintain. Different functions are split in sheets of the Excel. In this way, every step of the calculation is very clear, and the outputs of each step can be compared. The other idea of this tool is that everything is linked, so that when the inputs change, the outputs will change automatically. All the analysis is delivered on this tool.

1.5 Outline

The outline of the thesis is as follows: Chapter 1 introduce the background, literature review, problem and methodology for this work. Chapter 2 described

9

the characteristics and specifications of the cellular and non-cellular technologies. Chapter 3 presents the method we brought out and used in the thesis. Chapter 4 illustrate the cost analysis results of the calculation under given scenarios. Chapter 5 is the conclusions, findings and potential future work of this analysis are stated.

10

2 Overview of Technologies

In various technologies to support IoT, NB-IoT and LTE-M are the two main licensed technologies. At the same time, 3GPP also released EC-GSM-IoT, it is an enhancement of EGPRS, together with PSM, which is the preparation of the GSM/EDGE for IoT market. As for the unlicensed technologies, LoRaWAN and SigFox are competing with each other, which are both efficient solutions to connect IoT devices. In order to play a role in the rapidly growing market, IEEE 802.11ah, also called WiFi-HaLow, is designed to provided extended range Wi-Fi network but with higher data rates by WiFi alliance. These six technologies are chosen because they are the representative ones and worth comparing. In this section, more detailed descriptions of them will be introduced. The technology specifications are collected form literature review[9], [26], [27], standardization specifications [28] [29] and white papers[30] [31]. Key characteristic parameters are listed in the Table 1, which will be used in the cost structure calculation in the following chapter.

Table 1 Specifications of Technologies

Parameters SigFox LoRaWAN NB-IoT LTE-M EC-GSM-IoT

WiFi-HaLow

Frequency Band (MHz) 868 868 868 700 940 900

Receiver Sensitivity (dbm) 164 154 150 146 150 146

Registered Device Capacity/cell 100000 10000 150000 90000 150000 8191

Spectrum (kHz) 200 1175 180 1080 600 1000

Modulation D-BPSK FSS/CSS Chrip OFDMA OFDMA OFDMA OFDMA

BW per message or channel (Hz) 100 125000 15000 180000 200000 -

Spacing(kHz) 0 200 3.75 15

UL Payload (Bytes) 12 51 125 1000 150 256

DL Payload (Bytes) 8 14 125 1000 150 256

Data Rate(bps) 100 1760 50000 1000000 24000 300000

Duty Cycle/ Tx Restriction 140 msg/day 1%-10% - - - -

Number of UL channel or subbands 25 3 12 6 3 1

Control traffic 0 0 0.4 0.05 0.05 -

Bidirectional HalfDuplex HalfDuplex HalfDuplex FullDuplex HalfDuplex FullDuplex

Number of Radio Unit per site 3 1 3 3 3 1

2.1 SigFox

SigFox is a start-up company and an LPWAN operator which developed the SigFox technology in 2010 in Toulouse, France[9]. “Ultra-narrowband” signal is used to reach wide range and pass through solid objects. SigFox offers an end-to-end IoT connectivity solution based on its own technology or with other

11

network operators. It deploys proprietary base stations and connect them to the back-end servers. The end-devices can connect to the base stations using binary phase-shift keying (BPSK) modulation. It uses unlicensed ISM bands, for example, 868 MHz in Europe, 915 MHz in North America, and 433 MHz in Asia[9]. By making use of the ultra-narrow band, it has very low power consumption, high receiver sensitivity, and low-cost antenna design because of the efficiency and little noise. But SigFox has its limitation, the number of messages over the uplink is up to 140 per day, while the number of messages over the downlink is up to 4 per day[9]. Besides, the maximum payload length for each uplink message is 12 bytes.

2.2 LoRaWAN

According to LoRa Alliance, LoRa is the physical layer or the wireless modulation utilized to create the long-range communication link[32]. Based on chirp spread spectrum modulation, LoRa have the same low power characteristic as FSK modulation but increase the communication range at the same time. LoRaWAN is a different from LoRa, it defines the communication protocol and system architecture for the network while the LoRa physical layer enables the long-range communication link[32]. The battery lifetime of a node, the network capacity, the quality of service, the security, and the variety of applications served by the network can be influenced by the protocol and network architecture. In a LoRaWAN network, data from a node will typically transmit to multiple gateways. Every gateway will then forward the received package from the end-device to the cloud-based network server through backhaul. The server will filter redundant packets, perform security checks, schedule acknowledgements, etc. LoRaWAN can save a lot battery compared to other LPWAN technologies, because it communicate when they have data ready to send whether event driven or scheduled and it is asynchronous [32]. LoRaWAN also utilized adaptive data rate to gain high capacity and scalability. As for the security, there are two layers of security for LoRaWAN, the application layer of security guarantee the network operator cannot access the end user’s application data while the network security ensures authenticity of the node in the network[32]. In addition, LoRaWAN provide three different device classes. Class A is end-devices that enable bidirectional communications, their uplink transmission is followed by two short downlinks receive windows. Class B is bi-directional end-devices with scheduled receive slots, adding extra receive windows at scheduled times. Class C is bi-directional end-devices of maximal receive windows, closed only when transmitting. The LoRaWAN specification varies slightly from region to region based on the different regional spectrum allocations and regulatory requirements. And the LoRa Alliance is continuously working on the next version.

