How to Cut the Electric Bill in Mobile Access Networks: A Mobile Operator's Perspective

How to Cut the Electric Bill in Mobile Access Networks: A Mobile Operator’s Perspective

KONSTANTINOS CHATZIMICHAIL

Master’s Degree Project Stockholm, Sweden September 2014

TRITA-ICT-EX-2014:167

i

How to Cut the Electric Bill in Mobile Access Networks:

A Mobile Operator’s Perspective

KONSTANTINOS CHATZIMICHAIL chat@kth.se

Master of Science Thesis in Communication Systems Stockholm, Sweden, 2014

Academic advisor : Dr. Cicek Cavdar, Royal Institute of Technology cavdar@kth.se

Academic Examiner: Prof. Jens Zander, Royal Institute of Technology

jenz@kth.se

Abstract

One of the major challenges that mobile operators are facing is the in- creasing power consumption costs as a consequence of the network densifica- tion experienced in current and future mobile access networks. This power increase causes both financial and environmental concerns to operators, since both the operational expenses and the CO ₂ emissions are affected.

This Master Thesis investigates and analyses various deployment archi-

tectures in urban and suburban areas, considering both the radio access and

backhaul segments, as well as system solutions that are expected to increase

the energy efficiency of Long Term Evolution (LTE) networks, such as Dis-

continuous Transmission - Discontinuous Reception (DTX-DRX) enhance-

ment of future mobile networks. During this dissertation, it is investigated

whether power efficient solutions can also be cost efficient in terms of Total

Cost of Ownership (TCO) in urban environments. Additionally, the DTX-

DRX enhancement of HetNets is studied on a dense urban case, providing

both financial and environmental benefits by its utilization. The main re-

sults indicate that significant power and TCO savings can be achieved by the

deployment of Heterogeneous Networks in urban environments. Fiber optics

backhauling seems to be more attractive compared to Microwaves in these

areas. Finally, DTX-DRX enhanced HetNets result in high power savings

and can be financially attractive under particular conditions.

Abstrakt

En av de fr¨ amsta utmaningarna som mobiloperat¨ orerna st˚ ar inf¨ or idag

¨

ar elkonsumtions¨ okningen p˚ a bekostnad av n¨ atverksf¨ ort¨ atning i dagens och framtidens mobila n¨ atverk. Denna ¨ okning f¨ ororsakar b˚ ade finansiella och milj¨ om¨ assiga bekymmer f¨ or operat¨ orerna eftersom den p˚ averkar b˚ ade de op- erativa kostnaderna och CO ₂ utsl¨ appen. Denna avhandling presenterar och analyserar diverse utplaceringsarkitekturer i stads- och f¨ orortsomr˚ aden d¨ ar h¨ ansyn tas till b˚ ade radio access- och backhaulsegmenten, samtidigt som arbetet presenterar systeml¨ osningar som f¨ orv¨ antas kunna ¨ oka energieffek- tiviteten i LTE n¨ atverk, som till exempel Discontinuous Transmission - Dis- continuous Reception (DTX-DRX) f¨ orb¨ attringar av framtidens n¨ atverk. Det unders¨ oks ¨ aven om energieffektiva l¨ osningar ¨ ar kostnadseffektiva i termer av TCO i stadsomr˚ aden. En unders¨ okning av DTX-DRX f¨ orb¨ attringar av Het- Nets g¨ ors i t¨ atstadsomr˚ aden ur ett finansiellt och milj¨ om¨ assigt perspektiv.

Resultaten visar att signifikanta besparingar kan g¨ oras i termer av elkonsum-

tion och TCO n¨ ar HetNets placeras ut i stadsomr˚ aden och n¨ ar fiber anv¨ ands

som backhaulteknik. Fiber optik backhauling visar sig vara mer attraktivt

j¨ amf¨ ort med mikrov˚ agor i dessa omr˚ aden. Slutligen s˚ a resulterar DTX-DRX

f¨ orb¨ attrade HetNets i stora energibesparingar och kan d¨ armed bli finansiellt

l¨ onsamma om ett antal villkor uppfylls.

CONTENTS v

Contents

1 Introduction 1

1.1 Background . . . . 1

1.2 Literature Review . . . . 3

1.2.1 Energy Efficient Network Configuration: Macro-sleep . 3 1.2.2 Energy Efficient Base Stations . . . . 5

1.2.3 Energy Efficient Network Deployment . . . . 8

1.3 Problem Formulation and Research Questions . . . . 9

1.4 Methodology . . . . 10

2 Mobile Network Architectures 14 2.1 General Description . . . . 14

2.2 Radio Access Network . . . . 14

2.2.1 Macro Base Stations . . . . 14

2.2.2 Micro Base Stations . . . . 15

2.2.3 Pico Base Stations . . . . 16

2.2.4 Femto Base Stations . . . . 17

2.3 Backhaul Technologies . . . . 18

2.3.1 Copper . . . . 19

2.3.2 Microwave . . . . 19

2.3.3 Fiber . . . . 21

2.4 Core Network . . . . 23

3 Network Deployment and DTX-DRX enhancement 25 3.1 Deployment Model . . . . 26

3.1.1 Forecast Level . . . . 26

3.1.2 Radio Access Network Level . . . . 27

3.1.3 Backhaul Network Level . . . . 29

3.2 Deployment Architectures . . . . 29

3.3 Power Model . . . . 30

3.4 Cost Model . . . . 32

3.5 Discontinuous Transmission - Discontinuous Reception (DTX- DRX) . . . . 35

3.6 DTX-DRX in LTE . . . . 36

3.7 DTX-DRX Power and Cost Analysis . . . . 37

3.7.1 Data Traffic Model . . . . 38

3.7.2 Power Model . . . . 38

CONTENTS vi

4 Simulation Results 41

4.1 Network Deployment . . . . 41

4.1.1 Net Present Value Analysis . . . . 47

4.1.2 Conclusions . . . . 48

4.2 DTX-DRX enhanced HetNets . . . . 50

4.2.1 DTX-DRX Power Analysis . . . . 50

4.2.2 DTX-DRX Cost Analysis . . . . 54

4.2.3 DTX-DRX Environmental Impact . . . . 57

4.2.4 Conclusions . . . . 60

5 Conclusion 62

6 Future Work 64

A Appendix: MATLAB tool description 65

B Appendix: Simulation Results in Urban and Suburban Ar-

eas 66

LIST OF TABLES vii

List of Tables

1 RBS Power Parameters . . . . 31

2 Backhaul Power Paramenters . . . . 32

3 Cost parameters . . . . 34

4 Scenario Assumptions . . . . 41

5 Energy and Cost Efficient Deployment Solutions . . . . 49

LIST OF FIGURES viii

List of Figures

1 Traffic-revenue gap . . . . 2

2 Macro-Sleep . . . . 4

3 Radio Station Block Diagram[19] . . . . 6

4 Scenarios . . . . 11

5 LTE Macro Base Station [30] . . . . 15

6 Micro Base Station [30] . . . . 16

7 Pico Base Station [30] . . . . 17

8 LTE femtocell [31] . . . . 18

9 Microwave Point-to-Point (PTP) . . . . 20

10 Microwave Point-to-Point (PTP) . . . . 21

11 FTTx Deployment . . . . 22

12 Deployment Model . . . . 26

13 Deployment Architectures . . . . 30

14 DTX-DRX main concept . . . . 37

15 Data traffic profile in Europe [19] . . . . 38

16 Cell Power Consumption Model[42] . . . . 39

17 Comparison of power consumption cost between heterogeneous and homogeneous scenarios using MW and fiber in the dense urban area . . . . 42

