Location via proxy:   [ UP ]  
[Report a bug]   [Manage cookies]                

Marran Computer Science 2017 PDF

Download as pdf or txt
Download as pdf or txt
You are on page 1of 75

UNIVERSITY OF TARTU

Institute of Computer Science


Computer Science Curriculum

Siim-Toomas Marran

Simulating Cellular Network Behaviour


in the Context of Urban Environment

Master’s Thesis (30 ECTS)

Supervisor(s): Amnir Hadachi, PhD


Artjom Lind, MSc

Tartu 2017
Simulating Road Traffic for Generating Cellular Network Logs in
Urban Context

Abstract:
In the last years, the use of mobile phone data logs start to attract a lot of researchers’
attentions from various disciplines. Those logs help the scientist to understand and pre-
dict human behaviour. The mobility logs, like Call Detail Records and GPS data, show
where to people commute, how often do they commute and, usually, those logs also
say why. These logs hold knowledge about our society, from that data the knowledge
could be extracted and used for multiple purposes. The scientists could analyse through
the movement how to plan the road infrastructure, generate target advertisement based
on forecasting peoples displacement, new positioning technology, population control
software, etc.
But there are limits on the people’s mobility data. Those information logs are heav-
ily protected by the government privacy data laws to protect the personal rights. Ad-
ditionally, the mobile operators are interested in their own commercial solutions and
therefore their interest to share vital information is low. Here, in this thesis, we show
that this cumbersome problem can be over-stepped by prototyping a cellular network
behaviour simulator to generate the logs for us through different scientific commuting
models inherited from the traffic simulation program.
The result of this thesis reveals that this approach is feasible and shows multiple
expansion possibilities how to produce even more real-life like mobility logs. The de-
velopment of the cellular network behaviour simulation has shown huge potential and
even bigger possibilities than predicted in the beginning. Since, our cellular network be-
haviour simulation is integrated with already existing open-source, highly configurable,
road traffic simulator basing on the scientific human behaviour models produce with
considerable value data.
Keywords: The traffic simulation; Cellular network behaviour simulation; The mobility
logs
CERCS: P170

2
Liikluse simuleerimine mobiilside võrkude logide genereerimiseks
linnakeskkonnas
Lühikokkuvõte:
Viimastel aastatel on hakanud mobiilside andmestik paeluma aina rohkem teadlasi
erinevatelt teadusdistsipliinidelt. Need andmed aitavad mõista inimeste käitumis- kui
ka liikumismustreid. Mitmed mobiilsusandmestikud (nagu näiteks Call Detail Records
mobiilside andmed) ning GPS andmed näitavad inimeste liikumissagedust ja -põhjusi.
Need andmestikud sisaldavad endas väärtuslikku informatsiooni ühiskonna kohta. Töö-
deldud informatsiooni saab kasutada mitmel otstarbel. Teadlased saaksid andmestiku
põhjal planeerida teedevõrgustikke, paremini suunata inimestele reklaame arvestades
nende paiknemist, luua uusi positsioneerimistehnoloogiaid, arendada rahvastikukont-
rolli tarkvara jne.
Vaatamata tehnoloogilistele võimalustele on inimeste mobiilsusandmestikud väga
raskesti kättesaadavad, sest need on kaitstud riiklike regulatsioonide poolt, kuna riivavad
inimeste privaatsust. Teine tegur on mobiilioperaatorite enda huvi luua inimeste mobiil-
susandmetel põhinevaid kommertslahendusi. Selline situatsioon ei innusta operaatoreid
jagama äriliselt vajalikku informatsiooni kolmandate osapooltega. Antud magistritöö
käigus näidatakse, kuidas sellest raskest probleemist üle saada arendades mobiilside-
võrgu käitumissimulatsiooni prototüüpi. Genereerides andmeid läbi erinevate teaduslike
liikumismudelite, mida võimaldab meile liiklussimulatsiooni tarkvara.
Uurimistöö tulemusena selgus, et selline lähenemine on resultatiivne ja omab mit-
meid laienemisvõimalusi. Täheldati mitmeid võimalusi koostööks teiste uurimisvald-
kondadega, et muuta genereeritavaid mobiilsusandmeid reaalelule sarnanevateks. Mo-
biilsidevõrgu käituvussimulatsioon on näidanud suurt potentsiaali ning arendamise käi-
gus avaldusid võimalused, mida algselt ei osatud oodata. Mainitud mobiilsidevõrgu käi-
tuvussimulatsioon on integreeritud eksisteeriva liiklussimulatsiooni tarkvaraga, mis on
vabavara ning mida on võimalik laialdaselt konfigureerida. Liiklussimulatsiooni tark-
varas kasutatavad inimkäitumise mudelid põhinevad erinevate teadustööde tulemustel
ning seetõttu mobiilsidevõrgu käituvus- ning liiklussimulatsiooni sümbioosi tulemusel
genereeritud andmed on märkimisväärse väärtusega.
Võtmesõnad: Liiklussimulatsioon, Mobiilsidevõrgu simulatsioon kasutaja perspektii-

3
vis, mobiilsed logid
CERCS: P170

4
Contents
Acknowledgements 7

Abbreviations and Acronyms 8

1 Introduction 11
1.1 Background . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
1.2 Motivation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
1.3 Objectives . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
1.4 Limitations and challenges . . . . . . . . . . . . . . . . . . . . . . . . 13
1.5 Road map . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13

2 State of the Art 15


2.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
2.2 SUMO . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
2.2.1 Vehicular communication . . . . . . . . . . . . . . . . . . . . 18
2.2.2 The SUMO Suite . . . . . . . . . . . . . . . . . . . . . . . . . 18
2.2.3 SUMO technologies . . . . . . . . . . . . . . . . . . . . . . . 22
2.3 Cellular Communication . . . . . . . . . . . . . . . . . . . . . . . . . 22
2.3.1 Mobility data . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
2.3.2 Cellular networks . . . . . . . . . . . . . . . . . . . . . . . . . 23
2.3.3 Basic cellular characteristics . . . . . . . . . . . . . . . . . . . 24
2.3.4 GSM architecture . . . . . . . . . . . . . . . . . . . . . . . . . 26
2.3.5 Generations . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
2.3.6 Mobility management . . . . . . . . . . . . . . . . . . . . . . 31
2.3.7 Description of main mobility events . . . . . . . . . . . . . . . 34
2.4 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35

3 Cellular communication network simulation 37


3.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37
3.2 Used technology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37
3.3 System architecture and design . . . . . . . . . . . . . . . . . . . . . . 38
3.3.1 Scope . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38

5
3.3.2 New models . . . . . . . . . . . . . . . . . . . . . . . . . . . 38
3.3.3 Visualization . . . . . . . . . . . . . . . . . . . . . . . . . . . 40
3.3.4 Core mechanics . . . . . . . . . . . . . . . . . . . . . . . . . . 40
3.3.5 Data import, output and feature configurations . . . . . . . . . 42
3.3.6 Extent of the development . . . . . . . . . . . . . . . . . . . . 43
3.4 MESGEN and HexagonGen tools . . . . . . . . . . . . . . . . . . . . 43
3.4.1 MESGEN . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
3.4.2 HexagonGen . . . . . . . . . . . . . . . . . . . . . . . . . . . 46
3.5 Mobility behaviour and management . . . . . . . . . . . . . . . . . . . 46
3.5.1 Microscopic simulation devices . . . . . . . . . . . . . . . . . 46
3.5.2 Mobility management . . . . . . . . . . . . . . . . . . . . . . 48
3.5.3 Wave propagation from the transmitter . . . . . . . . . . . . . . 52
3.5.4 Location update, Cell selection and reselection, Handover . . . 53
3.5.5 Mobility events and the simulation cycle . . . . . . . . . . . . 55
3.5.6 Mobility CDR-like logs . . . . . . . . . . . . . . . . . . . . . 55
3.6 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56

4 Results and Analysis 57


4.1 The mobility logs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57
4.1.1 GPS logs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57
4.1.2 Cellular network simulation logs . . . . . . . . . . . . . . . . . 58
4.1.3 MESGEN and HexagonGen . . . . . . . . . . . . . . . . . . . 60
4.2 Graphical upgrade . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62
4.3 The benchmarking . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62

5 Conclusion and Future Perspectives 63


5.1 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63
5.2 Future work . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64

References 74

Appendix 75
I. Licence . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75

6
Acknowledgements
First, I want to thank my supervisor Dr. Amnir Hadachi for instructing the thesis and
guiding me through the Dark Times. My eternal gratitude.
The thesis would not have been so interesting and educating without soon-to-be-a-
doctor Artjom Lind’s passionate talks about the cellular networking, its related tech-
nologies and algorithms. Thank you, Artjom!
Lastly, I would like to share my sincere gratitude to my friend Adam A. Waggoner
for his expertise in English language and for being supportive during the difficult times.

7
Abbreviations and Acronyms
This section clarifies main technological and philosophical terms used in the paper.

3GPP - 3rd Generation Partnership Project


AT - ATtention; this two-character abbreviation is always used to start a command line
to be sent from TE to TA
BS - Base Station
BSC - Base Station Controller
BSS - Base Station Subsystem
BTS - Base Transceiver Station
CC - Call Control
CDR - Call Detail Record
CM - Connection Management
CS - Circuit Switched
DBPSCH - Dedicated Basic Physical Subchannel
EDGE - Enhanced Data rates for GSM Evolution, 2.5G
EIR - Equipment Identity Register
ETSI - the European Telecommunications Standards Institute
GERAN - GSM EDGE Radio Access Network
GIS - Geographic Information System
GMM - GPRS Mobility Management
GPRS - General Packet Radio Service, 2.5G
GPS - Global Positioning System
GRA - GERAN Registration Area
GSM - Global System for Mobile Communications, originally Groupe SpécialMobile,
2G
GUI - Graphical User Interface
HLR - Home Location Register
IE - Information Element
IMSI - International Mobile Subscriber Identity
ITS - Intelligent Transport Systems
IVC - Inter Vehicle Communication

8
LAI - Location Area Identity
LAPDm - Link Access Procedure on the Dm
LMSI - Local Mobile Station Identity
LTE - Long-Term Evolution, 4G
MAC - Medium Access Control
ME - Mobile Equipment
MESGEN - Mobility Event Simulation Generator
MM - Mobility Management
MS - Mobile Station
MSC - Mobile-service Switching Center
NSS - Network Switching Subsystem
OpenGL - Open Graphics Library
OSM - OpenStreetMap
PD - Public Data
PDCP - Packet Data Convergence Protocol
PDN - Public Data Network
PDP - Packet Data Protocol
PLMN - Public Land Mobile Network
POI - Point Of Interest
PRAT - Rated Output Power
PSTN - Public Switched Telephone Network
RFID - Radio-Frequency IDentification
RLC - Radio Link Control
RRC - Radio Resource Control
RSCP - Received Signal Code Power
RSSI - Received Signal Strength Indicator
SACCH - Slow Associated Control Channel
SIM - Subscriber Identity Module
SRBCC - Single Radio Voice Call Continuity
SUMO - Simulation of Urban MObility
TI - Transaction Identifier
TMSI - Temporary Mobile Subscriber Identity
TraCI - Traffic Control Interface