12

2.3 NB-IoT

To enable a wide range of cellular devices and services, Narrowband IoT (NB-IoT) is a licensed LPWAN radio technology standard released by 3GPP[33]. There are other two technologies in release 13 by 3GPP, eMTC (LTE Cat M1) will enhance LTE, EC-GSM-IoT is designed to improve GSM. However, NB-IoT can be considered as a new track based on the existing 3GPP technologies. An advantage of cellular solutions compared to other LPWAN technologies is that they don’t have the duty cycle regulations, because of operating on licensed bands. In NB-IoT, to optimize the low end of the market, new radio is added to the LTE platform [34]. The objectives of releasing NB-IoT is to offer even lower cost than eMTC as well as extended coverage(164dB). NB-IoT can also support long battery life (10 years) and massive number of devices (50000 per cell). It is optimized in some way, such as reduced data rate/bandwidth, mobility support and further protocol. There are 3 modes of operation of NB-IoT:

• Stand-alone: utilizing stand-alone carrier. It is deployed in a stand-alone spectrum of 200kHz. All transmission power is consumed at the base station to increase coverage[10].

• Guard band: utilizing the unused resource blocks within an LTE carrier's guard-band. It is co-located with LTE cell and shares the transmission power.

• In-band: utilizing resource blocks within a normal LTE carrier[34]. Wideband LTE and NB-IoT share the transmit power at the base station.

As for the features of physical layer in the OSI model (PHY), NB-IoT is using narrow band and can support 180kHz. There are two modes for uplink, one is single, while the other one is multiple tone. But it doesn’t support Turbo code f0or the downlink. Single transmission mode of SFBC for PBCH, PDSCH, PDCCH is also supported. 3GPP continuously try to improve NB-IoT in new releases. It is said it is to be extended to include localization methods, multicast services, mobility, etc.[9]

2.4 LTE-M

LTE-M is one of the candidates to support M2M communications in Long Term Evolution (LTE) cellular networks. Many cellular operators and companies such as Nokia, Ericsson and Qualcomm chose LTE-M to deliver IoT services by optimizing LTE. LTE-M is the short for LTE Cat M1. 3GPP previously released LTE Category 0 for MTC, LTE-M is improved based on it.

13

LTE-M also supports long battery life (10 years), low device cost and extended coverage(155.7dB). One more thing, LTE-M offers variable rates, from 10kbps to 1Mbps depending on coverage needs[34]. The deployment for LTE-M is easier because it is based on current LTE network. It can coexist with other LTE services within the same bandwidth and be deployed in any LTE spectrum. FDD, TDD and half duplex (HD) modes are offered[34]. The most convenient and cost-efficient feature is, it can reuse existing LTE base stations only with software update. As for its PHY features, it is of narrowband operation with 1.08MHz bandwidth[34]. There are a lot of main enhanced features over LTE-M, it supports positioning (E-CID and OTDOA), multicasting (SC-PTM) and voice over LTE(VoLTE).

2.5 EC-GSM-IoT

EC-GSM-IoT, also known as EC-GSM, is another LPWA technology in development by 3GPP. It is designed as an enhancement to GSM, and most of the current GSM design are reused but some changes are made to meet the requirements of LPWA[10]. EC-GSM-IoT is with the objectives of offering long battery life, low device cost compared to GPRS/GSM devices, extended coverage and variable rates as well. It can support massive number of devices and improve security compared to GSM/EDGE. EC-GSM-IoT provides new logical channels that is designed for extended coverage and overlaid CDMA to increase cell capacity[34]. EC-GSM-IoT has been futher enhanced in the 14th release of 3GPP. Radio interface enhancements for EC-GSM-IoT is specified, which support positioning. It has made at least 3 dB MCL improvement for low power devices on all uplinks. Alternative mappings of blind physical layer transmissions for higher coverage classes is used. Actually, because EC-GSM-IoT is based on GSM network, it can be updated via software. After softeare update, support for new devices can be achieved in exsiting GSM deployments[10].

2.6 WiFi-HaLow

In order to support IoT applicaitions, a new Wi-Fi standard also known as WiFi-HaLow (or IEEE 802.11ah) was introduced by Wi-Fi alliance. Many MAC features are added to support large number fo devices, extended range of operation and less energy consumption compared to existing Wi-Fi standards[35].

14

As for the its specifications, it can support up to 8191 devices associated with an access point (AP). It is operated over sub-1 GHz carrier freguencies. Up to 1 km transmission range in outdoor areas is supported. The data rates of WiFi-HaLow is higher than other LPWA technologies, it reaches at least 100 Kbps. Definitly, it can be a cost-effective solution with very low energy consumption[35]. However, devices operated in the Sub-1 GHz ISM bands must comply with the maximum duty cycle limit of 2.8% in Europe. They can also support Liste Before Talk(LBT) and Adaptive Frequency Agility(AFA) features[11]. In order to optimize long battery time and a large number of STAs, some features was designed for WiFi-HaLow espcially. It offers short frame format, short control/management, asymmetric and bi-directional transmissions which are more efficient. It reduces the power consumption by Non-TIM operation, target wake time (TWT) mechanism and extended sleeping and listen interval[36].