18 Comparison of Total Cost of Ownership between heteroge- neous and homogeneous scenarios using MW and fiber in the dense urban area . . . . 43

19 Comparison of annualised costs using MW and fiber as back- haul in the dense urban area . . . . 44

20 Total Cost of Ownership Breakdown in the dense urban area . 45 21 Electric Bill as part of the OpEx in the dense urban area in 2014 and 2024 when homogeneous deployment and microwave backhauling is utilized . . . . 46

22 Electric Bill as part of the OpEx in the dense urban area in 2014 and 2024 when HetNet deployment and fiber backhauling is utilized . . . . 46

23 Power Consumption Cost and TCO (NPV for 10 years) in the dense urban area . . . . 47

24 Power Consumption Cost and TCO (NPV for 10 years) in the

urban area . . . . 47

LIST OF FIGURES ix

25 Power Consumption Cost and TCO (NPV for 10 years) in the suburban area . . . . 48 26 Daily power savings in a dense urban area for different de-

ployment strategies and when the Cell Dtx feature is enabled under different parts of the network . . . . 51 27 Power cost savings on a HetNet in a dense urban area when

the Cell Dtx feature is enabled under different parts of the network . . . . 52 28 Normalized power consumption sensitivity analysis for differ-

ent traffic loads in a dense urban area with DTX-DRX en- hanced HetNet deployment . . . . 54 29 Cost analysis on one DTX equipped macro base station-CapEx

increase versus Electric bill savings for several unit electricity cost values . . . . 56 30 Cost analysis on one DTX equipped femto base station-CapEx

increase versus Electric bill savings for several unit electricity cost values . . . . 57 31 Annual CO 2 savings in one DTX equipped macro base station,

IN-India, GR-Greece, CN-China, US-USA, DE-Germany, SE- Sweden . . . . 58 32 Annual CO 2 savings in one DTX equipped femto base station,

IN-India, GR-Greece, CN-China, US-USA, DE-Germany, SE- Sweden . . . . 59 33 Annual CO 2 savings in DTX equipped HetNet in a dense

urban area, IN-India, GR-Greece, CN-China, US-USA, DE- Germany, SE-Sweden . . . . 60 34 Comparison of power consumption cost between heterogeneous

and homogeneous scenarios using MW and fiber in the urban area . . . . 66 35 Comparison of Total Cost of Ownership between heteroge-

neous and homogeneous scenarios using MW and fiber in the urban area . . . . 67 36 TCO Breakdown the urban area . . . . 67 37 Electric Bill as part of the TCO in the urban area in 2014 . . 68 38 Electric Bill as part of the TCO in the urban area in 2024 . . 68 39 Comparison of power consumption cost between heterogeneous

and homogeneous scenarios using MW and fiber in the subur-

ban area . . . . 69

LIST OF FIGURES x

40 Comparison of Total Cost of Ownership between heteroge-

neous and homogeneous scenarios using MW and fiber in the

suburban area . . . . 69

41 TCO Breakdown in the suburban area . . . . 70

42 Electric Bill as part of the TCO in the suburban area in 2014 70

43 Electric Bill as part of the TCO in the suburban area in 2024 71

LIST OF FIGURES xi

ACKNOWLEDGEMENTS

I would like to honestly express my sincere gratitude to my supervi- sor, Dr.Cicek Cavdar, for the tremendous opportunity that she gave me to work on this field, be part of the 5GrEEn project-Towards Green 5G Mobile Networks-, as well as present part of this work at the European Wireless (EW) Conference 2014 in Barcelona. I am grateful for her continuous sup- port, guidance and patience in advising my study. I am particularly thankful to Professor Jens Zander for his fruitful comments and immense knowledge that have been of great importance in this work.

During my Master Thesis, I had the chance to cooperate with excellent professionals and kind people in the 5GrEEn project-Towards Green 5G Mo- bile Networks-. Our discussions and their vision on mobile networks have been a great inspiration and guidance for this work. I would like to offer my special thanks to Ashraf Awadelkarim Ahmed for his valuable comments and support.

Finally, I would like to give the biggest thanks to my parents, Pantelis and Vasiliki, and my sister, Annika, for their unconditional love, never-ending support and patience. Having always been by my side, supporting all my decisions, has been their greatest offer to me.

Konstantinos Chatzimichail

Stockholm, Sweden, 2014

1 INTRODUCTION 1

1 Introduction

1.1 Background

During the recent years tremendous data traffic increase is experienced. Ac- cording to Cisco forecast [1] the overall mobile data traffic is expected to grow to 15.9 exabytes per month by 2018, which is nearly an 11-fold increase over 2013. Meeting this exponential traffic growth in an affordable way is the challenge that mobile operators experience nowadays.

Offering these growing traffic services to the end-users can be achieved by different ways, such as acquisition of additional spectrum, increase of spec- tral efficiency or network densification. Since spectrum is a limited and very expensive source, while the spectral efficiency is technology and distance de- pendent, the main solution that has been utilized by the operators is network densification.

An increasing variety of mobile applications, growing number of con- nected terminals as well as new terms such as Machine-to-Machine Commu- nication (M2M) and Internet of Things (IoT) are expected to put even higher pressure on this capacity providing solution so that the operators can meet the extended need of coverage and high data rates.

On the other side, the exponential growth in data traffic and the follow-

ing increase in the number of base stations is not followed by the same pace

of revenue growth threatening the financial viability of the operators, since

this is constrained and dependent on new services and social factors. This

traffic-revenue gap, as presented at Figure 1, is a key challenge for future

mobile networks, generating numerous research questions for telecommuni-

cation engineers. The research focuses on minimising the network cost so

that cost-revenue gap makes the industry profitable for operators.

1 INTRODUCTION 2

Figure 1: Traffic-revenue gap

In the telecommunications field, there has been given high emphasis on utilizing the existing radio spectrum as efficient as possible, since it is a scarce natural resource with limited availability. Lower emphasis has been given though on another fundamental enabler, energy, that is expected to play crucial role in future mobile networks [2]. Estimations provide a clear idea regarding telecommunications contribution in the environmental issues. 3%

of the worlds annual electrical energy consumption and 2% of CO ₂ emissions are caused by the Information and Communication Technology (ICT) infras- tructure, while one tenth of that is caused by cellular mobile communication systems. It is perceived that cellular network infrastructure incurs impor- tant electrical energy consumption nowadays. This phenomenon is tempted to grow since estimations show that ICT energy consumption is rising at 15-20% per year, doubling every five years [3].

From an operator perspective, reducing the power consumption imple-

ments a double target, since both the Operational Expenditure (OpEx) is

substantially reduced, as well as contribution against the increasing aware-

ness of the harmful effects caused by CO ₂ emissions is provided, improving

the image of the telecommunication companies to customers and public or-

ganizations. The following chapter summarizes different areas of interest

regarding energy efficient mobile networks, considering both network archi-

tectures and system solutions.