9
TRAU - Transcoder and Rate Adaption Unit
UE - User Equipment
UMTS - Universal Mobile Telecommunications System, 3G
USIM - Universal Subscriber Identity Module
VANET - Vehicular Ad Hoc Network
V2D - Vehicle-to-Device
V2I - Vehicle-to-Infrastructure
V2V - Vehicle-to-Vehicle
V2X - Vehicle-to-Everything
VLR - Visitor Location Register
XML - Extensible Markup Language
XSD - XML Schema Definition

10
1 Introduction

1.1 Background
Modern lives are full of so much commuting and travelling that it has generated a
huge need for a large variety of technologies and solutions to manage, help and/or even
entertain the population. This preposterous quest has been adapted over time by the de-
velopment and spread of mobile technology, which has slowly alienated people from the
Public Switched Telephone Network (PSTN) and normalized cellular communication in
masses.
In 1982, The Confederation of European Posts and Telecommunications formed
Global System for Mobile Communications (GSM), which should render previous mo-
bile technologies (so-called 1G technologies) obsolete and lead Europe into the future
under one standard, what we nowadays call 2G. Less than 10 years later, Finnish Ra-
diolinja made the first GSM call in Finland, thereby announcing new possibilities to
researchers and enterprises, whom had been previously divided between different wire-
less telephone technologies. [70]
Standardization and the wide access of the mobile technology has boosted range of
research fields, most notably in this thesis’ point of view, like intelligent transportation
systems, geospatial technologies with access to Global Positioning System (GPS) data,
reproductively producing and enabling new technologies and solutions in the mobile
sphere itself. In 1996, the United States Federal Communication Commission issued a
set of rules to govern the availability of basic 911 services and the implementation of
wireless services, especially interesting for us, the locating the caller, who is in need
of help [7]. The positioning of people has been one of the key challenges in the field
of mobile technology. All phones do not have GPS or wireless network functionalities
and therefore, this sets multiple limitations in terms of possibilities to provide location-
based services. Before the smartphone revolution, government institutions and mobile
operators relied on GSM mobility data, stressed the necessity to use cellular information
for developing localizationing applications [33].
Cellular operators and mobile users have vital information to produce the next-
generation of mobile positioning algorithms, but unfortunately, there are major lim-
itations of accessing such data, because of various private data related laws and the

11
operators’ self-interest to develop money earning solutions. A further limitation is the
extent of effort it takes to gather data, even with the help of volunteers situated in various
research scenarios. The solution is to produce the mobility data through simulating the
human behavioural mobility in the urban environment, while attributing the simulated
model with a mobile device, which has its own characteristics.

1.2 Motivation
The main motivation of this thesis lies in the reality that different legal institutions
strongly protect cellular network related mobility data mainly because of users’ privacy
concerns, and limited due to competitive strategies between companies. Therefore, in
order to investigate wireless communication solutions and applications, we need an easy
way to produce training and testing datasets.
Through simulating people’s real life commutes and social behaviours we could
generate a source of information which could be used for understanding human mobility
patterns. These patterns could assist in various goals: designing and engineering cellular
networks, urban infrastructure and transport planning, advertising, event and population
management, epidemics monitoring and control, and analyzing and predicting issues
related to human migration.
By adding cellular network on top of the traffic simulation, we get a possibility to
simulate Vehicle-to-Everything (V2X) communication through the mobile technology,
more accurately Vehicle-to-Infrastructure (V2I), while a cellular device is attached to
the vehicle and communicates with a mobile operator.

1.3 Objectives
This thesis has two main divisions: open-source SUMO traffic simulator [4] and
cellular network behaviour. The primary objective is to research the cellular network
components’ and its devices’ functioning to prototype their operation principles in the
traffic simulator and produce synthetic cellular network logs.
This requires us to work through the vast codebase to analyze, design and imple-
ment new tools, integrations, and features to finally have, from 3GPP [1] specifications
derived, mobility data producer.

12
The objectives:

• Research SUMO codebase for upcoming developments;

• Design XML importers and exporters;

• Design of cellular network;

• Representation of cellular towers and cells graphically;

• Mobility event simulator for depicting human’s mobile behaviour;

• Simulation of Mobile behaviour inside of user defined cellular network;

• Investigate and prototype of basic vehicle-to-infrastructure communication and


data generation.

1.4 Limitations and challenges


The challenges comprised of SUMO and cellular network related documentation.
SUMO has a very long development history, and therefore, some of the technologies
have become legacy and even deprecated. Over the years, there have been many engi-
neers with different styles and preferences, but general guidelines for the development
have kept somewhat order in the project.
Another issue lies in the development pipeline, the lack of feature branching and ac-
cepting only tested code into the master could lead into better project health. Currently,
regression tests in the project note if it is a known bug or not. In software engineering,
the master branch is typically reserved for review and tests passing code.
3GPP specifications about GSM and newer standards had information divided into
many documents and therefore, for example, internal device communication between
different layers and cellular network components were not always easily understand-
able, resulting in ambiguity or certain corner cases being unexplained or explained in
another specification document.

1.5 Road map


The thesis dissertation has been structured in the following manner.

13
Chapter 2: Presents the State of the Art about the SUMO traffic simulator and cellular
communication technologies.

Chapter 3: Outlook of the work which has been done during the implementation of the
cellular network simulation.

Chapter 4: Describes the results which have been achieved during the development
and testing.

Chapter 5: Concludes the work, supplying results and giving an insight into future
perspectives.

14
2 State of the Art
The chapter is describing applications and technologies behind the traffic simulation
package and gives basic knowledge about how mobile technology works on the view-
point of this thesis. Providing an overview of relevant literature and existing alternating
works, definitions and their properties used in the paper for designing and developing
the cellular network behavioural simulation prototype in cooperation with existing traf-
fic simulation software.

2.1 Introduction
The need for mobility data for science research to determine user location, predict
or classify human behaviour in various scenarios can be exhibited by a few examples
which describe, with some variety, the dependence on mobile specific data. Authors in
search of a localization algorithm, are stressing a need for of minimal network location
technology as an alternative to current triangulation or trilateration based technologies,
because those technologies are incapable of localizing the targets when there is only one
distinct base station. In their dataset, approximately 50% of Call Detail Records (CDR),
were served by only one base station. People who are developing mobile applications
and depend on the localization feature do not have available full and easy access to
CDRs, due to privacy concerns and anonymization challenges. Therefore, the authors
emphasize the need to develop technologies which are qualified to work with minimal
network information. One article proposes a solution by using Bayesian inference for
the localization problem and managed to achieve a 20% growth in accuracy when they
switched from blind approach to their newly implemented solution. [59] Needless to
say, access to generated data with one base station antenna coverage is exactly a scenario
which can be resolvable with cellular network behaviour simulation software.
Positioning and human behaviour research have been quite intertwined over the re-
cent decades when it comes to extracting knowledge how people act and move in so-
ciety. One paper proposes a methodology to detect social events by using the cellular
information with their Bayesian location inference framework. They differentiated mo-
bility behavior among social event attendees and people’s everyday routine by retracing
attendees steps. This approach has been deemed useful in managing crowds and detect-

15
ing potential emergency situations. [75] The simulation of human behaviour could be
defined through the traffic simulation, which has vehicle, public transport and pedes-
trian support with a possibility to define every entities’ mobility pattern from various
researches, which has been done in the respective field.
One example of cellular network’s designing and altering capabilities to suit re-
search scenario needs, is the changing of 2G cellular network macro cells to 5G micro
cells to produce training and test data for model development. This is a very realis-
tic scenario, as several researchers have raised the need for optimal network planning
for future heterogeneous networks with high number of small cells since growth in the
number of users will increase the density inside the cell. Underlining the importance of
geometric modeling for network design and analysis, because currently heterogeneous
networks are moving towards to a highly irregular network infrastructure and therefore
causing interference, optimal resources allocation may be negotiated and that all results
in difficult load balancing between small cells and macro cell due to random network
topology and non-deterministic users’ behavior. [72] The thesis’ prototype serves as a
possibility to model and test the new infrastructure and how it responds in compare of
previous versions.
The need has been verified by slightly different application examples from the mo-
bile technology sphere. To solve the problem there has been taken SUMO traffic sim-
ulation software to overcome human movement in the urban context and implemented
cellular networking behaviour to generate mobility logs.

2.2 SUMO
SUMO (Simulation of Urban MObility) is a traffic simulation package which have
been developed by the employees of the Institute of Transportations Systems at the
German Aerospace [2] and SUMO community. It is an open source project, licensed
under GNU General Public License (GPL) [30], with considerable support from exter-
nal authors who appreciate the software’s high portability, extendibility, performance,
microscopic design to handle serious large road network simulations without inbuilt
limitations towards limiting the network size and number of simulated entities.
Academics from German Aerospace Center started developing SUMO in 2001 and
released the first open source version in 2002. The idea behind the open source is simple

16
Figure 1. SUMO (Simulation of Urban MObility), example of running real-life graphi-
cal simulation, location is Riia-Turu streets’ intersection in the city of Tartu.

and honest: to give people a free simulation software. For students and researchers, this
is a possibility to implement their own research tasks in SUMO environment without re-
inventing all the simulation steps again for the projects, which usually get lost in void
eventually after author’s graduation or end of research. Furthermore, SUMO allows
for comparison and benchmarking against other technological approaches. [68] As an
example of continuous research and development, Stefan Krauß developed time-discrete
microscopic traffic flow model for his PhD thesis; this model is now being used for
SUMO vehicle simulation [64], meanwhile other simulation specific maneuver: lane-
changing movement was done during the development of the core simulation [62]. The
scientific collaboration and coherence has expanded the simulator, while concentrating
infrastructure, vehicular, and traffic behaviour related knowledge into different models
to assist other researchers.
To understand the extent of SUMO community and its doings, the author of this the-
sis would like to point out this year’s SUMO User Conference - “Towards Simulation
for Autonomous Mobility (May 8-10, 2017)” [3], where people from numerous uni-

17
versities and vehicle manufacturers, like Volkswagen and Volvo, held talks on various
topics relating to SUMO. A detailed wiki page [25] about the application package and
an active e-mail mailing list also exist.

2.2.1 Vehicular communication

This thesis works on the cellular communication, setting a cellular device to every
simulated vehicle (there could also be another simulation routable entity, like pedestrian,
bicycle driver). Combining a vehicle with a mobile phone makes it possible to conduct
more technical simulation scenarios from the cellular networking point of view, which
opens us to Vehicle-To-Everything (V2X) research possibilities.
V2X basically means communication between vehicle and an entity, which might
affect the vehicle or vice-versa. Vehicle-To-Vehicle (V2V) communication has been
simulated with SUMO in various research scenarios with the help of third-party soft-
ware. Bidirectionally coupled framework, which consists of network simulation and
road traffic microsimulation, has been built to evaluate VANETs. VANETs are Vehicu-
lar Ad Hoc Networks, which were first introduced in 2001 as “car-to-car ad hoc mobile
communication and networking” applications [74]. SUMO was used for the traffic sim-
ulation, while vehicle network simulation was done with OMNeT++ /citeomNet. The
paper concluded that bidirectionally coupled traffic and network simulation provided
major advantages and benefits to simulate Inter Vehicle Communication (IVC) cite-
sommer2008need The bidirectional networking proved to be successful and gave im-
portant feedback to the researchers, resulting in more detailed future simulations where
different VANET protocols were introduced and significant confirmation about IVC’s
impact for the vehicles’ mobility was received [71].
The traffic simulation combined with different types of simulation software offers a
new dimension for research, including the possibility to concentrate on more problems
and problems in wider area that the tool was not at first intended to use for.