15

3 Calculation method

As we have mentioned before, there is no existing method we can use directly for the IoT network. We get this method by enhancing the existing methods of other wireless network and use it to get the results, which is divided into three parts as the Figure 4. They are scenario assumption, network dimensioning and cost structure calculation. In scenario assumption parts, all the possible scenarios are listed, especially their diverse parameters we have considered. When it comes to network dimensioning, both coverage dimensioning, and capacity dimensioning are needed, the limiting factors are coverage range, number of devices, data volume and number of messages. All inputs coming from the scenario assumption lead to the number of sites as the output. As for the cost structure, the data of both scenario assumption and networking dimensioning contribute to this part. The cost specification and economic concepts are brought in when calculating the cost structure.

Figure 4 Analysis Method Overview

The cost analysis result using the method is shown in the next section.

3.1 Scenario Assumption

When we consider the scenarios, we would like to bring out a general scenario model. In this model, the scenario can be divided into two dimensions, one is capacity, the other one is coverage, as in Figure 5. Four kinds of scenarios are generated according to these two dimensions. And, you can find matched real use cases to these general scenarios. For example, smart city is like the small coverage but large capacity scenario, while connected cars might be the large coverage but small capacity scenario.

16

Figure 5 Scenario Partition

When we dig into how different parameters are related to these two dimensions, we find out that there are four factors in the network. The number of elements within the assumed network is limited by them. Only the maximum value of the number of elements got from the factors can achieve all the requirement of the network, which is taken for the following calculation. They are coverage area, device density, number of transmitted message and message payload. In this way, we can cover the boundary use cases. Some may be very common, while others might be rare in the reality. And in the cost structure calculation, what is needed is to input the required parameters. When we think from the network confinity, there are four scenarios as the above. However, the actors in the market of this area should also be taken into consideration. Different actors in the industry may take different strategies. For the incumbent actors, they can reuse already exiting sites, if the exiting sites cannot fulfil the network requirements, they can either build or lease more sites. However, for the new entrants, there are no sites for them to reuse.

3.2 Network Dimensioning

To estimate the required number of radio base stations needed to support a specified traffic load in a certain area is the purpose of dimensioning[37]. Based on the assumed scenario, network dimensioning can be delivered. After dimensioning, we can calculate the cost of building a network of the estimated number of base stations. The number of the base stations should fulfill the requirements of certain scenarios from two aspects, both the coverage and the capacity, which is

17

NumberSites= Maximum { NumberSitesCoverage, NumberSitesCapacity}

(1) Coverage Dimensioning The coverage dimensioning is to get the number of sites by dividing the deployment area according to site coverage. The unit within the network is called a cell. There are three types of cell deployment strategies among the technologies as the Figure 6. LoRaWAN and WiFi-HaLow usually are of Omnidirectional antennas. Equal radio power is radiated in all directions. However, SigFox choose to deploy three- sector cell, the sectors are at the corners of the cell. For SigFox, a message packet from a device in a cell shall be transmitted to three nearby base stations. Then they are forwarded to OSS/BSS, who will decide and keep only one packet[38]. This kind of deployment limit the transmitted number of messages but guarantee the delivery. QoS, link availability and accessibility rate are needed to be improved based on this structure. Cellular-IoT technologies consider the sectorized cell, which can improve capacity and coverage performance. The three-sector cell is not enclosed in a single hexagon, each cell is represented by its own hexagon. So actually, sectors and cells are the same thing in a three-sector situation[39].

Figure 6 Three types of sites (Omni directional, 3-RAT null-sector, Three-

sector)[38]

So, the cell of all the three types can be calculated as a hexagon. From the specifications in the references, we can find out the coverage range of each technology as shown in the Table 2. The site area is the hexagon size, the coverage range is the radius of the hexagon. So, the formula is:

𝑆𝑖𝑡𝑒𝐴𝑟𝑒𝑎 =3√3

2𝑑2

the result of which is also displayed in Table 2.

Table 2 Coverage range [40]

SigFox LoRaWAN NB-IoT LTE-M EC-GSM-IoT WiFi-HaLow

Coverage(km) 13 11 15 11 15 1

18

Area(km2) 439 314 585 314 585 3

The number of sites can be easily calculated from site area when inputting the assumed deployment area.

𝑁𝑐𝑜𝑣𝑒𝑟𝑎𝑔𝑒 =𝐷𝑒𝑝𝑙𝑜𝑦𝑚𝑒𝑛𝑡𝐴𝑟𝑒𝑎

𝑆𝑖𝑡𝑒𝐴𝑟𝑒𝑎

(2) Capacity Dimensioning In this case for IoT connectivity network capacity, we consider from three perspectives, device capacity, data capacity and message transmission capacity. Since IoT services consume very a small amount of data, using the data capacity as a boundary condition may not be enough. So, we consider message transmission capacity at the same time. The network should cover all the devices and support the message transmission of certain data within a day in the area. So, the number of sites that can guarantee the capacity should be the maximum number of elements calculating from these three perspectives.

𝑁𝑐𝑎𝑝𝑎𝑐𝑖𝑡𝑦= Maximum {𝑁𝑑𝑒𝑣𝑖𝑐𝑒, 𝑁𝑑𝑎𝑡𝑎, 𝑁𝑇𝑥𝑀𝑠𝑔}

1) Device Capacity A radio station can only support a certain number of devices connected to it. The number of it is the device capacity. The registered device capacity per cell is shown in the table in Section 2 Technology.