1 INTRODUCTION 3

1.2 Literature Review

1.2.1 Energy Efficient Network Configuration: Macro-sleep The network densification needed in order to cover the growing traffic demand results in a rapid increase on the energy consumption of the mobile networks.

One of the main ideas that research has dealt with in order to minimize the energy needed to operate mobile cellular networks is macro-sleep, as the technique during which low traffic base stations are switched-off and the traffic is served by neighbouring nodes, is known.

Cellular networks are designed in such a way that each base station (BS) provides coverage and capacity to a specific area, consisting of one or more subareas, called ”cells” depending on the transmitters available per base station. Usually, three transmitters are utilized per base station, resulting in a three-sector design. Neighbouring cells are then assigned different radio frequencies so that interference can be eliminated. Mobile network design assumes that the cells are hexagonal, since uniform area polygons can be used to tessellate the plane without overlap or missed areas and the hexagon is usually chosen as the cell shape because it maximizes area coverage.

Macro sleep is taking advantage of this network design in order to save

energy by switching off cells during periods of light traffic. The traffic that

was served by the switched-off cell is then provided by neighbouring cells

according to the algorithms that the designer has implemented. The base

stations that are switched off result in lower energy consumption, but at the

same time network configuration is needed so that the end user experience

is not negatively affected by long delays. The evaluation of the trade-offs

between energy consumption and customer delay has been given high focus

on the research [10].

1 INTRODUCTION 4

Figure 2: Macro-Sleep

Since traffic load in mobile networks significantly varies during a work- ing or weekend day, the influence of these variations on the base station power consumption is high. Real traffic measurements on GSM and UMTS base stations have been conducted in order to show the difference on the instantaneous power consumption according to their respective traffic load.

A direct relationship between base station traffic load and power consump- tion has been observed, proving the necessity for smart design of switch on and off mechanisms, that put in an idle or sleep mode the base stations that experience low traffic profile for a certain time period [18].

Plenty of studies have been dedicated on the optimization of macro sleep, following different approaches. Switch-off mechanisms have been modelled as a binary integer linear programming (BILP) problem when interference is considered to be constant. Allowing the interference to be a function of the user assignment, which depicts a more realistic approach of the problem, necessitates a heuristic method. Genetic algorithms have been created to deal with it, resulting in linear complexity models [4][18].

Majorization-minimization (max-min) algorithms have also been devel- oped in order to find the smallest set of active base stations that can pre- serve the quality of service (minimum data rate) required by the users [5].

Moreover, distributed cooperative algorithms have been designed so that

neighbouring cells are switched-off when traffic variation is observed result-

1 INTRODUCTION 5

ing in both higher energy efficiency and better outage throughput[6]. One key aspect of switching on-and-off a base station is the additive traffic that has to be served by the rest of the network.

Heuristic algorithms have been proposed that turn off a base station tak- ing into account the additional load increments brought to its neighbouring base stations. According to these studies, significant reduction of 50-80% can be achieved in the total energy consumption on a real traffic profile from a metropolitan urban area [7]. Moreover, significant research has been made on minimizing the total network energy consumption considering the distance between the User Equipments (UEs) and the associated base stations [9].

1.2.2 Energy Efficient Base Stations

Energy efficiency is a significant requirement for the design and management of mobile networks and has recently gained substantial attention from both network operators and the research community. Emphasis has to be given on various components of the telecommunication systems so that their energy efficient design does not prevent the end user experience. To this direction, modules such as radio network equipment, backhaul network equipment as well as the user terminals have to be designed in a way that minimizes the power consumption while providing the desirable services.

Nowadays a variety of terminals are part of our everyday life. Smart- phones, tablets and laptops are important components of both professional and personal aspects of our lives. The user equipment, which is usually pow- ered by batteries, has therefore to be as energy efficient as possible, so that the battery life duration is as high as possible. The growing demand on mobile traffic and the following exponential growth of battery consumption (150% every two years) is not equally followed by the slow development of battery technology (10% every two years). As a result, an exponentially in- creasing gap between the energy demand and supply has to be faced by the terminal designers [2].

It has been shown that the biggest part of the energy consumption in a mobile network, around 80 pecent of the total power consumption, is con- sumed by the radio access network, namely the base stations [11]. Inter- national projects, such as METIS[38] and EARTH [19] have provided the essential framework so that power models can be adopted in future research.

EARTH results are mainly concentrated on LTE base stations, following the

trend in the mobile networks but the framework provided can be easily gen-

1 INTRODUCTION 6

eralized for other radio technology equipment as well.

A simplified block diagram of a complete base station that can be gener- alized to all base station types, including macro, micro, pico and femto base stations is provided on figure 3.

Figure 3: Radio Station Block Diagram[19]

A base station (BS) is generally made of multiple transceivers (TRXs) with multiple antennas, since modern mobile networks use sectorization tech- niques in order to provide coverage and capacity at an area. One TRX com- prises an Antenna Interface (AI), a Power Amplifier (PA), a Radio Frequency (RF) small-signal transceiver section, a baseband (BB) interface including a receiver (uplink) and transmitter (downlink) section, a DC-DC power supply, an active cooling system and an AC-DC unit (Main Supply) for the connec- tion to the electrical grid. The various TRX parts are further analyzed so that their contribution in the total power consumption is realised.

• Antenna Interface (AI)

Power efficiency is modelled through a certain amount of losses on the var-

ious modules of the antenna interface. The feeder, antenna band-pass filters,

1 INTRODUCTION 7

duplexers and matching components are parts of this power consumption model. A feeder loss of about feed = 3 dB needs to be added for a macro base station, while the feeder loss for smaller base station types, such as micro, pico and femto base stations is typically negligible. A power efficient design technique that is usually followed demands the power amplifier to be located at the same physical location as the transmit antenna so that the feeder loss is mitigated.

• Power Amplifier (PA)

Non-linear effects and OFDM modulation with non-constant envelope signals force the power amplifiers to operate in the linear region, instead of operating close to the maximum output power (near saturation), where they are typically most efficient. This prevents Adjacent Channel Interference (ACI) due to non-linear distortions, avoiding performance degradation at the receiver. One the other side, this results in poor power efficiency.

• Small-Signal RF Transceiver (RF TRX)

The Small-Signal RF Transceiver (RF-TRX) comprises a receiver and a transmitter for uplink (UL) and downlink (DL) communication. Parameters with highest impact on the RF-TRX energy consumption PRF , are gener- ally the required bandwidth, the allowable Signal-to-Noise Ratio (SNR), the resolution of the analogue-to-digital conversion and the number of antenna elements for transmission and reception.

• Baseband Interface

The Baseband Interface performs a plethora of digital signal process-

ing processes, such as up/down conversion, filtering, FFT/IFFT for OFDM,

modulation/demodulation, digital pre-distortion (only in DL and for large

base stations), signal detection and channel coding/decoding. For large BSs

the Baseband Interface also includes the power consumed by the serial link

to the backbone network. This interface is usually made of silicon, which af-

fects significantly the power consumption PBB, since the increasing leakage

sets limits on the power reduction that can be achieved through technology

scaling. The power consumption on the Baseband Interface is also affected

by various other factors apart from the technology, naming the signal band-

width, number of antennas and the applied signal processing algorithms.