2.2.2 The SUMO Suite

The authors’ do not only classify SUMO as a traffic simulation, but rather, a bundle
of applications which help to perform the traffic simulations to user’s specific scenarios.
The main application has two versions: command-line based batch simulation program

18
and graphical interface powered live simulation.
The traffic simulation is not limited to road vehicles. There is also support for pedes-
trians and the community is currently working on bicycle, railway, and waterway simu-
lations. All those subjects are good candidates for researchers to propose dynamic traffic
flow models. Additional features of SUMO are logistics support, where the user has to
define goods shipment stations, where the truck typed vehicles can stop to load/unload
the merchandise, parking, public transport, emergency vehicles and electric vehicles
support. Those mentioned features are not definitive number of features; there are more
smaller features, plus vast configuration possibilities to all features to alter the traffic
simulation scenarios to the user’s specific needs.
The application bundle has multiple core programs along with the simulation and
they vary from generating road network to creating meaningful mobility routes for the
simulated entities. Below are the core programs, so-called main applications, needed
for traffic researchers:

SUMO command-line software for simulating a defined scenario.

SUMO-GUI adds to SUMO program graphical user-interface with multiple view op-
tions to overview running simulation and make sure everything works according
to the defined scenario.

MESO refers to mesoscopic simulation model which does not have all the features like
SUMO microscopic simulation. Known limitations are missing TraCI support,
not existing public transport, electric model, wireless model, opposite-direction
driving, sublane-model.

NETGENERATE support application to generate abstract traffic networks, which are


usable by other programs. Has interesting solutions for generating grid networks,
which reminds Manhattan’s design, spider styled road networks, reminding like
the name says spider web, which in turn reminds Moscow’s road infrastructure on
huge scale, and finally random network.

NETCONVERT is one of the most important tools of SUMO, because the community
has put big effort to build multiple import styled integrations with other traffic
simulators to enhance compatibility and collaboration. Currently, the importing

19
supports the following formats: SUMO’s own XML format, OpenStreeMap’s
[18] XML file, third party traffic simulators network files from VISUM [22],
Vissim [21], OpenDRIVE [15] and MATsim [13], ordinary shapefiles and some
lesser known formats. NETCONVERT also supports exporting to some external
formats, like the previously mentioned OpenDRIVE and MATsim. In the Fig-
ure 2a we can see selected area of city of Tartu, which we would like to import
to the program, and in the Figure 2b we can see the successful result after the
process. The city of Tartu with its vehicle roads, pedestrian ways, and even train
tracks, have been imported to SUMO.

(a) Selected region, what we want to import to


(b) Previously selected area after the import pro-
SUMO, the city of Tartu in OpenStreetMap.
cess in SUMO GUI.

Figure 2. Selecting a region from OpenStreetMap to import into SUMO for simulation.

NETEDIT is a visual network editor to create and modify SUMO networks. NETEDIT
is NETCONVERT’s extension to debug and fix broken network attributes.

OD2TRIPS imports trip distribution related origin-destination matrices to generate


trips, which can be converted into simulation ready vehicle routes.

DUAROUTER imports different demand definitions, which help to define vehicles’


behaviours and how they will act (selecting paths) in the traffic on a microscopic
scale.

20
JTRROUTER computes routes from the junction probability values and the traffic
volume. Some junctions have probabilities to which road the vehicle proceeds.

DFROUTER generates routes from the statistics collected from high-way related en-
tering and leaving flow data, which is being actively collected and therefore ap-
plicable to generate real-life like simulations.

MAROUTER calculates routes from the origin and destination matrices.

POLYCONVERT is being used for importing geometrical shapes to visualize SUMO-


GUI simulation.

ACTIVITYGEN generates people’s mobility in the simulation, for example, people


going to work, school, having free time with various means of transportation like
car, bike, bus, and even walking.

EMISSIONMAP tool produces matrices of emissions, which could be used in the sim-
ulation for emission simulation along the general traffic movement.

EMISSIONDRIVINGCYCLE is another emission tool, which computes emissions


given a timeline of speeds and accelerations for a single vehicle.

TraCI is an abbreviation coined from “Traffic Control Interface” and behind that term
is a tool which helps to access and control running traffic simulation. During
the simulation, the user could debug the simulated entities and manipulate their
behaviour to alter scenarios for research needs. [27]

SUMO also has additional tools. For example, this thesis used randomTrips.py tool
for generating random trips with generated vehicle ids, input calculated departing times,
random starting lane-locations, and destination lane-locations. With DUAROUTER
program, the random trips generated file was converted into a route file, where vehicles’
routes have been already calculated to save the performance during the simulation.
More details about SUMO and its applications’ functionality can be acquired from
its documentation page in wiki and its related articles. [27] [68] [63]

21
2.2.3 SUMO technologies

SUMO main applications have been programmed in C++ programming language,


while some additional tools have been scripted in Python. The unit tests are done with
C++ library Google test. The regression tests of the applications have been done by
Python framework TextTest [28], which is an open source tool for functional testing.
The following frameworks are most notable for main simulation software:

FOX Toolkit [8] is a cross-platform framework for graphical user interface develop-
ment.

Xerces-C++ [29] is a validating XML parser to read and write XML data.

GDAL [10] is an abbreviation for Geospatial Data Abstraction Library. It is a translator


library for raster and vector geospatial data formats.

proj.4 [20] is a library which converts geographical coordinates into cartesian coordi-
nates and in the opposite manner as well.

OpenGL [16] is an application programming interface for rendering two and three-
dimensional vector graphics. It has a cross-language and cross platform support.

2.3 Cellular Communication


2.3.1 Mobility data

One of the results of the thesis project is simulated cellular mobility data logs. Mo-
bility data is a set of different types of information entries recorded with telecommuni-
cations and ubiquitous computing where the locations of mobile user are continuously
sensed. The regulations about data privacy are very serious; it has been regulated thor-
oughly by the legislators to protect people’s’ anonymity and privacy. The information is
usually shared by the mobile operators under the user agreement, which impose limits
to the research. That is the reason why there anonymity data processing is necessary
before sharing the datasets. [32] Another possibility is to generate your own dataset,
which relies on other science fields works to produce real-life like datasets to train and
test your work.
Some mobility data examples:

22
• Mobility data can be produced from the mobile phone conversations, as the mo-
bile operator knows your location in the cellular network through positioning
techniques.

• Data recorded with GPS devices; modern smartphones in these days have many
applications which rely on geospatial information.

• Data exchange between vehicles in VANET environment generates entries about


the vehicular communication.

• Radio-Frequency Identification (RFID) systems generate logs, which have your


location determined and timestamped by RFID reader.

• WiFi access points give information about user’s position to service provider. [69]

2.3.2 Cellular networks

Figure 3. Example of pre-cellular mobile technology.

Modern mobile network, also known as a cellular network, is a communication net-


work which connects the mobile - User Equipment (UE) to the infrastructure in the
wireless manner. The name cellular comes from the network distribution over land: it
creates cells of coverage to provide services in every location where needed. [56]
The cellular network concept was created in 1949 by Bell Labs with suggestions
to fix pre-cellular mobile technology problems like handoffs between transmitter with

23
Figure 4. Example of cellular mobile technology.

reuse of the frequencies [70]. Before that, you had to re-initialize your connection when
you moved to new base station’s (BS) zone, which is visible in the Figure 3. In 1971,
Bell Labs revealed a potential cellular system’s design [65] [67] and 8 years later, the
first commercially usable cellular system started to work in Tokyo in 1979, by Nippon
Telegraph and Telephone [70].
The first cellular networks have been titled as first generation (1G) networks. In
the Figure 4 we can see an example of cellular network, with two transmitters covering
through sectorization technique 6 cells (more precisely they are called sectors and the
cells in that case are the group of one transmitter’s sectors). From 1979 onwards to 1991,
when the first 2G call was made in Finland, lots of different 1G technologies were in
use. Eventually, the Confederation of European Posts and Telecommunications formed
Global System for Mobile Communications, a pan-European standard, because Europe
was clustered between different analog and only voice call supporting standards and
technologies, which caused problems to roam seamlessly from one country to another.
GSM was meant to fix this problem.

2.3.3 Basic cellular characteristics

Network architecture is based on lots of cells covering large or small geographical


regions, establishing the service region where the mobile operator offers its services. A

24
Figure 5. The visual presentation of umbrella cell and 120 ◦ sectored cell.

cell is a resemblance of radio coverage emitted by the base station (BS) [35]. Cells are
shaped as hexagons because the circular energy pattern emitted by the omnidirectional
antenna from the base station depicts what is close to a hexagon shape [66]. While
designing the hexagons help to avoid overlapping or non-filled areas. A group of cells
produce a cluster. The cellular coverage has been achieved by using two approaches:
omnidirectional and sectorized cells. Omnidirectional cells are usually huge umbrella
cells covering large territories, classified as macro and boomer cells, defined in Table 1.
Sectorization means that instead of 360 ◦ omnidirectional umbrella cell, which covers
all-around with one antenna, the 360 ◦ around the tower is divided into, for example,
three 120 ◦ sectors. In the Figure 5 it visible that how Cell 2 is covering three sectors,
while using directed antennas, while Cell 1 creates “umbrella” kind of coverage over
the area.

Table 1. Non-standardized cell classification. [70]

Class Radius Typical deployment scenario


Boomer > 100 km (up to 240 km) Large, open space (e.g., plains)
Macro Typically 1 to 20 km, sometimes up to 100 km Rural and suburban areas
Micro 100 m to 1 km Urban areas
Pico Less than 100 m Inside buildings
Femto Few meters Homes

For the second, the communication link between the Mobile Station (MS) and BS

25
over air by transferring radio waves.
Key characteristics is frequency reuse. Adjacent cells must have different frequen-
cies, because if the neighboring cell provides the same frequencies, then that would
cause co-channel interference. Therefore, frequency reuse means that adjacent cells
have different frequencies in use, but further away residing cells can reuse the frequen-
cies without worrying about negative effects, since the distance and low powered trans-
mitting equipment are withdrawing the interference related problems [56]. The topic
of how frequency spectrum is allocated between different technologies and finally be-
tween users is out of this paper’s scope. If the reader is interested, then the topic could
be explored in multiple books which cover the fundamental concepts of cellular net-
working [70] [56].
The last basic characteristic, which this paper will address is the concept of ar-
chitecture, which will be described separately in the next chapter with GSM, second
generation cellular network. GSM was picked because it was the first pan-European
(even global, with some exceptions) standard on top of which upcoming generations
have been built with backwards compatibility. Therefore, it is still active and funda-
mentally important to know, even though new technologies have introduced some figu-
rative aggregative simplifications to the architecture and could be easier to explain and
understand.