𝑁𝑑𝑒𝑣𝑖𝑐𝑒=

𝐷𝑒𝑣𝑖𝑐𝑒𝑁𝑢𝑚𝑏𝑒𝑟

𝐶𝑑𝑒𝑣𝑖𝑐𝑒

2)Data Capacity Data capacity is the data consumed per day per site. The formula of it for uplink is,

𝐶𝑑𝑎𝑡𝑎 = 𝑁𝐶ℎ𝑎𝑛𝑛𝑒𝑙𝑠 ∗ 𝑅𝑑𝑎𝑡𝑎 ∗ (1 − 𝑓𝑑𝑒𝑔𝑟𝑎𝑑𝑎𝑡𝑖𝑜𝑛) ∗ 𝑁𝑠𝑒𝑐𝑡𝑜𝑟 ∗ 𝑡𝑠[38]

𝑁𝐶ℎ𝑎𝑛𝑛𝑒𝑙𝑠 is the number of channels allocated for uplink transmission, 𝑅𝑑𝑎𝑡𝑎 is the max data rate can be obtained per channel, 𝑓𝑑𝑒𝑔𝑟𝑎𝑑𝑎𝑡𝑖𝑜𝑛 is the coefficient to

determine the rate of data rate degradation, 𝑁𝑠𝑒𝑐𝑡𝑜𝑟 is the number of sectors per site, 𝑡𝑠 is the time of a day in seconds which is 24h*3600s[38].

𝑁𝑑𝑎𝑡𝑎=

𝐷𝑎𝑡𝑎𝑉𝑜𝑙𝑢𝑚𝑒

𝐶𝑑𝑎𝑡𝑎

19

3)Message Transmission Capacity It is the number of messages that can be transmitted to the radio station within one day, which can be calculated as “time on air”.

𝐶𝑇𝑥𝑀𝑠𝑔 =𝐶𝑑𝑎𝑡𝑎 ∗ (1 − 𝑃𝑐𝑐) ∗ 𝑆𝑝𝑎𝑐𝑘𝑒𝑡

𝐷𝑎𝑡𝑎 𝑝𝑎𝑦𝑙𝑜𝑎𝑑 + 𝑂𝑣𝑒𝑟𝑙𝑜𝑎𝑑[38]

𝑃𝑐𝑐 is the percentage of resources allocated for the control channel and 𝑆𝑝𝑎𝑐𝑘𝑒𝑡 is the

packet success rate for a certain use case[38].

𝑁𝑇𝑥𝑀𝑠𝑔=

𝑀𝑒𝑠𝑠𝑎𝑔𝑒𝑁𝑢𝑚𝑏𝑒𝑟

𝐶𝑇𝑥𝑀𝑠𝑔

3.3 Cost structure Calculation

The last step is to do the cost structure calculation to answer our research question. The cost segments of CAPEX and OPEX used in the tool are listed in the Table 3. The cost data are from three sources, the cost assumption of NB-IoT, LTE-M and EC-GSM-IoT are from METIS-II[9][41]. And we took the cost of SigFox, LoRaWAN and WiFi-HaLow from[42]. Capital expenditure or capital expense (CAPEX) is the money a company or a business spent to buy, maintain, or improve its fixed assets, such as buildings, vehicles, equipment, or land[43]. So, equipment cost, installation cost, spectrum cost and transmission cost are considered as CAPEX. An operating expense, operating expenditure, operational expense, operational expenditure or OPEX is an ongoing cost for running a product, business, or system[44]. Site lease cost, electricity cost, transmission cost, operation and maintenance cost are considered as parts of OPEX. And we don’t consider spectrum cost here in the analysis.

Table 3 Cost Assumptions

SigFox LoRaWAN NB-IoT LTE-M EC-GSM-

IoT WiFi-

HaLow

Equipment Cost(K€) 4 1 10 6 10 1

Installation Cost(K€) 6 2 10 10 10 2

Spectrum Cost (K€/kHz/site) 0 0 0 0.001 0.001 0

Transmission Installation Cost(K€) 0.5 4 0 0 0 1

Site Build Cost (K€) 10 2 20 20 20 1

Site Lease (K€/year) 1 0.4 1 1 1 0.4

Electricity Cost (K€/year) 1 1 1 1 1 0.1

Transmission Cost (K€/year) 0.12 0.1 0.1 0.1 0.1 0.1

Operation&Maintenance Cost 10%-15% of CAPEX

10%-15% of CAPEX

5%-10% of CAPEX

5%-10% of CAPEX

5%-10% of CAPEX

20%-25% of CAPEX

20

CAPEX= (EquipmentCost+ InstallationCost+ TransmissionInstalltionCost+ SiteBuildCost) * NumberSites+ SpectrumCost* Bandwidth OPEX= (SiteLeaseCost+ElectricityCost+TransmissionCost) * NumberSites+ OperationMaintenanceCost*CAPEX

The end-device cost is a large part in the IoT connectivity network cost structure especially the number of devices is increasing rapidly nowadays. The cost of the connectivity module in the end-device is shown in the Figure 7. It’s easy to notice that for cellular-IoT system, the cost is higher, while the cost is the cheapest for the LPWAN network. This should also be considered when it comes to the real use case. If the device density is large, cheaper end-device could be a better choice. However, as a network operator, it’s their own business strategy to sale the end-device or not. If they only provide the network service, the end-device is not within their scope. In this thesis, the focus is on the network layer, so the end-device cost isn’t included in the cost structure.