1 INTRODUCTION 8

• Main Supply, Cooling and DC-DC

Power supply losses at the main supply, the active cooling system and the DC-DC unit scale linearly with the power consumption of the other components. Active cooling is only applicable to macro base stations, while it is omitted in smaller BS types, where the temperature increase is not a major problem for the operation of the radio network components.

1.2.3 Energy Efficient Network Deployment

The deployment of low cost and low power base stations has been recognized in recent years as a promising cost-efficient and energy-efficient strategy. Het- erogeneous Network (HetNet) deployment architectures, where small cells of- fer the necessary capacity, while macro cells operate us an umbrella covering the area, are expected to be a major part of future mobile networks. Homoge- neous (macro only) and heterogeneous deployment architectures have been compared under various traffic and environment conditions, while various energy saving schemes have been tested on top of them [14][15][17].

The existence of a higher number of base stations that the densification demands leads in higher expectations from the backhaul network as well.

Base stations of different size and various locations (indoors/outdoors) have to be fed by backhaul networks that need to be designed in a cost and power efficient way as well.

Different scenarios have been studied, where power consumption reduc- tion is the main target of the designers. To this direction, combined wireless and optical access networks have been investigated in areas with business and residential customers in various traffic environments [12]. Moreover, mi- gration strategies, where GSM (Global System for Mobile communications - 2G) and UMTS (Universal Mobile Telecommunications System 3G) hard- ware is replaced by modern equipment have been presented and analyzed. It is shown that the total energy consumption can be reduced by around 60%

by 2020 compared to today, despite deploying LTE (Long Term Evolution -

4G), if modernization of the existing legacy equipment is made at the same

pace as LTE rollout [16]. Finally, base station deployment schemes have been

proposed based on the whole traffic variation process, where the optimal base

station density is investigated regarding power consumption [13].

1 INTRODUCTION 9

1.3 Problem Formulation and Research Questions

Extensive research has been made on energy efficiency in mobile networks.

Various aspects have been covered, considering macro-sleep algorithms that put a radio base station into sleep or idle mode when it is underutilized, micro-sleep mechanisms, where parts of the base stations are switched off for a certain time period that the base station does not need to be fully active, energy efficient hardware design, where both radio access network modules and end user terminals are taken into consideration, as well as deployment architecture scenarios have been considered so that power consumption is eliminated.

This Master Thesis will investigate the importance of energy savings in mobile networks when cost is also considered. It will be examined from an operators perspective if energy-efficient deployment design can also be cost efficient. Since energy-efficient equipment is usually more expensive but re- sults in lower power consumption costs, this analysis will try to investigate if such an investment can pay back the operators. Both radio access network deployment strategies and backhaul network architectures will be considered for various traffic and environmental scenarios in urban and suburban cases.

Moreover, energy-efficient system solutions, such as Discontinuous Transmis- sion - Discontinuous Reception (DTX-DRX), will be assumed to be adopted in future radio equipment. A cost analysis will be implemented providing an indication on the additional investment that an operator should provide for this type of equipment, regarding the potential power savings that can be offered. Finally, the carbon footprint savings by DTX-DRX utilization will be investigated provided an environmental aspect of the problem.

This dissertation will try to cover the gap on DTX-DRX enhanced Het- Nets, providing a study on power savings achieved by their utilization in dense urban cases. Additionally, a cost analysis will be implemented to provide an indication on the conditions that make this option financially attractive. Moreover, carbon footprint savings by their utilization will be studied in countries with different energy mix, providing an indication of their environmental influence.

Finally, differentiating the point of view presented on previous studies,

where main emphasis was given on the power savings achieved by various

deployment options, without considering the Total Cost of Ownership, a

unified power and TCO study will be implemented on various radio and

backhaul deployment combinations in urban and suburban areas.

1 INTRODUCTION 10

The research questions that this Master Thesis will try to investigate are:

• Can energy efficient radio access deployment strategies and backhaul technologies be also cost efficient in terms of Total Cost of Ownership in urban environments?

• Which is the additional acquisition cost of a DTX equipped RBS that is justified by electric bill savings in dense heterogeneous networks?

• Which are the carbon footprint savings by DTX-DRX utilization in dense urban environments?

1.4 Methodology

This Master Thesis is a study that will try to examine whether energy ef- ficient mobile network strategies can be cost-efficient as well, considering different usage cases and scenarios. In the beginning an extensive descrip- tion of the various parts consisting a mobile network will be provided. Main emphasis will be given on the radio access network, where different types of base stations will be compared in terms of technical characteristics, energy efficiency and deployment options.

Following, different backhaul technologies will be examined. Since het- erogeneous networks with a variety of numerous base stations are expected to play important role in future mobile networks, power consumption and cost of the backhauling that will provide the necessary capacity to these base stations becomes an important part of future networks as well. Different backhaul networks architectures will be examined with a strong emphasis on microwave and fiber technologies. The major characteristics, topologies and architectures will be presented and analyzed.

Next, various scenarios will be examined, where various traffic profiles, environments and deployment architectures will be considered. This study will consider dense urban, urban and suburban areas, where different radio ac- cess network architectures (homogeneous versus heterogeneous) and different backhaul architectures (microwave, fiber) will be considered. The scenarios will assume 1Km ² areas, where the number of users and number of buildings will differ. In the dense urban case, 3000 users will be considered to be uni- formly distributed in 6 5-floor buildings, corresponding to a business scenario.

In the urban case, 1000 users will be considered in 50 2-floor houses, whereas

in the suburban case 500 users are located in 20 single-floor apartments.

1 INTRODUCTION 11

Figure 4: Scenarios

A techno-ecomomic analysis will be implemented where power consump- tion cost and Total Cost of Ownership (TCO) as a sum of Capital Expen- diture (CapEx) and Operational Expenditure (OpEx) will be provided and compared. This analysis will try to investigate if power efficient radio access and backhaul deployment architectures turn to be also cost efficient when high traffic requirements are conisidered.

In the following chapter, emphasis will be given on future radio base sta-

tions. Since this part of the radio access network is the main power consum-

ing component, it is important to examine power efficient solutions. EARTH

project [19] has already provided various power models for future radio base

stations. Power saving techniques, such as DTX-DRX, that are expected to

be part of future radio base stations will be presented and analysed. Cost-

Power saving analysis will be implemented on these modules to investigate

the acquisition cost margin that is justified by the electric bill savings that

they provide. Moreover, carbon footprint savings due to DTX-DRX utiliza-

1 INTRODUCTION 12

tion will be analyzed for different RBS types.

Finally, conclusions will be drawn and future work as expansion of the presented one will be suggested.