2.3.4 GSM architecture

Public Land Mobile Network (PLMN) is a network operated by a government owned


administration or private mobile operator for providing collection of land mobile telecom-
munications service to the public. [35] PLMN of GSM has been described visually in
the Figure 6, where we can see the radio network and fixed network (Network Switching
Subsystem, abbreviated as NSS) with interface names from the user equipment to NSS.
PLMN may have one or more Mobile-service Switching Centres (MSC), which or-
ganize the communication between the radio system and the fixed networks as the inter-
face. MSC performs all the signaling and switching to the mobile phone, which resides
in its area. Main procedures are the location registration and handover between different
cells. [40] [37]
Several MSCs may be required to cover an entire country, but MSC is just merely a

26
Figure 6. The basic architecture of GSM, second generation cellular network.

sub-unit in the communication between the radio system and the fixed network. There-
fore, to achieve radio coverage of geographical areas, the mobile operator needs Base
Station Systems (BSS).
Communication between MSC and BSS occurs through A-interface, which car-
ries information about BSS management, call handling, and mobile mobility manage-
ment. [38]
The base station system is a system containing vital equipment to build communica-
tion between MSC and Mobile Station (MS). Vital equipment consists of Base Station
Control (BSC) and one or more Base Transceiver Station (BTS), which each serve one
cell; therefore, BSC controls multiple cells through BTSs.
The Base station control keeps track of its one or more BTSs through Abis interface.
This interface manages the services offered to the mobile stations, allows control of
radio equipment, and allocates radio frequencies in the BTS.
The mobile stations are physical user equipment to interact in PLMN network. The
physical user equipment consists of Mobile Equipment (ME) and the subscriber identity
module (SIM).
Um-interface administers the communication between the mobile stations and the
BSSs.
Inside of PLMN’s fixed network, also known as network switching subsystem or

27
GSM core network, there are components which carry out the call switching and mobil-
ity management functions. MSC is one of them, but there are many other components:

Home Location Registration (HLR) is a central location register which stores sub-
scriber information, location information to charge and route the call to appro-
priate MSC, where the MS is positioned. The subscriber information holds In-
ternational Mobile Subscriber Identity (IMSI) and one or more Mobile Station
International ISDN (MSISDN) numbers [39]. Integrated Service Digital Network
(ISDN) is a technology which enabled to have simultaneous digital transmissions
over the circuit-switched public switched telephone network [34].

Visitor Location Register (VLR) is location register closely bound to MSC, which
oversees the user equipment’s area transfers. MSC uses VLR for retrieve infor-
mation. For example, it uses VLR to handle calls for the mobile stations situ-
ated inside of MSC’s jurisdiction. VLR stores IMSI, MS ISDN number, Mobile
Station Roaming Number (MSRN), Temporary Mobile Station Identity (TMSI),
Local Mobile Station Identity (LMSI) and the location area of MS [39].

Gateway MSC (GMSC) is, during the delivery of the call to PLMN, the specific MSC
which after the interrogation about the MS location with HLR, performs the rout-
ing to the actual location of the MS. [36]

Equipment Identity Register (EIR) is a logical entity which stores and shares the In-
ternational Equipment Identities. It contains MS classification states. MS may be
white listed, grey listed, black listed or unknown. [35]

Camel Service Environment is another logical entity which offers operator-specific


services. [35]

Short Message Service Center (SC) stores, forwards, converts and delivers SMS mes-
sages. [35]

SMS Interworking MSC is between PLMN and SC, transferring SMSs from the MSs
to SC. [35]

SMS Gateway MSC is a unit which determines suitable GSM PLMN that can send
SMS from the short message service center to the MS. [35]

28
Authentication Center (AuC) handles and stores each mobile station’s IMSI, respon-
sible for authentication and ciphering communication over the radio path - ci-
pher key management. During the authentication procedure, the ciphering key is
set. [42]

2.3.5 Generations

We have already discussed the pre-cellular era, the shortcomings of the first radio
telephone systems, and how the Bell Labs led research towards the first cellular net-
works. Pre-cellular era technologies are called 0G, since they are predecessors to the
first generation technologies. 1G had issues as well; the roaming all over the Europe
was made difficult due to following facts. At the time 1G was to be deployed, there
were newer multiplexing and channel accessing technologies available. Therefore, the
European Telecommunications Standards Institute (ETSI) was created with a focus on
working out the unique solution with the latest technologies. As a result, 2G solution
called GSM was declared the pan-European standard of mobile communication.
On top of GSM, multiple levels of technologies have been built, maintaining back-
wards compatibility. After decades, there have been numerous upgrades and currently
there is ongoing work to release 5G commercially in 2020 [73]. As a result, we are
exposed to different generations of mobile technologies, which have been or are being
used by Estonian mobile operators :

1G is a code-name for the analog signal based telecommunications standards. 1G in-


troduced the cellular network concept. A very important concept related to this
topic is Frequency-division multiple access (FDMA). FDMA is a channel access
method where the users get a channel which is divided into two frequency bands:
one band for uplink (MS to BTS) and one band for downlink (BTS to MS). In
FDMA, the unique channels are multiplexed into one carrier frequency a using a
modulation. Multiplexing is a method used to combine multiple analog or digi-
tal signals into one shared medium and, with multiple access methods makes it
possible for numerous transmitters to connect to the same physical medium and
share its capacity. This approach’s main flaws are a small maximum data rate, and
the use of multiple frequency bands and band guards to avoid interference limits
overall capacity of users. [60]

29
2G went from analog signal to digital, bringing along a new multiplexing method to im-
prove voice capacity and quality. GSM presented Time-division Multiple Access
(TDMA). Time-division duplexing is a concept where at any given time happens
only downlink or uplink transmission between two connected parties, but with
TDMA the resource can be accessed by multiple parties. [60]

2.5G generation introduced two new technologies: General Packet Radio Service (GPRS)
and Enhanced Data Rates for GSM Evolution (EDGE). Before, GPRS users ac-
cessed data services in a circuit-switched data approach, which means that two
network nodes have to establish a dedicated channel for communication, known
as a circuit. This approach has clear downsides, there are possibilities that the
channel is under-loaded when the user is less active (the channel is allocated for
the user). In contrast to that, the channel can be overloaded by a single active user.
Old circuit-switched systems often limited themselves to manage only one circuit
connection per device. GPRS established a new packet-switched core network to
support packet data services in a more efficient manner than old circuit-switched
solution. Packet-switching is a concept where data gets divided into packets and
transmitted over the network independently. Channel dedication became obso-
lete and packets were transmitted through channel which were now shared by
multiple simultaneous communication sessions. EDGE brought radio network
enhancements to GPRS, which resulted in higher data rates while using packet
data services. EDGE added no new NSS components. From 2G to 2.5G we saw
better resource management and, over time, more efficient and improved packet
data services. [41] [46]

3G managed to grow the capacity of the users and brought data rates in hundreds
of kbps. ETSI consortium called this generation mobile technology as Univer-
sal Telecommunications System (UMTS). Another channel access method called
Code-division Multiple Access (CDMA) was standardized. CDMA used wide-
band transmission, introduced new universal frequency reuse, improved inter-
ference control, reliable handovers, and power settings. [52] [43] This spread-
spectrum multiple access method allows several users to share a band of frequen-
cies, since a special coding scheme assigns to each transmitter its own code [56].

30
4G has high data rates in Mbps, using yet another multiplexing technique called Or-
thogonal Frequency-Division Multiplexing (OFDM), an IP-based network archi-
tecture, beared down packet-switched core network. Efficient and cost-effective
solution for modern high data demand. [52] OFDM encodes data on multiple
carrier frequencies. Each sub-carrier uses modulation schemes like quadrature
amplitude modulation or phase-shift keying [55].

5G is coming up in the future and has currently mapped out some of the requirements
which the new standard should address.

2.3.6 Mobility management

Figure 7. Overview of mobile station’s communication layers with its another end
(BTS).

Mobility management is an important concept. This section describes MS related


mobility management (the network also has its own mobility management), because
during the development of the simulation, the devices’ behaviour was programmed ac-
cording to GSM/GPRS mobility management, while GPRS related session management
was not included.
As visible in the Figure 7, communication between MS and BTS through Um inter-
face - GSM signaling protocol has been gathered into three layers:

31
Layer 1 is the physical layer, which offers transmission on a limited set of logical chan-
nels. Logical channels are multiplexed on physical channels. [44]

Layer 2 is the data-link layer. Layer 2 bases on Link Access Procedure on the Dm
channel (LAPDm), which provides a reliable signalink between the MS and the
network . [51]

Layer 3 has three sub-layers:

Radio Resource Control (RRC) manages the common transmission resources


(physical channels) and the data link connections on control channels. It
contains of procedures to establish, keep running, and release the dedicated
radio connections. [50]
Mobility Management (MM) sublayer supports the mobility of MS; it informs
the network about the location through location update (also periodic loca-
tion update), manages authentication, identification, and IMSI related pro-
cedures etc. [54]
Connection management (CM) is a sub-layer, which also consists of three sub-
layers Call Control (CC), short message services (SMS) and supplementary
services. [49]

Mobility management depends on the RR connection. It calls out the establishment


of RR connection to perform its own procedures. Import aspect is MM states, which
were essential for the simulation. While RRC has two main states, IDLE MODE and
CONNECTED MODE, the MM has many more states, which are dependent on the RRC
states. During RRC IDLE MODE the MS mobility management resides in its own MM
IDLE state. But when the mobility management conducts its procedures, it establishes
RR connection and is maintained until the MM returns back to its IDLE mode or the
user equipment gets turned off. Mobility management has following main states:

• NULL,

• LOCATION UPDATING INITIATED,

• WAIT FOR OUTGOING MM CONNECTION,

32
• MM CONNECTION ACTIVE,

• IMSI DETACH INITIATED,

• PROCESS CM SERVICE PROMPT,

• WAIT FOR NETWORK COMMAND,

• LOCATION UPDATE REJECTED,

• WAIT FOR RR CONNECTION (LOCATION UPDATING),

• WAIT FOR RR CONNECTION (MM CONNECTION),

• WAIT FOR RR CONNECTION (IMSI DETACH),

• WAIT FOR REESTABLISH,

• WAIT FOR RR ACTIVE,

• MM IDLE,

• WAIT FOR ADDITIONAL OUTGOING MM CONNECTION,

• MM CONNECTION ACTIVE (GROUP TRANSMIT MODE),

• WAIT FOR RR CONNECTION (GROUP TRANSMIT MODE),

• LOCATION UPDATING PENDING,

• IMSI DETACH PENDING,

• RR CONNECTION RELEASE NOT ALLOWED. [54]

MM IDLE has its own substates, which are important because they reflect the state
of the device even more accurately, e.g., not having the connection or no valid subscriber
data(SIM card missing). The substates are:

• NORMAL SERVICE,

• ATTEMPTING TO UPDATE,

33
• LIMITED SERVICE,

• NO IMSI,

• NO CELL AVAILABLE,

• LOCATION UPDATE NEEDED,

• PLMN SEARCH,

• PLMN SEARCH, NORMAL SERVICE,

• RECEIVING GROUP CALL (NORMAL SERVICE),

• RECEIVING GROUP CALL (LIMITED SERVICE),

• eCALL INACTIVE. [54]

2.3.7 Description of main mobility events

In this subsection, we are going to discuss project related mobility procedures (do
not confuse with mobility management procedures, but rather think about overall mo-
bility procedures), which were important for the simulating cellular network behaviour.
The simulation relevant mobility procedures:

Handover is an RRC specific procedure to change channels (intercell, intracell), deter-


mined by the MS or the network on the basis of signal level, connection quality,
power level propagation delay, and the traffic reduction between the transmit-
ters. [48]

Cell selection is an RRC function during which MS tries to find a suitable cell to camp
on, while cell reselection is a process where the MS is periodically looking for
a better cell to camp on. [45] In this thesis scope, the cell selection and the cell
reselection are synonym since their workflows in the code level are the same.