Figure 7 End-Device Cost

0

2

4

6

8

10

12

14


End-device Cost(€)

21

4 Analysis Results

The results by using the method stated in the previous section is shown in this chapter. The assumed scenarios are mentioned. The dimensioning and cost structure results as well as the CAPEX/OPEX and leasing/building sites comparison are also analyzed in the following.

4.1 Scenarios

As we mentioned in the scenario assumption part, the scenario can be assumed from two dimensions and totally four parameters should be the inputs. We have made the assumptions with the aim of finding the boundaries as in the Table 4. We take all the six technologies described above to the analysis.

Table 4 Scenario List

Scenario Deployment Area (Km2)

Device Density (Number/Km2)

Message (Number)

Message Payload (Bytes)

1 Small Coverage+ Small Capacity 100 500 10 12

2 Small Coverage+ Large Capacity 100 5000 100 300

3 Large Coverage+ Small Capacity 10000 500 10 12

4 Large Coverage+ Large Capacity 10000 5000 100 300

The device penetration rate is set to 5% at year 0 then it increases based on the CAGR which is set to 57% in our assumptions. From these, we can get the device penetration rate each year as Table 5. But this parameter is an educated assumption, it shall change in the reality, dimensioning and cost structure calculation results may be affected.

Table 5 Device Penetration Rate 2020 2021 2022 2023 2024 2025 2026 2027 2028 2029 2030

Device Penetration Rate

5% 9% 16% 27% 48% 85% 150% 265% 467% 823% 1450%

4.2 Dimensioning Results

Before going into details of the cost, we can look at the dimensioning result to find out how these four scenarios are limited by the parameters in Figure 8-11. The horizontal axis represents the different technologies, and the vertical axis represents the required number of cells.

22

Figure 8 Dimensioning of Scenario 1


When the coverage is small, it is obvious that under the small capacity cases, the technologies are limited by the supported number of devices. WiFi-HaLow requires the largest number of cells, LoRaWan has slightly lower number. SigFox and other three cellular technologies are in need of small number of cells under this scenario, because they can support large number of devices per site. However, SigFox and LoRaWAN are limited by the number of messages transmitted and data when it comes to large capacity and needs largest number of cells. It is because their data rates are small, especially SigFox, whose is 100 bps, the smallest among the six technologies. Meanwhile, the data limited number of cells is the same as event limited number of cells, the reason is the message payload is multiple times of the payload SigFox supported. WiFi-HaLow are limited by supported devices, which is based on its low device

0

10

20

30

40

50

60

70

80

90

100


Dimensioning of Scenario 1

Coverage Limited Device Limited Data Limited Events Limited

0

1000

2000

3000

4000

5000

6000

7000

8000

9000




23

capacity, high data rate and message payload characteristics. For Cellular-IoT technologies, they have the best performance, while EC-GSM-IoT is limited by data capacity among them.



We can find out scenario 1 is similar to scenario 3 and scenario 2 is similar to scenario 4. This indicates that the two large coverage scenarios share the similar trends as the two small coverage scenarios, only the number of cells is larger. The different comparison results are caused by the capacity instead of coverage.

0

1000

2000

3000

4000

5000

6000

7000

8000

9000

10000




0

100000

200000

300000

400000

500000

600000

700000

800000

900000




24

4.3 Cost Structure Results

With the acknowledge of the number of sites, we can calculate the cost including both CAPEX and OPEX. Figure 12-15 show all the cost segments for a new market entrant building the network under scenario 1-4. The horizontal axis represents the different technologies, and the vertical axis represents the required cost.

Figure 12 Cost of Scenario 1

For scenario 1, we can find that the cost is somehow follow the same tendency of number of cells, except LoRaWAN cost more than WiFi-HaLow because of large transmission cost. SigFox is of best performance, which means it is suitable for the small capacity scenarios. LoRaWAN cost the most in this case. Although the number of sites of NB-IoT, LTE-M and EC-GSM-IoT are small, their site building cost is a large part. The operation and maintenance cost are extremely enormous which contribute to its overall large cost. It is obvious that under this low coverage and low capacity scenario, the transmission cost of LoRaWAN and the maintenance cost of WiFi-Low becomes the main cost driver. SigFox is the most cost-efficient technology under this scenario.

0

200

400

600

800

1000

1200


Cost of Scenario 1 (K€)

Equipment Cost Installation Cost

Transmission Cost Site Build Cost

Electricity Cost Operation&Maintenance Cost

25


However, for scenario 2, SigFox has a largest amount of cost due to the largest number of sites which leads to the worst performance under large capacity scenario. Site build cost becomes one large part of the total cost. LoRaWAN is also of bad cost performance. Cellular-IoT technologies and WiFi-HaLow are of good performance, especailly LTE-M, which make them a good choice.


0

50000

100000

150000

200000

250000

300000


Cost of Scenario 2(K€)




0

20000

40000

60000

80000

100000

120000






26


For the cost structure comparison of scenario 3 and 4, they are similar to that of scenario 1 and 2. We can draw a conclusion here, the cost performance is largely dependent on the capacity. If two scenarios share the same capacity conditions, the comparison results between the six technologies are more or less the same.

4.4 CAPEX/OPEX Comparison

Figure 14-17 shows the CAPEX and OPEX comparison results for scenario 1-4. The horizontal axis represents the different technologies, and the vertical axis represents the required cost.