This Master Thesis’ contributions are mainly summarized on

• the study, verification and extension of the work presented at [61]

• the implementation of an extendible MATLAB tool for the deployment of HetNets, their backhauling, as well as their DTX-DRX enhancement (Appendix)

• the study of DTX-DRX enhancement on HetNets

• the translation of DTX-DRX equipped base station electric bill savings

into carbon footprint terms, providing an environmental aspect of the

problem

2 Mobile Network Architectures

2.1 General Description

Different components are combined when forming a mobile network. The main parts of the network are: the radio access part, which mainly contains the radio base stations transmitting and receiving the electromagnetic sig- nals, the backhaul part, which is responsible to provide adequate capacity to the radio access part, and the core (backbone) part, which connects differ- ent parts of the access network and provides the gateway to other networks.

Each network component is further described in the following sections.

2.2 Radio Access Network

Two different radio access deployment architectures will be considered in this study. Homogeneous deployment is the network architecture that utilizes only one type of radio base stations. In practice, homogeneous networks are formed by macro base stations, following the traditional pattern of mobile network dimensioning.

When different sizes of base stations are utilized to build mobile networks, heterogeneous network deployment is implemented. This option is expected to form future mobile networks, where the proximity to the end-user will satisfy the quality of experience requirements.

2.2.1 Macro Base Stations

Radio Access Characteristics A macro base station is a base station

that provides the largest area of coverage within a mobile network. Its cov-

erage area depends on the frequency used and the physical terrain but is

usually an area with a radius of maximum 35 kilometres, creating a macro-

cell of that radius. It is deployed outdoors and the antennas can be mounted

on ground-based masts, rooftops or other existing structures. These struc-

tures are usually positioned at a height that is not obstructed by terrain or

buildings, since the main advantage of the macro base stations is the cover-

age that they can provide in an extended area, providing service to a high

number of users.

2 MOBILE NETWORK ARCHITECTURES 15

Power Characteristics, Cost and Backhauling Options Usually macro base stations are an expensive option, since their high price, site acquisition, rental costs and their energy consumption levels are higher than any other type of base stations. CapEx and OpEx are therefore higher than other ra- dio access network elements. On the other side, a small number of them are needed to cover an extended area, because of their large radiation ra- dius, when coverage is the main design concern. The radiated power varies between 5 and 40 Watts depending on the network planning, the coverage purposes and the interference management [20].

Today macro base stations can provide 2G, 3G and 4G services. Legacy 2G sites are usually backhauled using copper, since 2G networks were mainly deployed during 1980s, when copper was the leading backhauling solution.

Being in the initial phase of mobile networks with low traffic requirements and low functional complexities, copper was an adequate option providing the necessary capacity to macro base stations. On the other side, 3G and newly deployed 4G base station sites typically use microwave or fiber when deployed, since the high expenses of copper deployment and the limited ca- pacity that can be offered make this backhauling solution inadequate to meet today needs.

Figure 5: LTE Macro Base Station [30]

2.2.2 Micro Base Stations

Radio Access Characteristics Micro base stations are also usually de-

ployed outdoors. They are radio access equipment with a coverage radio of

maximum 2 Kilometers and it is usually placed below-rooftop levels, such as

external walls, lamp posts, etc. covering a specific area of interest, such as a

street, a block or a square with high traffic demand.

2 MOBILE NETWORK ARCHITECTURES 16

Power Characteristics, Cost and Backhauling Options Micro base stations are mainly used for covering a specific area, while their operation is usually additive to umbrella macro base stations, which cover a more extended area. To this direction, radiated power levels of the micro base stations are much lower than the ones by macro base stations, since lower interference problems should be caused to the end users. Typical radiated power varies between 0.5 and 2 Watt for a micro base station.

The size and the cost of this type of antennas are also lower compared to macro base stations, since both their price and their operational expenditure are limited. On the other side, their limited coverage would make them an expensive option if coverage is the main design concern. This type of base stations should be mainly used for capacity issues additively to a macro base station, when the last one does not meet the traffic requirements in this area.

Micro base stations are usually backhauled via microwave nowadays.

Figure 6: Micro Base Station [30]

2.2.3 Pico Base Stations

Radio Access Characteristics Pico base stations provide even lower cov- erage than micro base stations and therefore lower radiated power is needed.

Typical coverage radius is lower than 200 meters, making this type of base

stations flexible for both indoor and outdoor usage, depending on the needs

of the network designer. They are often deployed to remedy the coverage

or capacity holes in a given area, like a hot-spot, since they can provide the

coverage/capacity needed without causing any interference problems, when

they are placed at a reasonable distance from a macro base station.

2 MOBILE NETWORK ARCHITECTURES 17

Power Characteristics, Cost and Backhauling Options As expected, the dimensions, cost and radiation power are lower than the ones of a micro base station. Typical values of radiated power are 0.25-2 Watt for indoor use and 0.1 Watt for outdoor use.

The backhauling provision depends on the area that the pico base station is planned to cover. When it is placed indoors, it is usually backhauled through an existing broadband infrastructure, under the condition that it can provide enough capacity (i.e. Fiber To The Home/Curb combined with Ethernet). When outdoor usage is planned, the pico base stations are usually backhauled via microwave, operating in a similar way as the micro base stations.

Figure 7: Pico Base Station [30]

2.2.4 Femto Base Stations

Radio Access Characteristics Femto base stations are different than the rest base station types, since they are meant for the consumer market, in contrast to prior described equipment that is mainly used by network operators in an industry oriented way.

As a large number of small cells or femtocells would be deployed in order

to provide the same coverage as one macro cell, the economic perspective

of macro cells differs from that of small cells. Indoor femtocell devices are

less expensive, lower power and more portable equipment than that of the

macro cells. In order to provide the same coverage as the macro layers, a

large number of femto cells would be required, but on the other side they

would play a crucial role in solving the capacity issues, which arise due to

high data rate requirements [21] in enterprises and public hotspots. With the

deployment of small cells, the end user would experience higher data rates

2 MOBILE NETWORK ARCHITECTURES 18

as one would operate close to the radio equipment (0 to 15 metres) unlike that of macro cells, which are equipped for solving the coverage issues but not the future higher data rate requirements.

Figure 8: LTE femtocell [31]

Power Characteristics, Cost and Backhauling Options A totally dif- ferent business model should be followed when femtocells come into picture, since they are customer oriented, plug-and-play devices that provide indoor coverage and capacity services to residential and business areas. Their di- mensions and cost is lower than larger base stations with advanced network functionalities. Radiated power of less than 0.1 Watt is typically assumed, since their proximity to the end user and their short coverage radius can be satisfied with low power radiation, making these devices extremely interest- ing from an energy perspective as well.

Femto base stations, being placed indoors in the customers premises, are usually backhauled via the users existing broadband infrastructure, such as a digital subscriber line (DSL), cable modem, Ethernet or fiber.

2.3 Backhaul Technologies

In traditional mobile networks the backhaul contribution to the total power

consumption is usually neglected because of its limited impact compared to

that of the radio access network equipment. However, the almost exponential

increase in mobile data traffic demand is expected to require a large number

of base stations, since small base stations are expected to be an important

part of future mobile networks. It has been shown that the total power

2 MOBILE NETWORK ARCHITECTURES 19

consumption of a heterogeneous network deployment is affected not only by the presence of the backhaul [22] but also by the specific technological and architectural choices [23].

In this chapter, some of the most popular backhaul options will be de- scribed, namely copper, microwave and fiber.