Location update is MM specific procedure, which occurs during normal location up-
date, other procedures, periodic location update, and when IMSI is attached (SIM
card inserted). The location update registers or updates the mobile station’s actual
location area in the network. Periodic update procedure is controlled by a T312

34
Figure 8. Simple overview about initiating a call and what happens on the connection
management layer. [54] [47] [53]

timer in the mobile station. When the timer hits the constant with a counting, then
it starts location update procedure. [54]

During the cellular network simulation, there are happening the mobile events like
calling, sending SMS messages or web connection open over the communication man-
agement layer. On the top of the connection management, during the call, there is the
call control managing the state, which handles the incoming and outgoing call events. In
the Figure 8 we see the process when the call is originating from the MS. In the Figure 9
we can examine what is happening in the terminating end of the call.

2.4 Conclusion
In this chapter, we introduced the microscopic traffic simulator SUMO and the back-
ground of cellular networking, including some specific details about the architecture in
order to familiarize the reader with the topic and to present the scope’s boundaries to

35
Figure 9. Simple overview about terminating end of a call and what happens on the
connection management layer. [54] [47] [53]

understand on what extent the practical part of the work has been implemented while
reading the next chapter about the cellular communication network simulation. There
was also some important extra reading about the progress of cellular mobile technology,
and a subsection about generations, without which we cannot explain the essence of
cellular networking.

36
3 Cellular communication network simulation
The intent of this chapter is to describe the core of this thesis; what, how and why
something was done and eventually what kind of technical outcome was achieved in
goal to produce cellular network behavioural data simulator.

3.1 Introduction
The extent of the work to understand the cellular network and to model it into SUMO
traffic simulation had considerable difficulties. The learning curve to understand the
basic concepts of both components demand lots of resources. Modelling architecture of
over decade old code to realize the work flow and to design new features on top of that.
Development was burdened with many project specific demands about documentation
and C++ related robustness about writing; slow writing speed and poor stack traceless
error management. Learning cellular network required sketching and visioning different
systems and related technologies to grasp how the mobile technologies are working
nowadays.
The vastness of cellular network made it clear that for the simulation of cellular
behaviour there is a need to define a scope, which is achievable during the thesis limits.
The NSS specific functionality has been left out or delegated into the mobile station.
Primary goal of the thesis is to manage the mobile device to generate the cellular logs,
similar to CDR through the symbiosis of traffic and cellular simulations.

3.2 Used technology


The technologies used in this thesis are primarily influenced by SUMO traffic sim-
ulator, in the perspective of the development task means that technologies described in
the Chapter 2.2.3 were also used by the author.
For C++ there was used GNU Compiler Collection version 4.8.4 [9], project code
was managed with Git [11] and integrated development environment was Eclipse CDT [5].
Some minor scripting was done in Python 2.7 [24] with PyCharm [23]

37
3.3 System architecture and design
3.3.1 Scope

In the Chapter 3.1, the scope was incoherently defined; in this subsection, the exact
borderline will be explained. The central piece of this entire simulation is the user
equipment, which produces CDR-like activity logs.
The network has been constructed in a manner that resembles a radio system, with
one component from the NSS - mobile-service switching center. In the code level de-
sign, tasks of BSC and MSC are aggregated into one entity - towers control (SUMO
architecture influenced buildup), while their main functionalities are removed or del-
egated to the mobile device class, where they have been also pruned and emulated in
lower extent.
Mobile devices communicate with transmitters (called antennas in the code level).
Transmitters are tied with physical tower entities, which are representing base station
systems. BTSs are container type of classes to store tower’s antennas and its location
related information.
From the Chapter 2.3.6, we know that the mobile device and its communication with
BTS over Um interface is vast and complicated due to the different protocol layers. The
user equipment mainly has the mobility management layer, which is defined with some
radio resource layer specific functions. The mobility management will be described in
the Chapter 3.5.2.
Upgrading SUMO GUI to visualize the tower position and its transmitters coverage
areas - hexagons. SUMO has since version 0.26.0 mesoscopic simulation model [26],
but in this thesis, we want to produce individual entity level mobility logs about its
cellular behaviour through the microscopic traffic simulation. Therefore, we are not
utilizing mesoscopic model.

3.3.2 New models

Cellular tower is the presentation of the physical tower with the following attributes:
point of interest, which contains the cartesian and real life geospatial information;
list of cellular antennas; and connected mobile devices connected with tower’s
transmitters.

38
Cellular antenna , according to the cellular network definitions a transmitter, is a data
structure that facilitates information about its coverage as polygon shape (e.g.
hexagon, custom polygons are also supported). Antenna also have these following
attributes: signal strength, frequency, radius, and reference to its tower. It also has
an id value, which represents Common Gateway Identifier (CGI) value.

GPS device is defined as in-vehicle device producing data, known as geographical co-
ordinates. The device and its subclasses have been explained more thoroughly in
the Chapter 3.5.1.

Mobile device is another in-vehicle device which is used for the cellular network be-
haviour simulation. The device and its subclasses have been explained more thor-
oughly in the Chapter 3.5.1.

Mobility Event Data is a data structure model that contains a single mobility event
which has been generated by Mobility Event Simulation Generator (MESGEN),
which has been developed within this thesis. The model has three attributes: event
start, event end and event type. The mobility event data has been labeled into
following types:

NONE refers to error state, the event is in the void state;


CALL_INIT is an event for the call initiation process;
CALL_REC , incoming call event;
SMS_SEND - event describing that the device sends a SMS;
SMS_REC is for the device received a SMS;
WEB_COMM describes the ongoing web communication;
KEEP-ALIVE event is not generated, but it occurs during the simulation when the T312
timer initiates periodic location update;
RADIOLINK_FAILURE is another event which occurs during the simulation, bound to the signal
propagation and losing the connection to the transmitter and no handover
candidates were found. Radio link failure is an RRC layer technical term for
losing the connection.

39
3.3.3 Visualization

The scope of this thesis aims to support the command-line and the visual simulation.
Therefore, during the simulation, we needed to extend some core simulation models into
the graphical user-interface supporting models.

GUI Cellular Tower depends on the SUMO graphical class GUIPointOInterest. It


contains the cartesian position coordinates and methods about visualizing the lo-
cation. Therefore, GUI cellular tower basically wraps itself around the graphical
user interface point of interest.

GUI Cellular Antenna has been implemented through wrapping itself around SUMO’s
GUIPolygon. SUMO’s GUI class, polygon gives us the visualizing possibilities
for drawing polygons in various styles.

3.3.4 Core mechanics

The main class of SUMO creates the central entity called MSNet and loads the pro-
gram specific configuration options. Setting up the simulation network and performer
requires simulation data and logical entities to manage them. The core SUMO simula-
tion depends on the work of the Chapter 3.3.5, where we describe the data import into
the traffic simulation program. Data import occurs during the road network loading,
which is done by NLBuilder class. For SUMO GUI ,the loading has been delegated to
the thread GUILoadThread, where NLBuilder exists and does its work.
During the importing and network building, various entities with a wide area of
functionality and commitment regions are created. There are control entities, which are
created by their respective builders. The control entities are, e.g., logical control objects
which administer simulation specific entities like road edges, lanes, junctions, detectors
for triggering, traffic lights, and cellular towers. There are two types of dynamic controls
for the routes and the mobility event data. The dynamic control means that they are
loader, actively loading new routes and events during the simulation according to the
step of the simulation. During loading the routes the vehicles are also created and added
to MSNet related vehicle control unit.
Before the simulation, after building the network, the graphical SUMO builds up the
visual presentation for the user while using given data and graphics library OpenGL. For

40
graphical version of SUMO GUINet is extending MSNet, adding attributes and methods
for the visual simulation.
The simulation cycle is happening in MSNet. The simulation step has been, by de-
fault, defined to be equal with one second in real life (there is a possibility to configure to
alter the value of SUMO time, but it has been noted that it might break some simulation
models and, therefore, not advisable).
During the simulation step, MSEventControl entities are used for various simula-
tion actions. In the Chapter 3.5.5, the essence of the mobility events simulation shall
be explained more thoroughly. The problem is that certain actions have presumptions
and dependencies on other resources, which need resources’ states updated before their
dependent events can work properly.

Figure 10. The simulation steps in the SUMO simulation cycle.

The simulation has been divided into four components (visualized in the Figure 10):

Begin of time step events is an event running cycle during which SUMO executes, for
example, traffic light specific events, trigger type events (lane speed, rerouting),
routing device, and pedestrian move events.

Intermediate actions are dependent on the previous event running cycle. During the
time between two event running cycles, SUMO is detecting collisions, checking
for the need of changing traffic lights, check if the edges are active (for simulation
purposes the user can turn off the lanes), planning vehicle movements, executing
vehicle movements, checking for the collisions again, run lane changes, check

41
if the previous movements created collisions, load new routes and mobile events
for the cellular simulation, check if there are people or containers waiting for the
transport, check and insert new vehicles to the network, insert new events, check
about inserted vehicles. During the execution of vehicles movement, there is a
case when the vehicle has arrived to the destination and, immediately afterwards,
the vehicle shall be removed from the simulation. This function has been altered
to have checks for existing mobility devices (GPS, mobile) to run their last events
directly and then invalidating those events to stop them from recurring. This entire
code alternation prevents losing data (otherwise, if the vehicle is removed before
the mobility events, no arriving of the last event) and null pointer exceptions.

End of time step events is an event running cycle which runs the events dependent on
the updated location of the routable elements (vehicles, pedestrians etc). There
are also step by step logger events, vehicular devices’ specific events (Bluetooth,
GPS, mobility management related state update) and vehicle flow calibrator events.

Mobility events is a newly added event running cycle entirely for the purpose of mobil-
ity event simulation. The events are created by the mobile devices from the data,
which has been attached to the vehicle after it has been loaded into the simulation
during the intermediate actions.

3.3.5 Data import, output and feature configurations

Running the simulation requires its configurations and data to be imported into the
program. For this, and altered multiple XML Schema Definition (XSD) files have been
created [31]. XSD files specify how to build Extensible Markup Language (XML)
files [6]. Describing how the importable elements should look like for being prepared
for the reading process into SUMO.
There has been defined cellular network related XSD file, defining the cellular tow-
ers and their antennas. Mobility event data from MESGEN has been also specified for
the import purpose. To include new configurations for SUMO, alterations in existing
XSD files were reqiored to enable data import through files.
During the building of the network to perform the simulation with NLBuilder so-
called handler elements are created for handling the XML data sources. For importing

42
the cellular physical network data and the mobility event data, base classes were created
under the utils for importing the data. The creation of them into the utils is because
the tools might need to import similar data and might have different data structures and
workflows, so they can make their own extensions for those purposes. Cellular and
mobility specific handlers are extending Xerces library class as a base class for XML
importing. For SUMO, microsim extended versions were implemented to suit for the
needs of the microscopic traffic simulation program.
Builder classes, which oversee the creation and populating of MSNet control entities,
extract data from the XML files and transform them into the models governed by control
entities. Our control classes covered command-line version and GUI version control
creations. GUI cellular builder, for example, had to deal graphics centric classes like
GUIPolygon and GUIPointOfInterest to be compatible with MSNet which operates with
ordinary Polygon and GUIPointOfInterest. Upcasting techniques were required.
Output after the simulation has been written in the dynamic manner without XSD
definition. The output has been built around C++ std::ostream (output stream) inside
of SUMO’s OutputDevice. OutputDevice has support for writing data inside of XML
elements.

3.3.6 Extent of the development

In the Figure 11 is described this thesis contribution in the overall scale compare to
SUMO. The white rectangles show the entities, which have not been altered during the
development, orange that this section has new features and the green blocks mean that
those are entirely new contributions.