Figure 16 CAPEX and OPEX Comparison for Scenario 1

0

5000000

10000000

15000000

20000000

25000000

30000000






0

100

200

300

400

500

600

700


CAPEX and OPEX Comparison for Scenario 1(K€)

CAPEX OPEX

27


When we compare the CAPEX and OPEX under scenario 1, CAPEX is always larger than OPEX except for WiFi-HaLow. WiFi-HaLow is somehow special here because of its large operation and maintenance cost. LoRaWAN has the most CAPEX, while SigFox has the least CAPEX. For OPEX, WiFi-HaLow cost the most and EC-GSM-IoT cost the least. For scenario 2, CAPEX cost more than OPEX of any technologies. SigFox are of the largest CAPEX, while LTE-M are of the smallest CAPEX. It is the same for OPEX. These are the two low coverage cases. We can find out the total cost are mostly defined by the CAPEX.


0

20000

40000

60000

80000

100000

120000

140000

160000

180000



CAPEX OPEX

0

10000

20000

30000

40000

50000

60000

70000



CAPEX OPEX

28


For the two large coverage cases, the CAPEX and OPEX comparison results are similar to the two small coverage cases.

4.5 Leasing/Building Comparison

From Figure 20 and 21, it is obvious that leasing sites is always cheaper than building sites especially for cellular-IoT technologies. The horizontal axis represents the different technologies, and the vertical axis represents the required cost.

Figure 20 Site Leasing Cost Compared to Site Building Cost of Scenario 1

0

2000000

4000000

6000000

8000000

10000000

12000000

14000000

16000000

18000000



CAPEX OPEX

0

50

100

150

200

250

300

350


Site Leasing Cost Compared to Site Building Cost of Scenario 1 (K€)

Site Build Cost Site Lease Cost

29


But there is one exception, WiFi-HaLow cost more when leasing from others under low capacity cases. Even under high capacity cases, leasing and building has nearly the same cost. It is because the leasing cost is not largely less than building cost. And when paying the building cost, only the additional sites counts every year, but the total number of sites need to be paid the leasing fee. Under scenario 1, the site building cost is the most for NB-IoT and least for SigFox. WiFi-HaLow cost the most and NB-IoT cost the least when leasing. This indicates that leasing sites can largely save the cost and so as other cellular technologies. And under scenario 2, the site building cost of SigFox is the most and which of WiFi-HaLow is the least. As for the site leasing cost, SigFox is still the most, NB-IoT is the least.


0

10000

20000

30000

40000

50000

60000

70000

80000

90000




0

5000

10000

15000

20000

25000

30000

35000




30


Figure 22 and 23 illustrates the leasing and building cost comparison under scenario 3 and 4. As the analysis above, these two scenarios are of similar results.

Table 6 Reusable Site Number[38]

SigFox LoRaWAN NB-IoT LTE-M EC-GSM-IoT WiFI-HaLow

Reuseable site number 20 39 26 39 26 77

The above calculation is for a new market entrant. However, in the telecom industry, there are already a lot of incumbent actors, they can reuse existing sites which can save the cost and is eco-friendlier. The number of sites that can reuse for 100 𝑘𝑚2is shown in the Table 6.

0

1000000

2000000

3000000

4000000

5000000

6000000

7000000

8000000

9000000




31

5 Conclusions

In this thesis, the calculation method is proposed as well as the cost performance is compared of LPWAN, LPLAN and cellular-IoT. Firstly, we bring out the calculation method from scenario assumption, network dimensioning to cost structure calculation, which is one main contribution. In the scenario assumption part, the two dimensions, coverage and capacity, is used to divide the scenarios. Based on the assumed scenarios, the dimensioning is carried out also from these two aspects. The network is required to meet all the demands. The segments of the cost structure states of the cost of deploying IoT network, the calculation method is also introduced. By using the method, knowing the input, we can get the output. If we can get more accurate data, the more accurate performance analysis can be delivered. The dimensioning results show under low capacity, each technology is limited by the supported number of devices per site, which can be one of improvements can be made. Unlicensed technologies, SigFox and LoRaWAN are largely limited by data rate causing the poor performance under high capacity scenarios. As for the cost analysis, one key finding is that operation and maintenance cost become one cost driver of the total cost, especially for WiFi-HaLow. From a long-term perspective, actors should focus on reduce the maintenance and operation cost. And for Cellular-IoT technologies, site building cost is a large part. So, reusing and leasing sites seems like a good choice for actors applying them. The other finding is that site leasing is always cheaper than building own sites except for WiFi-HaLow. And incumbent actors have advantages over new market entrant since they can reuse a large number of existing sites. SigFox has the best performance under low capacity case, while it is of the worst performance under high capacity cases. This indicates SigFox is more suitable for low capacity case because its data rate and payload limitation. LoRaWAN are of poor performance overall. Cellular-IoT technologies has very good performance under high capacity cases, and since its cost can be largely deduced by leasing and reusing, it can be an even better choice. WiFi-HaLow is some kind of special among all these six technologies, it is of poor performance for low capacity cases and better performance for high capacity cases. And its OPEX is larger than CAPEX, because of operation and maintenance cost. Meanwhile it cannot benefit from leasing sites from others. There is another interesting finding, when other conditions are the same, changing the coverage cannot influence the comparison results, only causing larger cost. For future study, the platform layer of the IoT network can be analyzed. One should focus on the cost breakdown of a platform and the core network to get a more comprehensive cost structure of the IoT communication systems. More interviews can be carried on getting a touch to the market.