2.3.1 Copper

As mentioned before, copper is the legacy technology that has been exten- sively used during the past decades to provide fixed broadband connectivity (i.e. ADSL, VDSL), as well as backhaul traffic in the early generations of mobile services. Even though copper was a sufficient choice regarding capac- ity provision to early mobile systems, its inability to provide high capacities over long distances has made its functionality to decrease over the years and it is mainly used as an already deployed infrastructure in sparse areas that capacity is not a big issue but the main concern is put on coverage aspects. As examples of backhaul provisioning data rates can be stated the VDSL2 bonding, which can provide up to 100Mbit/s downstream and up to 5.4Mbit/s upstream over 1500m [24], whereas even the latest improvement in DSL technology [25] may enable rates up to 500Mbit/s but only over very short distances (around 100m).

Even though copper does not seem promising for future mobile networks, it is still an important part of today networks, since it can provide back- hauling solutions to a wide range of scenarios, especially in the short term future that the traffic demand is not that high, while more capacity efficient backhauling technologies are deployed.

2.3.2 Microwave

Microwave backhauling is a usual choice in urban and rural regions [25] [26]

mostly because of its low deployment cost and its ability to provide rather high capacity services to the base stations. Even though it is a rather cheap backhauling option, the operators have to lease spectrum resources that are needed for the communication of the microwave links, resulting in higher Operational Expenditures (OpEx), depending on the amount of spectrum that is needed.

Two categories are identified when microwave is considered, namely Mi-

crowave Point to Point (PTP) and Microwave Point to Multiple Point (PMP)

2 MOBILE NETWORK ARCHITECTURES 20

links. These two categories will be further presented and analyzed.

Microwave Point-to-Point (PTP) Microwave Point to Point requires a dedicated link (in the 2-30 GHz range) to connect each Radio Access Network (RAN) site to a hub node that is in turn connected to the metro/aggregation segment. If the RAN site is too far from the hub, or if there is no Line of Sight (LOS) connectivity, the backhaul may include multiple hops [27].

A special subcategory within this category is the E-Band PTP links, which is a backhaul solution that presents the same topology and architecture characteristics as the conventional PTP microwave links, while utilizing a different spectrum range, namely 70-80 GHz. Due to this spectrum range, E- band PTP links can reach a shorter distance and are affected more drastically by environmental factors, such as rain and hail.

Figure 9: Microwave Point-to-Point (PTP)

Microwave Point-to-MultiPoint (PMP) In contrary to the previous

analyzed option, Microwave Point to Multiple Point is the backhaul option

that allows one access point (AP) in the hub to connect simultaneously to

multiple RAN sites, leading therefore to savings in terms of radio equipment

needed.

2 MOBILE NETWORK ARCHITECTURES 21

Figure 10: Microwave Point-to-Point (PTP)

2.3.3 Fiber

Fiber is the most promising backhaul solution when future high traffic de- mand networks are considered, since it can provide the radio sites with virtu- ally unlimited capacity, which might be a special need in dense urban areas [28]. Even though fiber is a capacity and power saving promising technol- ogy, an adequate fiber infrastructure deployment level can take several years, while the fiber rollout and equipment acquisition cost might be a big draw- back for mobile operators. Fiber is expected though to be part of mixed backhaul technologies, since operators may still benefit from an already de- ployed last mile infrastructure (i.e. copper), especially for small radio base stations, whose aggregated bandwidth is not that high.

A large family of fiber deployment architectures, named Fiber To The-x (FTTx) is nowadays available. The selection of each option depends on how near to the subscriber the fiber reaches. Some typical FTTx architectures are further presented.

Fiber-To-The-Node (FTTN) Fiber-To-The-Node is the optical solution when the optical signal reaches the node of the network and it is terminated in a street cabinet in a large distance from the customer premises.

Fiber-To-The-Curb (FTTC) Fiber-To-The-Curb is the deployment so-

lution when the optical signal reaches the curb, which is usually located in

a distance of less than 300 meters from the building. Once again, another

type of backhauling has to be used in order to reach the end user.

2 MOBILE NETWORK ARCHITECTURES 22

Fiber-To-The-Building (FTTB) Fiber-To-The-Building is the option when the optical signal reaches the building where the end user is located, but does not go into it. This means that another type of backhauling (Ethernet, copper, etc.) has to be used to reach the end user.

Fiber-To-The-Home (FTTH) Fiber-To-The-Home is the optical fiber deployment, where the optical signal reaches the end subscribers equipment situated in the subscriber home.

Figure 11: FTTx Deployment

From a power consumption perspective, an attractive FTTx solution is a

Passive Optical Network (PON) that utilizes passive optical splitters in order

to connect a number of users, usually up to 32 or 64, to a single port of an ag-

gregation switch in the Central Office (CO). Some of the optical access tech-

nologies that have been implemented during the last years are Point-to-Point

(P-t-P) optical Ethernet and Point-to-Multipoint (P-t-MP) passive optical

networks such as ATM based PON (APON/BPON), 1 Gbit/s Ethernet-based

PON (GPON), 10 Gbit/s PON (10GPON) and Wavelength-Division Multi-

plexed PON (WDM-PON). Hybrid PONs that combine WDM and TDM

in a single network and Long-Reach PONs (LR-PONs) are also expected

to play an important role in future networks, providing more cost-efficient

and extended-reach operation. WDM technologies and optical amplification

techniques allow an extended network reach from the traditional 20 Km to

100 Km [29].

2 MOBILE NETWORK ARCHITECTURES 23

2.4 Core Network

Core network (or backbone network) is the part of the network that connects different parts of the access network. It also provides the gateway to other networks. It is mainly implemented by the utilization of routers and switches.

Aggregation, authentication, switching, charging and service invocation are

the main functions offered by this part of the network. The core network

will not be considered for the rest of this Master Thesis project, where the

emphasis will be given on the radio access and backhaul network.

3 Network Deployment and DTX-DRX en- hancement

The main layout of network deployment and the DTX-DRX enhancement of HetNets are described in this chapter. Initially, it is investigated whether energy efficient deployment architectures can also be cost efficient, in terms of Total Cost of Ownership (TCO), when 4G Long Term Evolution (LTE) rollout is implemented in a greenfield deployment scenario, namely when no previous network infrastructure is present and the operators have to build their network from scratch. Three different types of environment will be investigated: dense urban, urban and suburban areas. Both radio access and backhaul deployment dimensioning will be implemented and power con- sumption as well as cost analysis, in terms of TCO as the sum of CapEx and OpEx will be provided.

The study implemented in the beginning of this chapter is based on the work conducted on [61]. In that work, high emphasis was given on spectrum utilization in mobile networks as well as power and deployment costs for a time period of 10 years. This study, part of which is presented at [62], will fo- cus on the annualised investments needed for the network deployment. High emphasis will be given on two main characteristics: power cost in terms of electric bill and Total Cost of Ownership. It will be investigated whether these two costs can be simultaneously eliminated under one single RAN de- ployment and backhaul technology choice. Moreover it will be investigated how long it takes for a particular network deployment choice to become power efficient and cost efficient. The study will focus on urban and suburban areas of 1Km ² and the traffic demand during the time period 2014-2024 will be considered.