3.4 MESGEN and HexagonGen tools


3.4.1 MESGEN

Mobility Event Simulation Generator is an independent program for preparing pre-


requisites for the cellular network behaviour simulation. Some mobility user types have
been defined to reflect real life. For example, the program describes casual user, idle
user, and frequently SMS sending and receiving teenager. Each type has been described
by three attributes: call frequency, SMS frequency, and web frequency. Each of the

43
Figure 11. The state of the SUMO application bundle after the contribution from the
author.

fields can be distributed into four categories, presenting the user’s activity: none, low,
medium and high.
The mobile events which will be generated are similar to mobile event data model’s
events, but with some reductions. The generated mobile event types are:

• SMS_REC,

• SMS_SEND,

• CALL_INIT,

• CALL_REC.

MESGEN takes as an input network file and route file to determine the vehicles
and generate cellular mobility events. While reading in the routes, the network data is
needed for validating purposes. The route file is important because of the vehicle id and

44
its departing time. The result file of MESGEN is an XML file which can be used as an
input file for SUMO to simulate cellular network behaviour.
The program generates logs in the following manner:

1. Pick a random character (mobility user type).

2. After picking a character, the event type control sum will be calculated from the
character mobility type properties (e.g., if high level calling property equals with
3, medium level SMS frequency equals with 2 and low level web communication
frequency equals with 1, the sum will be 6).

3. Events for the person in the vehicle will be generated through simple while loop.
The current time, which at the beginning is equal to the vehicle’s depart time, will
also be initialized. The will loop long until when the simulation end time minus
our defined current time variable condition equals less than 0.

4. In order to decide the event, an arbitrary number between 0 and previously calcu-
lated sum. The random number will then fall into the category of mobility user
type and then the occurring event has been decided.

5. The next step is to decide against the activity threshold if the chosen event happens
or not. If not, then the event does not occur and a random unit of time will be
added to the current time - narrowing the timeframe.

6. If the event happens, the length of the event will be calculated and the event will
be saved. Finally, the timeframe gets a random event buffer time added to the
current time - narrowing the timeframe between the simulation end time and the
vehicle entity’s current time.

Minimum length and maximum length of different events can be configured through
variables. It is also possible to define time buffers between occurring events. The values
of low and medium frequency events’ thresholds and random number seed can also be
set through the configuration file.

45
3.4.2 HexagonGen

HexagonGen is an accessory level script, written in Python, to generate 120 ◦ sectorized


cells for the cellular network behaviour simulation. It takes an input of latitude and lon-
gitude coordinate values, size (length of hexagon’s side), hexagon network width, and
height values. The given coordinate pair is the center of the drawn network (tower po-
sition). The result of the HexagonGen is an XML file which can be directly fed into
SUMO.

(a) Example of cellular network grid with lots of


(b) Example of macro cells with hexagon side
micro cells, where the hexagon side length has
as 20km.
been defined as 200 metres.

Figure 12. The village of Nõo with two types of cellular coverage with help of
HexagonGen.

3.5 Mobility behaviour and management


This subsection is about mobility devices, their soul-life, and how they have been
integrated with SUMO program.

3.5.1 Microscopic simulation devices

GPS device , also known as MSDevice_GPS in the code, is a test device, in the sense
that it was created with the objective to study SUMO simulation cycle for the
upcoming mobility device. Mapping out all the requirements for the develop-
ment while researching through SUMO simulation core while engineering the

46
GPS device. MSDevice_GPS is extending MSDevice, which is SUMO’s abstract
in-vehicle device. Therefore, it already has some basic interfaces to communicate
with the vehicle and abstract methods, which need to implemented to develop a
working device.
MSDevice_GPS has two subclasses called RouteGPSInfo and GPSSignalUpdate.
RouteGPSInfo is a data structure that records geographical coordinates and the
timestamp. GPSSignalUpdate represents GPS signal; it is an event by extend-
ing SUMO Command class (base SUMO microsim event class). During every
simulation step, if the vehicle has a device, the event gets ran after the vehicle’s
movement.

Mobile device , is known as MSDevice_Mobile in the code. It is a very enormous class


which simulates the mobile behaviour in multiple levels. It is emulating for the
simulation multiple functions, which in real life are actually done by other log-
ical units, or at least in cooperation with other entities (through chain of events,
clarifying states, allowing procedures, etc). New SUMO mobile device’s mobil-
ity management will be examined in the Chapter 3.5.2. Signal pinging during the
simulation pinging will be discussed in the Chapter 3.5.3. Primary state changing
procedures will be described in the Chapter 3.5.4. Events which direct the mobil-
ity behaviour will be elaborated in the Chapter 3.5.5. Finally, the result log object
will be analyzed and discussed in the Chapter 3.5.6.
Mobile device is collecting its activity logs (the primary objective of the thesis -
generation of logs), current signal strength, and boolean primitive flags if the mo-
bility event is currently occurring and the connection with the transmitter has been
interrupted. It also contains references to the mobility management entity and for
the active mobility event (if the vehicle, and therefore the device, gets removed
from the simulation, then through the both way reference we can invalidate an
event in the event cycle).
In the Figure 13 has been described the mobile device in the SUMO program in
the form of class diagram. It has defined the relationships with other mobility
classes.

47
Figure 13. The class diagram of MSDevice_Mobile and its sub-entities.

3.5.2 Mobility management

In the Figure 14, we can see mobility management’s reduced version. Layer 3’s three
sublayers, along with their actions, have been described in the context of this thesis.
In MSDevice_Mobile there is a subclass called MobilityManagement which gov-
erns mobile device’s state during the simulation. The mobile mobility state is managed
through the end of simulation step event cycle. During the creation of the vehicle into
the simulation, it shall be equipped with a wide variety of devices, according to the
configuration setup. If the vehicle has been equipped with a mobile device, then during

48
Figure 14. Reduced mobility management state machine. [54] [53]

the device creation, it will be created with mobility management entity. Mobility man-
agement entity has MobilityManagementUpdate, an entity which is extending SUMO
Command base abstract event class. The mobility management update event entity will
be added to a particular MSEventControl entity, which handles the end of the simulation
step events’ execution. Therefore, the mobility management is updated after the vehicle
movement in the end of the simulation step. Well almost at the end of the simulation
step, after mobility management update there will be the execution of mobility events
which are dependent on the latest state of mobility management.
The mobility management has the following states:

• MM_NULL,

• MM_IDLE_NORMAL_SERVICE,

• MM_IDLE_SEARCH_FOR_PLMN,

• MM_IDLE_NO_CELL_FOUND,

49
Figure 15. The simple work flow of the mobility management.

• MM_CONNECTION_ACTIVE,

• MM_WAIT_FOR_RR_ACTIVE,

• MM_WAIT_FOR_RR_CONNECTION,

• MM_LOCATION_UPDATING_INITIATED,

• MM_LOCATION_UPDATE_REJECTED,

• MM_WAIT_FOR_NETWORK_COMMAND,

• MM_PROCESS_SERVICE_PROMPT,

50
• MM_WAIT_FOR_OUTGOING_MM_CONNECTION.

Mobility management has a reference to the mobile device to call out mobility re-
lated procedures from the lower layers. Then there are mobility management state re-
lated variables; the T312 timer is for reminding a mobile device to make periodic lo-
cation updates after every user defined steps (by default 10 steps). Some state related
extra scenarios, define whether the mobile has been turned on, the SIM card inserted,
the IMSI attached, and define whether the mobile has PLMN related information.
In the Figure 15 has been the mobile mobility life cycle described visually for better
understanding.
At the start of the mobility management update, the creation of the current time
step’s mobility events for the simulation for this specific device takes place. The mobil-
ity events are explained in the Chapter 3.5.5.
During the mobility management update event, there will be called out management
core’s update function which increments instantly periodic location update and then,
according to the previous state, goes through state machine to determine the next step:

If the mobility management state has been classified as NULL, which actually should
be impossible, but nonetheless the core update will change mobile to the state as
it was just physically turned on.

MM_IDLE_NORMAL_SERVICE starts mobile cell selection method on the device


level. Described in the Chapter 3.5.4.

MM_IDLE_SEARCH_FOR_PLMN starts mobile cell selection method, but since the


device has never been connected to the network, it also waits for the surrounding
PLMN information.

MM_IDLE_NO_CELL_FOUND start mobile cell selection, since there is currently no


connectivity.

MM_CONNECTION_ACTIVE starts the cell selection process for the handover pur-
poses, if needed. Described in the Chapter 3.5.4.

As the reader could detect, in the state machine description there are less states
described than in the previously outlined list. That is because many of the states are

51
actually intermediate states, and during the simulation, there are no meaningful purpose
for them.
After cell selection, the workflow will move from the device back into the mobil-
ity management, where the state of MM will be updated and location update initiated.
Location update is described in the Chapter 3.5.4.
If the state machine has done its work, then there is a successive check to determine
if the periodic update is needed and, if it is, then it will be executed. The rule with
the periodic location update is that if a mobile has not had any location updates during
some certain amount of simulation steps, then the mobility management has to step in
and remind the transmitter that it still exists and has not been lost during the commute.

3.5.3 Wave propagation from the transmitter

During the cell selection, the mobile device has multiple challenges. It has to deter-
mine the nearest polygons, determine in which one of them it resides, and determine the
signal strength from the base station’s transmitter.
The signal strength has been envisioned by the free space wave propagation model.
It indicates the power received by the antenna under ideal conditions. The free space
model predicts the power’s decay to be the negative square root of the distance. In our
case, we have not included the system loss factor. This could be included in the future
work and allow real life like signaling in SUMO.
The signal strength is calculated with a formula described in the Definition 1. The
distance in the formula has been calculated between the tower and the vehicle and for
that, Haversine distance, defined in the Definition 2, was used. Haversine distance is
used for distance calculation on the sphere (in our scope planet Earth) using two points
longitudes and latitudes.

Definition 1. Friis free space equation:

Pt Gt Gr λ2
Pr (d) = ,
(4π)2 d2 L

where Pr (d) is the received power, Pt is the transmitter output power, Gt is transmitter
gain Gr is antenna gain, d is the distance between the MS and tower, and L is the system
loss factor. [70].

52
Definition 2. Haversine distance:
s    !
φ2 − φ 1 λ2 − λ1
d = 2r arcsin sin2 + cos(φ1 ) cos(φ2 ) sin2 ,
2 2

where φ is the latitude, λ the longitude and r is the Earth’s radius in meters (6 371 000
meters).

During the signal strength calculation, we use the configuration supplied tower trans-
mitter output strength (dBm) and the transmitter’s frequency. From the frequency, we
calculate the wavelength. Wavelength equals the speed of light divided by the transmit-
ter emitted frequency.

3.5.4 Location update, Cell selection and reselection, Handover

The procedures are key component of keeping track of the mobility management
level state, this can be seen also in the Figure 15.

Cell selection is a mobile device level method for measuring the current signal strength
value, determining the current antenna to which the mobile device is connected,
and assessing the need for the handover procedure.

Figure 16. The example of R-tree after all the shapes have been inserted into the tree.