32

References

[1] “Y.2060 : Overview of the Internet of things.” [Online]. Available: https://www.itu.int/rec/T-REC-Y.2060-201206-I. [Accessed: 26-Jun-2018].

[2] U. Raza, P. Kulkarni, and M. Sooriyabandara, “Low Power Wide Area Networks: An Overview,” IEEE Commun. Surv. Tutor., vol. 19, no. 2, pp. 855–873, Secondquarter 2017.

[3] “Ericsson Mobility Report June 2018 – Ericsson,” Ericsson.com, 04-Jun-2018. [Online]. Available: https://www.ericsson.com/en/mobility-report/reports/june-2018. [Accessed: 26-Jun-2018].

[4] E. AB, “Cellular networks for Massive IoT – enabling low power wide area applications,” p. 13.

[5] D. Niyato, X. Lu, P. Wang, D. I. Kim, and Z. Han, “Economics of Internet of Things: an information market approach,” IEEE Wirel. Commun., vol. 23, no. 4, pp. 136–145, Aug. 2016.

[6] “White paper-IoT platforms: The central backbone for the Internet of Things,” Nov-2015. [Online]. Available: http://iot-analytics.com/wp/wp-content/uploads/2016/01/White-paper-IoT-platforms-The-central-backbone-for-the-Internet-of-Things-Nov-2015-vfi5.pdf. [Accessed: 28-Oct-2018].

[7] Keysight Technologies, “The Internet of Things: Enabling Technologies and Solutions for Design and Test.” [Online]. Available: http://www.datatec.de/shop/artikelpdf/5992-1175en_d.pdf. [Accessed: 27-Jun-2018].

[8] “What is Cost Structure? - Definition, Types & Examples - Video & Lesson Transcript,” Study.com. [Online]. Available: http://study.com/academy/lesson/what-is-cost-structure-definition-types-examples.html. [Accessed: 30-Oct-2018].

[9] K. Mekki, E. Bajic, F. Chaxel, and F. Meyer, “A comparative study of LPWAN technologies for large-scale IoT deployment,” ICT Express, Jan. 2018.

[10] J. Finnegan and S. Brown, “A Comparative Survey of LPWA Networking,” ArXiv180204222 Cs, Feb. 2018.

[11] M. Qutab-ud-din et al., “Duty Cycle Challenges of IEEE 802.11ah Networks in M2M and IoT Applications,” in European Wireless 2016; 22th European Wireless Conference, 2016, pp. 1–7.

[12] S. Dawaliby, A. Bradai, and Y. Pousset, “In Depth Performance Evaluation of LTE-M for M2M Communications,” in The 12th IEEE International Conference on Wireless and Mobile Computing, Networking and Communications, WiMob’16, New York, United States, 2016.

[13] J. Zander, “On the cost structure of future wideband wireless access,” in Technology in Motion 1997 IEEE 47th Vehicular Technology Conference, 1997, vol. 3, pp. 1773–1776 vol.3.

[14] T. Giles et al., “Cost drivers and deployment scenarios for future broadband wireless networks - key research problems and directions for research,” in 2004 IEEE 59th Vehicular Technology Conference. VTC

33

2004-Spring (IEEE Cat. No.04CH37514), 2004, vol. 4, pp. 2042-2046 Vol.4.

[15] B. Timus, “Cost analysis issues in a wireless multihop architecture with fixed relays,” in 2005 IEEE 61st Vehicular Technology Conference, 2005, vol. 5, pp. 3178-3182 Vol. 5.

[16] H. Claussen, L. T. W. Ho, and L. G. Samuel, “Financial Analysis of a Pico-Cellular Home Network Deployment,” in 2007 IEEE International Conference on Communications, 2007, pp. 5604–5609.

[17] J. Markendahl and Ö. Mäkitalo, “A comparative study of deployment options, capacity and cost structure for macrocellular and femtocell networks,” in 2010 IEEE 21st International Symposium on Personal, Indoor and Mobile Radio Communications Workshops, 2010, pp. 145–150.

[18] Z. Cheng, M. Perillo, and W. B. Heinzelman, “General Network Lifetime and Cost Models for Evaluating Sensor Network Deployment Strategies,” IEEE Trans. Mob. Comput., vol. 7, no. 4, pp. 484–497, Apr. 2008.

[19] N. C. Luong, D. T. Hoang, P. Wang, D. Niyato, D. I. Kim, and Z. Han, “Data Collection and Wireless Communication in Internet of Things (IoT) Using Economic Analysis and Pricing Models: A Survey,” IEEE Commun. Surv. Tutor., vol. 18, no. 4, pp. 2546–2590, Fourthquarter 2016.

[20] M. I. Hossain and J. Markendahl, “IoT-communications as a service: Actor roles on indoor wireless coverage,” in 2017 Internet of Things Business Models, Users, and Networks, 2017, pp. 1–7.

[21] U. R. Kamboh, Q. Yang, M. Qin, and S. Rauf, “A techno-economic cost analysis of heterogeneous wireless network,” in 2017 IEEE 2nd Advanced Information Technology, Electronic and Automation Control Conference (IAEAC), 2017, pp. 1645–1650.

[22] A.-B. Syed, “Dimensioning of LTE Network,” p. 71. [23] D. Niyato, X. Lu, P. Wang, D. I. Kim, and Z. Han, “Economics of Internet

of Things (IoT): An Information Market Approach,” ArXiv151006837 Cs, Oct. 2015.