Following, DTX-DRX enhanced HetNets will be presented and analysed

in terms of power efficiency. Both electric bill savings and carbon footprint

reduction will be calculated on a dense urban area scenario. Finally, a cost

analysis will be implemented to show the potential financial benefit that

operators might experience by their utilization.

3 NETWORK DEPLOYMENT AND DTX-DRX ENHANCEMENT 26

3.1 Deployment Model

The deployment model is similar for all the environments under study. As depicted on figure 12, different levels of analysis can be identified namely the traffic demand level, radio access network level and backhaul level.

Figure 12: Deployment Model

3.1.1 Forecast Level

Various levels of uniformly distributed users in an area of one square Km are assumed. Following the EARTH standards, 3000 users are assumed in the case of the urban business area (dense urban), 1000 in the urban residential and 500 users in the suburban area case.

The traffic assumptions will follow the Swedish traffic demand expecta-

tions. According to the Swedish Post and Telecom Authority (PTS) there

was a traffic demand of 6.6 GByte per month per mobile subscription as a

standalone service in 2013 [44]. Considering that an increase of 70% per year

is expected, this traffic increase generates important research issues, even if

the traffic demand does not follow exactly the traffic forecasts. In our sce-

nario the traffic demand should be covered by the deployed infrastructure

for the period 2014 to 2024, namely a period of 10 years, while afterwards

3 NETWORK DEPLOYMENT AND DTX-DRX ENHANCEMENT 27

new technology is expected to be utilized for mobile access networks. As ex- pected [1] indoor traffic will dominate the total traffic demand (70-80%). For these purposes, the indoor/outdoor ratio will be assumed 3/1, namely 75%

indoor usage, whereas 25% outdoor usage, for the indoor deployment when indoor base stations will be deployed. 10 MHz spectrum will be assumed to be utilized by RBSs in the rest of the study.

Moreover, according to the studies conducted in the scope of EARTH project [19]; 10-30% of the data subscribers are active in the busy/peak hours in mobile networks. Respectively activity factors of 12.5% and 25%

are assumed for ordinary usage and heavy usage pattern respectively in this study; which can be translated to 8 and 4 hours per day.

The above mentioned input parameters: type of area, average data rate per user, user density, usage pattern as well as indoor/outdoor ratio, are used in the traffic forecast level, so that the user demand in the area will be provided as input to the next level of the model, the radio access network model.

3.1.2 Radio Access Network Level

The second level of the deployment model represents the radio access network dimensioning. In this part, the number of the required number of radio base stations, that are needed in order to meet the capacity and coverage purposes in each area, depends on the coverage and capacity constraints of the type of base stations used. Regarding the outdoor network dimensioning, the number of macro base stations needed depends on the capacity and coverage requirements, as well as the coverage and capacity characteristics of the radio stations, as shown at:

N _macro = max( N _user ∗ R _user

B _{T RX} ∗ N _{T RX} ∗ SE , A _service

A _RBS ) (1)

where A _service is the size of the service area, A _RBS is the coverage area of

specific RBS type , B _{T RX} is the available (spectrum) bandwidth per TRX,

N _{T RX} is the number of TRXs per RBS, SE is the spectral efficiency of the

adopted radio access technology, N _user is the number of uniformly distributed

users in the area and R _user is the average offered data rate per user. The

network dimensioning is made on a year-over-year basis, which means that

radio infrastructure is built every year, in order to follow the traffic increase.

3 NETWORK DEPLOYMENT AND DTX-DRX ENHANCEMENT 28

For the dimensioning of the indoor network, implemented by femtocells, one more restricting factor is taken into consideration, since one main limi- tation of the femtocells is the number of users that can serve simultaneously.

Adding this factor to capacity and coverage requirements, the following for- mula is utilized for the femtocell network dimensioning:

N _{f emto} = max( N _user ∗ R _user

B _{T RX} ∗ N _{T RX} ∗ SE , A _service

A _RBS , N _user

N _con ) (2)

where N _con stands for the number of users that a femtocell can support simultaneously.

As can be seen from the equations (1),(2) the number of base stations needed depends both on coverage and capacity requirements. As far as cov- erage is considered, the number of base stations needed, depends on the required data rate at the cell edge, thus the cell size can be calculated by considering the relationship between data rate and distance and the used frequency band (i.e. high or low frequency bands). This study is focused on the rollout of LTE. By using the same methodology as in [45][46], cell radius of 3km and 50m could be estimated for LTE macrocell and LTE-femtocell respectively.

Since the area that has to be covered is limited (1 square Km) and the environment corresponds to high traffic demand cases, capacity requirements are of major importance in the network deployment. Equation (1) provides the framework of the capacity dimensioning, where the used Radio Access Technology (RAT) with its corresponding spectral efficiency, the number of carriers (TRX) and the amount of allocated frequency bandwidth are used.

LTE macrocells are considered as 3-sector base stations with one TRX per sector (10 MHz), whereas the average spectral efficiency is considered as SE=1.6. When LTE femtocells come into the picture, these are modeled as 1-sector omni-directional base stations (only one TRX of 10 MHz), whereas the average spectral efficiency for indoor environments is regarded as SE=4, due to closer distances between the end user and the base station.

In this level, the number of base stations needed, as well as the peak

traffic that each base station should provide is extracted. According to these

requirements, the backhaul network dimensioning is implemented at level 3

of the model.

3 NETWORK DEPLOYMENT AND DTX-DRX ENHANCEMENT 29

3.1.3 Backhaul Network Level

The backhaul network comprises the intermediate link between the radio access network and the core network. It is composed of backhaul links and aggregation nodes that aggregate the traffic so that information is forwarded to the radio access part for the final transmission.

The number of required backhaul links can be calculated as function of the average backhaul load offered by each RBS ( R _BH ), maximum capacity of the backhaul link (C _BH ) and the number of radio bases station as illustrated in Equation (3):

N _BH = R BH

C _BH ∗ N _BS (3)

The number of required aggregation nodes is a function of the number of required backhaul links (N _BH ), maximum capacity of the backhaul link (C BH ) and maximum capacity of the aggregation node (C ag ) as illustrated in Equation (4):

N _Agg = C _BH

C _Agg ∗ N _BH (4)

As for the RAN deployment, the backhaul deployment also follows the year-over-year basis, since additional backhauling is built every year in order to meet the traffic demand of the increasing number of base stations. When both radio access and backhaul dimensioning is implemented, power and cost analysis will be made on top of them, so that power consumption and Total Cost of Ownership comparisons can be made and conclusions on power/cost efficient deployment architectures can be drawn.

3.2 Deployment Architectures

The deployment architectures studied at this Thesis include both RAN and

backhaul components. RAN dimensioning will be implemented comparing

homogeneous (macro only) versus heterogeneous (macrocells for outdoor cov-

erage and femtocells for indoor capacity services) architectures. Regard-

ing backhauling two different technologies will be taken into consideration,

namely p-t-p microwave and 10GPON Fiber-To-The-Home (FTTH) optical

fiber solutions.

3 NETWORK DEPLOYMENT AND DTX-DRX ENHANCEMENT 30

The deployment options are depicted on figure 13, where both RAN and backhaul components are included.