53
During the cell selection, the first step is to find surrounding cellular network
cells. Those cells (polygons) have been structured into the R-tree entity during
the creation of the mobility tower control unit, where they were added for a faster
search. Otherwise, there would have been brute force like approach, where we
would had to run through every single cell in the program with expensive point in
polygon algorithm. As visible in the Figure 16, R-tree is a tree data structure to
handle spatial data efficiently by indexing shapes for the future access. [57]
After searching the nearest polygons to the device, we check through the vector
of cells and determine whether the vehicle is in the cell or not. The vehicle might
be in multiple cells; therefore, we are calculating the best signal strength and
accordingly, we pick the transmitter and cell to camp on. The check if the vehicle
is in the polygon or not has been dealt with the winding number algorithm. In
essence, the winding number algorithm is an algorithm which counts the number
of times the polygon winds around the point of interest. If the point is not inside
of the polygon, then the resulting winding number is 0. [61]
During the cell selection, the registration to the tower or the deregistration from
the tower will be determined. This is basically emulating the registering the mo-
bile device’s location area into the visitor location register.
The final step is the status update of the mobility management.

Location update is the mobility management level procedure for updating the device’s
location in the network. Location update is called out during the status update, the
periodic update procedure, and the handover. Location update resets the periodic
update timer (sets the timer back to 0) and updates the mobility management
state. To avoid spaghetti code, a simplification has been made, so that during the
cell selection, the location of the device has been already updated in the network
(added to the certain tower’s private vector of connected mobile entities).

Handover occurs when a better quality cell has been found during a mobility event.
The flags have been set and during the status update, there is a state machine level
check about MM_CONNECTION_ACTIVE in regard to newly found high quality
signal strength cell. If the statement is true, then there the handover process will
occur. The handover in our reduced scope means calling out the location update.

54
3.5.5 Mobility events and the simulation cycle

We have covered the mobility management tied event object called MobilityMan-
agementUpdate. In this section, we will discuss about the mobile device related event.
MobilityEvent extends the SUMO Command abstract event base class. The mobility
events have their own MSEventControl container-event queue and the events are being
executed at the very end of the simulation step, after the mobility management update
events.
Each mobility event has a type, which are the same types declared in the enum
of MobilityEventType in the class of MSMobilityEventData. The event has the mobile
device reference to let the mobile know about the radio link failure and trigger the
log creation about the occurred event. Other attributes of the mobility event are the
event’s start time and end time. Calls and the web communication events depend on the
timeframe variables.

3.5.6 Mobility CDR-like logs

Call detail record is the information about incoming and outgoing mobile activities,
e.g., calls or SMS messages. The data in the CDR is about the event originated and the
terminated parties (both sides’ phone numbers), time of connection through the starting
time and the call duration, call event type, unique generated id for the record, etc. [58]
MobilityLog class is a data structure that, to some extent, emulates the essence of
CDR. The attributes of the mobile logs are the following (also visible in the Figure 13):

• timestamp,

• mobility event type,

• IMSI code, in our case usually known as the vehicle ID,

• Common Gateway Identifier (CGI) is the value indicating to the transmitter id


(could be interpreted also as a cell ID)

• signal strength during the event,

• latitude coordinate,

55
• longitude coordinate,

• vehicle speed during the event (in m/s).

3.6 Conclusion
This chapter explained the practical part of the work with extensive overview about
SUMO core mechanics. It was described how the new entities and their functions were
integrated into the existing system. Different tools were presented which help to gener-
ate data for the cellular network behaviour simulation. Those tools created XML files
which needed to be imported into the SUMO program. The entire work flow about
that process was thoroughly presented, including the models, which were used for the
simulation.
We took a deep look into the mobile device and its mobility management, as well
as how it was installed and fit according to the needs and the scope of this thesis. The
result of the cellular network behaviour simulation was the created mobility logs files:
one of them was the practice and learning task (GPS logs) and the other was the primary
task, which is required for vast area of research as an input. We will take deeper look
into the logs and the other tangible result in the Chapter 4.

56
4 Results and Analysis
We will now examine the results of the practical task and give an overview of what
has been accomplished. The mobility logs will be presented, along with a tutorial how
to produce them. The second part of the chapter takes a glimpse into the performance
benchmarking.

4.1 The mobility logs


4.1.1 GPS logs

...
< v e h i c l e id ="0" d e p a r t =" 0.00 ">
<param key = " h a s . g p s . d e v i c e " v a l u e = " t r u e " / >
<param key = " h a s . m o b i l e . d e v i c e " v a l u e = " t r u e " / >
<route . . . / >
</ vehicle>
...

Listing 1. Route file, an example how to add devices to the desired vehicle.

MSDevice_GPS related logs can be created in two ways. One way is to configure
the main SUMO configuration file in a manner that all the vehicles have a device, as
seen in the Listing 2. The output file must be defined into the SUMO configuration file.
Otherwise, the output will not be delivered because of the lacking destination. Another
possibility is to configure that specific vehicle has a device to the SUMO route file,
which has been also shown in the Listing 1.
In the Listing 3 GPS signal events which have been described, in the Chapter 3.5.1
are exhibited. RouteGPSLog, as we can see from the Listing, it contains of the times-
tamp, longitude, and latitude.
<input>
...
<mes f i l e s v a l u e = " m e s _ o u t p u t . xml " / >
< c e l l u l a r f i l e s v a l u e = " c e l l u l a r _ o u t p u t . xml " / >
</ input>
...
<output>

57
<gps o u t p u t v a l u e = " g p s _ o u t p u t . xml " / >
< m o b i l e o u t p u t v a l u e = " m o b i l e _ o u t p u t . xml " / >
</ output>
...
<positioning>
< d e v i c e . gps . enable a l l v a l u e =" f a l s e " / >
</ positioning>
...
<mobility>
< device . m o b i l i t y . enable a l l value =" t r u e " / >
</ mobility>
...

Listing 2. Sumo configuration snippet about activating the GPS devices in the all
vehicles

<gps o u t p u t >
< v e h i c l e id ="1">
<gps d e v i c e i d = " g p s _ 1 " >
< s i g n a l time=" 1.00 ">
< c o o r d i n a t e s l a t i t u d e =" 58.29626476 " l o n g i t u d e =" 26.44804043 " / >
</ signal>
< s i g n a l time=" 3.00 ">
< c o o r d i n a t e s l a t i t u d e =" 58.29628805 " l o n g i t u d e =" 26.44810251 " / >
</ signal>
< s i g n a l time=" 5.00 ">
< c o o r d i n a t e s l a t i t u d e =" 58.29622327 " l o n g i t u d e =" 26.44831755 " / >
</ signal>
...
</ vehicle>
...
< / gps o u t p u t >

Listing 3. GPS coordinates (longitude and latitude) with a timestamp

4.1.2 Cellular network simulation logs

The primary objective, MSDevice_Mobile class generated logs can be created when
the configuration values have been set true for the mobile device and the mobile data

58
events have been defined for the same device in the mes-file. If the user wants posi-
tive results, then the cellular network should be defined. Otherwise there would be no
connectivity for the devices and the events cannot occur.
The configuration of the mobile devices is similar to the GPS devices. One must
define the existing mobile device into the route file, as shown in the Listing 1. Otherwise
you can set all the vehicles to carry a mobile device and do as show in the Listing 2.
Also, the user must define the output file in the config file.
In the Listing 4 can be seen a snippet about the cellular network behaviour simula-
tion from the resulting XML output file. The log entity has been defined in the Figure 13
as the class MobilityLog and in the Chapter 3.5.6.
<mobile o u t p u t >
< v e h i c l e id ="1">
< e v e n t t i m e s t a m p = " 9 2 . 0 0 " e v e n t T y p e = "KEEP_ALIVE" i m s i = " 1 " c g i =
" t 4 3 8 1 " s i g n a l S t r e g n t h =" 2 1 . 0 1 " l a t i t u d e =" 58.29 "
l o n g i t u d e =" 26.46292491 " speed=" 13.68143750 " / >
< e v e n t t i m e s t a m p = " 9 4 . 0 0 " e v e n t T y p e = "SMS_REC" i m s i = " 1 " c g i = " t
4 3 8 1 " s i g n a l S t r e g n t h =" 21.70533381 " l a t i t u d e ="
58.29089506 " l o n g i t u d e =" 26.46336419 " speed=" 12.64468285 " / >
</ vehicle>
< v e h i c l e i d = " 27 " >
< e v e n t t i m e s t a m p = " 2 7 . 0 0 " e v e n t T y p e = " CALL_INIT " i m s i = " 27 " c g i =
" t 8 5 2 1 " s i g n a l S t r e g n t h =" 21.11894073 " l a t i t u d e ="
58.25022813 " l o n g i t u d e =" 26.43376626 " speed=" 0.00000000 " / >
< e v e n t t i m e s t a m p = " 2 8 . 0 0 " e v e n t T y p e = " CALL_INIT " i m s i = " 27 " c g i =
" t 8 5 2 1 " s i g n a l S t r e g n t h =" 21.21370266 " l a t i t u d e ="
58.25021502 " l o n g i t u d e =" 26.43379995 " speed=" 2.45816080 " / >
< e v e n t t i m e s t a m p = " 2 9 . 0 0 " e v e n t T y p e = " CALL_INIT " i m s i = " 27 " c g i =
" t 8 5 3 1 " s i g n a l S t r e g n t h =" 20.32984836 " l a t i t u d e ="
58.25019017 " l o n g i t u d e =" 26.43386379 " speed=" 4.65720239 " / >
...
</ vehicle>
...
< / mobile o u t p u t >

Listing 4. CDR-like mobility logs

59
4.1.3 MESGEN and HexagonGen

To generate mobility event data for the SUMO cellular network behaviour simula-
tion, then there is needed a configuration file. In the Listing 5, there are visible two
input files: network file and route file. There has been defined output file, start of the
simulation and the end, last parameter enables random number generator in the SUMO
tool.
<? xml v e r s i o n = " 1 . 0 " e n c o d i n g = "UTF 8 " ?>
<configuration >
<input >
< n e t f i l e v a l u e = " nyghen . n e t . xml " / >
< r o u t e f i l e s v a l u e = " r o u t e r 5 0 0 0 . r o u . xml " / >
</ i n p u t >
<output >
< o u t p u t f i l e v a l u e = " m e s _ o u t p u t 5 0 0 0 . xml " / >
</ o u t p u t >
<time >
<begin value ="0" / >
< end v a l u e = " 1000 " / >
</ t i m e >
<random_number >
<random > t r u e < / random >
</ random_number >
</ c o n f i g u r a t i o n >

Listing 5. MESGEN configuration file example

To run the MESGEN there is needed to run command line tool called mesgenerate.
The result of the generation is visible in the Listing 6. Vehicle with it list of mobile
events.
<vehicles>
< v e h i c l e id ="0" d e p a r t =" 0.00 ">
<mEvents >
<mEvent begin=" 114.00 " end = " 1 3 7 . 0 0 " t y p e = "CALL_REC" / >
<mEvent begin=" 363.00 " end = " 4 5 0 . 0 0 " t y p e = "WEB_COMM" / >
<mEvent begin=" 490.00 " end = " 5 1 6 . 0 0 " t y p e = " CALL_INIT " / >
<mEvent begin=" 627.00 " end = " 6 6 9 . 0 0 " t y p e = " CALL_INIT " / >
<mEvent begin=" 731.00 " end = " 7 4 2 . 0 0 " t y p e = "CALL_REC" / >
<mEvent begin=" 829.00 " end = " 8 5 6 . 0 0 " t y p e = " CALL_INIT " / >