[24] A. Håkansson, “Portal of Research Methods and Methodologies for Research Projects and Degree Projects,” presented at the The 2013 World Congress in Computer Science, Computer Engineering, and Applied Computing WORLDCOMP 2013; Las Vegas, Nevada, USA, 22-25 July, 2013, pp. 67–73.

[25] K. Johansson, J. Markendahl, and P. Zetterberg, “Relaying access points and related business models for low cost mobile systems,” p. 14.

[26] R. Akeela and Y. Elziq, “Design and verification of IEEE 802.11ah for IoT and M2M applications,” in 2017 IEEE International Conference on Pervasive Computing and Communications Workshops (PerCom Workshops), 2017, pp. 491–496.

[27] F. Adelantado, X. Vilajosana, P. Tuset-Peiro, B. Martinez, J. Melia-Segui, and T. Watteyne, “Understanding the Limits of LoRaWAN,” IEEE Commun. Mag., vol. 55, no. 9, pp. 34–40, Sep. 2017.

[28] A. Hoglund et al., “Overview of 3GPP Release 14 Enhanced NB-IoT,” IEEE Netw., vol. 31, no. 6, pp. 16–22, Nov. 2017.

[29] LoRa Alliance Technical Committee, “‘LoRaWAN 1.1 Specification.’ LoRa Alliance, Standard 1 (2017): 1.” [Online]. Available: https://lora-

34

alliance.org/sites/default/files/2018-04/lorawantm_specification_-v1.1.pdf. [Accessed: 13-Apr-2019].

[30] SigFox, “Sigfox technical Overview”, 2017. [Online]. Available: https://www.disk91.com/wp-content/uploads/2017/05/4967675830228422064.pdf. [Accessed: 13-Apr-2019].

[31] Semtech, “A. N. ‘120022.’ LoRa Modulation Basics,” 2015. [Online]. Available: https://www.semtech.com/uploads/documents/an1200.22.pdf. [Accessed: 13-Apr-2019].

[32] LoRa Alliance, “What is LoRaWAN.” [Online]. Available: https://www.lora-alliance.org/sites/default/files/2018-04/what-is-lorawan.pdf. [Accessed: 08-Jul-2018].

[33] “3GPP Low Power Wide Area Technologies White Paper.” [Online]. Available: https://www.gsma.com/iot/wp-content/uploads/2016/10/3GPP-Low-Power-Wide-Area-Technologies-GSMA-White-Paper.pdf. [Accessed: 15-May-2018].

[34] “3gpp Standards for IoT.” [Online]. Available: http://www.3gpp.org/images/presentations/2016_11_3gpp_Standards_for_IoT.pdf. [Accessed: 12-Jun-2018].

[35] R. Akeela and Y. Elziq, “Design and verification of IEEE 802.11ah for IoT and M2M applications,” in 2017 IEEE International Conference on Pervasive Computing and Communications Workshops (PerCom Workshops), 2017, pp. 491–496.

[36] “IEEE 802.11ah (Wi-Fi in 900 MHz License-exempt Band) for IoT Application,” IEEE Standards University. [Online]. Available: https://www.standardsuniversity.org/e-magazine/august-2016-volume-6/ieee-802-11ah-wi-fi-900-mhz-license-exempt-band-iot-application/. [Accessed: 28-May-2018].

[37] B. Olin, H. Nyberg, and M. Lundevall, “A novel approach to WCDMA radio network dimensioning,” in IEEE 60th Vehicular Technology Conference, 2004. VTC2004-Fall. 2004, 2004, vol. 5, pp. 3443-3447 Vol. 5.

[38] M. I. Hossain, L. Lin, and J. Markendahl, “A Comparative Study of IoT-Communication Systems Cost Structure: : Initial Findings of Radio Access Networks Cost,” in 2018 11th CMI International Conference: Prospects and Challenges Towards Developing a Digital Economy within the EU, 2018, pp. 49–55.

[39] “Cells, Sectors and Antenna Beamforming.” [Online]. Available: https://www.commscope.com/Blog/Cells--Sectors-and-Antenna-Beamforming/. [Accessed: 13-Jan-2019].

[40] Sami Tabbane, “IoT network planning.” [Online]. Available: https://www.itu.int/en/ITU-D/Regional-Presence/AsiaPacific/SiteAssets/Pages/Events/2016/Dec-2016-IoT/IoTtraining/IoT%20network%20planning%20ST%2015122016.pdf. [Accessed: 26-Jul-2018].

[41] SalahElAyoubi,etal, “Deliverable d1.2 METISII Quantitative techno-economic feasibility assessment.” 2017.

35

[42] K. Johansson, J. Zander, and A. Furuskar, “Modelling the Cost of Heterogeneous Wireless Access Networks,” Int J Mob Netw Innov, vol. 2, no. 1, pp. 58–66, May 2007.

[43] “What is a capital expenditure (CAPEX)? definition and meaning,” BusinessDictionary.com. [Online]. Available: http://www.businessdictionary.com/definition/capital-expenditure-CAPEX.html. [Accessed: 30-Oct-2018].

[44] “Operating expense - Wikipedia.” [Online]. Available: https://en.wikipedia.org/wiki/Operating_expense#cite_note-1. [Accessed: 30-Oct-2018].

TRITA -EECS-EX-2019:180

www.kth.se

cost structure of iot connectivity services1326313/fulltext01.pdf · cases. the cost driver is...

Documents