Figure 13: Deployment Architectures

3.3 Power Model

The total power consumption of the network consists of the power consump- tion at the radio access network part and the power consumption at the backhaul part. Each of them contains two parts: a fixed power part, which is totally independent from the traffic load, and a dynamic power consumption part, which depends on the traffic load of the system.

The power model developed within the scope of EARTH project is utilized in this study to estimate the power consumption of different types of RBSs [19]. Hence, the power consumption per Radio site (P _BS ) can be modelled as shown in the equation (5) below:

P _BS = N _{T RX} ∗ (P _o + ζ ∗ P _k ) (5)

3 NETWORK DEPLOYMENT AND DTX-DRX ENHANCEMENT 31

Parameters P _o [W ] P _k [W ] ζ Source Macrocell 118.7 20 5.32 [59]

Femtocell 4.8 0.05 7.5 [59]

Table 1: RBS Power Parameters

where P k denotes the power spectral density in a cell, ζ represents the por- tion of the transmit power due to feeder losses and power amplifier, whereas P _o is the traffic independent power consumption part. P _o , ζ and P _k have been estimated for different radio base station types, as presented on Table 1, while N _{T RX} represents the number of TRX per base station.

The power consumption in the backhauling network can be modeled as the power consumed per backhauling connection (which is either fiber optic cable or microwave link) and its associated aggregation nodes. The power consumption in the backhaul link can be modeled taking into account the fixed part of the power consumed independently from the traffic load and the dynamic part of the consumed power that depends on the traffic load as described by equation (6):

P BH = a ∗ P F + R _BH

C _BH ∗ P tx

H (6)

Where R _BH represents the average backhaul load offered by each radio base station, C _BH represents the capacity of the backhaul link α is the per- centage of the power consumed proportional to P _max in idle mode (indepen- dent from traffic load) and P _tx

_H depicts the maximum transmit power per backhaul link.

Finally, the power consumption in the aggregation node, P _Agg , can be calculated using the following model shown in equation (7) [22], and using the corresponding values in Table 2.

P _Agg = a ∗ P _max + N _BH ∗ R _BH

C _Agg ∗ (1 − a) ∗ P _max + P _DL ∗ N _DL + P _{U L} ∗ N _{U L} (7) Where C _Agg represents the capacity of the aggregation node/access mod- ule and P max is the power supplied to the aggregation node Moreover P DL

and P _u stand for the power consumed in a downlink port and uplink port

in the aggregation node respectively. N _{U L} , N _DL are numbers of the uplink

3 NETWORK DEPLOYMENT AND DTX-DRX ENHANCEMENT 32

Parameters a P _max [W ] P _{U L} [W ] P _DL [W ] C _Agg Source

Fiber Optic 0.6 95 20 4.5 32 Gbps [49][28]

LAN switch 0.6 57 2 1 3.6 Gbps [53][22]

Table 2: Backhaul Power Paramenters

interfaces and downlink interfaces in the aggregation node respectively. N _BH is the total number of backhaul links per aggregation node.

3.4 Cost Model

As soon as the network dimensioning is completed, cost analysis can be im- plemented considering both RAN and backhaul network components. Since the network deployment is annualized, the cost analysis is also implemented in an annualized way, which means that additional cost is calculated every year, considering that network densification is needed to meet the increasing traffic demand.

TCO analysis is implemented where both CapEx and OpEx are taken into consideration. CapEx includes the build-out cost, in terms of civil works, shelter, auxiliary systems and other related installation and commissioning costs, as well as the telecom equipment, namely the base stations, backhaul links and aggregation nodes. OpEx is composed of the cost of annual opera- tion and maintenance activities cost together with the cost of annual power consumption and leased lines or spectrum fees. As shown in Table 3, the dominant component in macrocell site build-out is the cost associated with the civil construction work and the cost of auxiliary systems; this includes costs for towers or masts, non-telecom equipment, power system, installation, and site lease. In the femtocell deployment case, the dominant component is the cost of the femtocell equipment.

CapEx is a function of the total number of modules Ni, the cost of each

module Ci and the equipment erosion factor q, per year, as presented at

Eq.8. The latter represents the price erosion factor of the equipment, as a

result of functionality and the commercial value of equipment through the

years. In this study equipment erosion factor=5% is assumed[47]. When

fiber is considered as backhauling an additional cost is added representing

the fiber rollout cost. For that purpose, the cost of the fiber rollout N j is

considered for the architecture (Central Office - splitter - ONU), where the

3 NETWORK DEPLOYMENT AND DTX-DRX ENHANCEMENT 33

distance between two components is regarded as d=0.25 Km.

CapEx = sum ^N _i=1 (N i ∗ C i ∗ q) (8) OpEx is a function of the annual Operation and Maintenance activities, the power consumption cost, the radio sites rent cost and the leased lines and/or spectrum fee per backhaul link. In this study the annual Opearation and Maintenace cost (OM) is assumed to be 10% of the CapEx of the year.

OpEx = C _OM + C _power + C _rent + C _bhl

_ee (9) The annual TCO is then composed of the annual CapEx and the annual OpEx, as presented at Eq.9

T CO = CapEx + OpEx (10)

Apart from the annualized analysis of the deployment cost, the deploy- ment options should be compared for the total period of 10 years (2014-2024), that this investment is planned to cover. In order to compare these deploy- ment instances, the Net Present Value (NPV) method is used considering a discount rate of r=10%. NPV is a central tool in discounted cash flow (DCF) analysis and is a standard method for using the time value of money to ap- praise long-term projects. NPV can be described as the difference amount between the sums of discounted: cash inflows and cash outflows. It compares the present value of money today to the present value of money in the future, taking inflation and returns into account.

N P V = sum ^T _t=1 T CO _t

(1 + r) ^t (11)

Where T CO _t represents the annualized total cost of ownership in terms

of the annualized CapEx and OpEx at year t, while r represents the discount

rate and T is the network operation period.

3 NETWORK DEPLOYMENT AND DTX-DRX ENHANCEMENT 34

Parameters Cost in kEuro Source

Femtocell base station and Cabling (1TRX) 1 [48]

Macrocell LTE base station (3TRX),10MHz 10 [49]

Macrocell LTE additional TRX 5 [49]

Macrocell Site Build Out 75 [48][50]

Macrocell Site Lease 8 [50]

Microwave Link (Antenna, ODU&IDU) 4 [46][28]

Microwavve Hub Site, 6*IDU, 24 Ports 8 [46][28]

10 Gbit Ethernet Switch, 24 ports 1.8 [51][52]

Microwave Link in the CO, 96 ports 14 [46][28]

Annual Spectrum fee per MHz and link 0.06 [53]

Optical Network Unit (ONU) 0.25 [46]

OLT Access Module ,4*10GPON ports 6.5 [54]

Fiber Optic Passive Splitter 0.1 [22]

OLT Shelf (Management Module Switching Module &

Rack)

35 [54]

Fiber Rollout per Km 3.8 [49]

Annual Fiber Optic Lease Line (around 80 Euro per 100 Mbps per month)

3 [50]

Electricity price per kWh in 2014 0.00007764 [58]

Table 3: Cost parameters