60
< / mEvents >
</ vehicle>
...
</ vehicles>

Listing 6. The appearance of the MESGEN output file

To generate the cellular network file, there is needed to run HexagonGen’ Python
file, visible in the Listing 7. The arguments are latitude, longitude, length of the hexagon
side and how big the cellular network grid should be with height and width.
hexagonGen / main . py 5 8 . 2 8 0 6 2 6 . 5 1 8 7 0 . 0 2 10 10

Listing 7. Command-line invoke of HexagonGen Python script

In the Listing 8, there is visible the result with one cell tower and three coverage
cells. Tower contains of cell coverages and point of interest, as the location of the
tower.
<? xml v e r s i o n = " 1 . 0 " e n c o d i n g = "UTF 8 " ? >
<cellularTowers>
<cellularTower>
<coverage id =" t 0 0 " shape=" 26.3697436305 ,58.3356
26.3680115797 ,58.3346 26.3662795289 ,58.3356
26.3662795289 ,58.3376 26.3680115797 ,58.3386
26.3697436305 ,58.3376 26.3697436305 ,58.3356 " c o l o r ="
1 3 7 , 1 3 7 , 1 3 7 " geo = " t r u e " l a y e r = " 0 " >
<param key = " s i g n a l S t r e n g t h " v a l u e = " 62 " >< / param >
<param key = " f r e q u e n c y " v a l u e = " 8 0 0 . 0 " >< / param >
< / coverage>
<coverage id =" t 0 1 " shape=" 26.3714756814 ,58.3326
26.3697436305 ,58.3316 26.3680115797 ,58.3326
26.3680115797 ,58.3346 26.3697436305 ,58.3356
26.3714756814 ,58.3346 26.3714756814 ,58.3326 " c o l o r ="
1 3 7 , 1 3 7 , 1 3 7 " geo = " t r u e " l a y e r = " 0 " >
<param key = " s i g n a l S t r e n g t h " v a l u e = " 62 " >< / param >
<param key = " f r e q u e n c y " v a l u e = " 8 0 0 . 0 " >< / param >
< / coverage>
<coverage id =" t 0 2 " shape=" 26.3680115797 ,58.3326
26.3662795289 ,58.3316 26.3645474781 ,58.3326
26.3645474781 ,58.3346 26.3662795289 ,58.3356

61
26.3680115797 ,58.3346 26.3680115797 ,58.3326 " c o l o r ="
1 3 7 , 1 3 7 , 1 3 7 " geo = " t r u e " l a y e r = " 0 " >
<param key = " s i g n a l S t r e n g t h " v a l u e = " 62 " >< / param >
<param key = " f r e q u e n c y " v a l u e = " 8 0 0 . 0 " >< / param >
< / coverage>
<poi id =" t 0 " l a t =" 58.3346 " lon =" 26.3680115797 " type ="
c e l l u l a r _ t o w e r " c o l o r = " 1 , 1 , 1 " l a y e r = " 0 " w i d t h = " 12 " h e i g h t =
" 16 " i m g F i l e = " t o w e r . bmp " >< / p o i >
</ cellularTower>
...
</ cellularTowers>

Listing 8. The appearance of the cellular file

4.2 Graphical upgrade


This thesis produced two new visual entities: cellular tower and cellular network
cell. In the Figure 17. The user-interface has been enriched with buttons for the cellular
network manipulation purposes. Those buttons are also visible in the Figure 17.
Drawing the towers depend on the SUMO class PointOfInterest and the hexagons
are drawn through Polygon class.

4.3 The benchmarking


There was created a little benchmark to see how much influence the cellular network
behaviour simulation has on the SUMO program. In the Figure 18, we can see that the
mobility data simulation with 1200 towers, 5997 cells and 200 vehicles resulted with
4.36s simulation. Meanwhile, when we removed the cellular towers and the respective
cells then the simulation time fell from 4.36s to 3.19s, as we can see on the Figure 19.

62
Figure 17. The SUMO simulation with the cellular tower with its cells.

5 Conclusion and Future Perspectives


This chapter concludes the thesis and presents the future perspectives in the research
field, speculating about the possibilities and what extra work needs to be done, as well
as the potential benefits from those upgrades.

5.1 Conclusion
The thesis aims at the possibility of simulating the mobile subscriber behaviour from
the cellular network point of view. There are vast fields of research which depend on the
mobile communication data. The work is often disturbed by the mobile operators’ lack
of interest and privacy of data constraints imposed by the law. The thesis investigates
the SUMO traffic simulator and modern mobile technologies to try to blend these two
components together to produce cellular network behaviour logs. The traffic simulator
program offered the vehicular mobility and the cellular network behaviour was devel-

63
Figure 18. The CPU benchmark with the cellular network simulation.

oped during this thesis. The mobile technology and the concepts were derived from the
3GPP/ETSI specifications and industry recommended books.
The cellular network architecture and the models were produced from the documents
and simplified for the implementing, reducing the scope and unwanted functionality
which would hinder the development and shift the thesis from the main objectives. The
mobile radio network was designed and developed, resulting in the mobile device, cel-
lular network physical infrastructure, and their resulting network cells with wave signal
propagation.
Finally, the graphical visualizations were made to understand the representation of
the cellular network in the SUMO graphical version. Additionally, the generation of the
mobility logs (mobile CDR-like and GPS) with storing was implemented.

5.2 Future work


The future perspective of this thesis work is to develop and visualize the wave prop-
agation. There are many models that must be added into the created simulation such as:
close electromagnetic properties to be emulated, for example, the path loss, reflection
(diffraction, scattering), to have a proper real-life representation of the Received Signal
Strength Indicator (RSSI) received at the mobile device in the simulation. The integra-
tion of wave propagation model, along with other cellular network behaviour models in
the created simulation, will add more scientific value for writing a journal publication.

64
Figure 19. The CPU benchmark without the cellular network simulation.

One of the most important works which must be done is the integration of this thesis
with other science projects. There is huge potential if we manage to integrate this thesis
with projects where human behaviour models are used. This, would help simulate the
vehicle movement and put the humans in vehicle communicate as they are in real life.
The human behaviour modelling in the communication could be one of the main
tasks for the mobile event data generator, which currently relies on arbitrary events. An-
other accessory created during the development of cellular network simulation, HexagonGen,
could be developed into a Bachelor thesis. During such a thesis, the objective would be
to investigate and design a cellular network tool. Drawing different types of cellular
networks with high functionality, e.g., angling hexagons and define sub-networks for
more complicated scenarios. One of the main criteria would be the export of the data
which should be compatible with SUMO and maybe with other tools in the market what
could benefit from it.
On the SUMO level, there is a possibility to simulate the entire network’s job. This
requires analysis, design, and development of the fixed network - network switching
subsystem. This could enable the base station level, MSC level, or even some other
components log generation.
One of the bigger scale future works involves third-party developed tools. There are
network technologies which simulate 2G, 3G, 4G networks. They could also be inte-
grated with SUMO. In addition, their cellular network transmitter waves could be prop-

65
agated with open source gprMax [12], which specialises in the electromagnetic wave
propagation. One of the network simulators is known as OsmocomBB [19], which is an
open source GSM baseband software implementation. OpenBTS [14] is another open
source project developing a new era of telecommunication protocols to substitute legacy
proprietary code. This means that this project code could be fitted for the simulation
software as well. An open source project also exists which is simulating 4G technology.
OpenLTE [17] depends heavily on the 3GPP LTE specifications. Such projects enable
researchers to take the cellular network simulation to next level, but the feasibility to
simulate entire systems depends on the requirements and expected output.
Finally, there are many possible opportunities that can be integrated to the created
system and how it could serve the scientific community.

66
References
[1] 3GPP. http://www.3gpp.org/. Accessed: 09.05.2017.

[2] DLR - DLR Portal. http://www.dlr.de/dlr//en/desktopdefault.


aspx/tabid-10002/. Accessed: 10.05.2017.

[3] DLR - Institute of Transportation Systems - SUMO User Conference


2017 - "Towards Simulation for Autonomous Mobility" (May 8-10,
2017). http://www.dlr.de/ts/en/desktopdefault.aspx/
tabid-3930/20125_read-47741/. Accessed: 10.05.2017.

[4] DLR - Institute of Transportation Systems - SUMO – Simulation of Ur-


ban MObility. http://www.dlr.de/ts/en/desktopdefault.aspx/
tabid-9883/16931_read-41000/. Accessed: 09.05.2017.

[5] Eclipse CDT - Eclipse CDT (C/C++ Development Tooling). http://www.


eclipse.org/cdt/. Accessed: 15.05.2017.

[6] Extensible Markup Language (XML). https://www.w3.org/XML/. Ac-


cessed: 15.05.2017.

[7] FCC ADOPTS RULES TO IMPLEMENT ENHANCED 911 FOR


WIRELESS SERVICES. https://www.fcc.gov/document/
fcc-adopts-rules-implement-enhanced-911-wireless-services.
Accessed: 08.05.2017.

[8] fox-toolkit.org. http://www.fox-toolkit.org/. Accessed: 11.05.2017.

[9] GCC 4.8 Release Series - GNU Project - Free Software Foundation (FSF).
https://gcc.gnu.org/gcc-4.8/. Accessed: 15.05.2017.

[10] GDAL:GDAL - Geospatial Data Abstraction Library. http://www.gdal.


org/. Accessed: 11.05.2017.

[11] Git. https://git-scm.com/. Accessed: 15.05.2017.

[12] gprMax, GitHub. https://github.com/gprMax. Accessed: 16.05.2017.

67
[13] MATSim | Multi-Agent Transport Simulation. http://www.matsim.org/.
Accessed: 11.05.2017.

[14] OpenBTS | Open Source Cellular Infrastructure. http://openbts.org/. Ac-


cessed: 16.05.2017.

[15] OpenDRIVE - Home. http://www.opendrive.org/. Accessed:


11.05.2017.

[16] OpenGL - The Industry Standard for High Performance Graphics. https://
www.opengl.org/. Accessed: 11.05.2017.

[17] openLTE download | SourceForge.net. https://sourceforge.net/


projects/openlte/. Accessed: 16.05.2017.

[18] OpenStreetMap. https://www.openstreetmap.org/. Accessed:


11.05.2017.

[19] Overview - OsmocomBB - Open Source Mobile Communications. http://


osmocom.org/projects/baseband. Accessed: 16.05.2017.

[20] proj.4 — proj.4 4.9.3 documentation. http://proj4.org/. Accessed:


11.05.2017.

[21] PTV Vissim. http://www.ptvgroup.com/en/solutions/


products/ptv-vissim/. Accessed: 11.05.2017.

[22] PTV Visum. http://vision-traffic.ptvgroup.com/en-us/


products/ptv-visum/. Accessed: 11.05.2017.

[23] PyCharm. https://www.jetbrains.com/pycharm/. Accessed:


15.05.2017.

[24] Python 2.7.0 Release | Python.org). https://www.python.org/


download/releases/2.7/. Accessed: 15.05.2017.

[25] Simulation of Urban MObility - Wiki. http://sumo.dlr.de/wiki. Ac-


cessed: 09.05.2017.

68
[26] Simulation/Meso - Sumo. http://sumo.dlr.de/wiki/Simulation/
Meso. Accessed: 15.05.2017.

[27] SUMO User Documentation. http://sumo.dlr.de/wiki/SUMO_User_


Documentation. Accessed: 09.05.2017.

[28] TextTest.org. http://texttest.sourceforge.net/. Accessed:


11.05.2017.

[29] The Apache XercesTM Project - xerces.apache.org. https://xerces.


apache.org/xerces-c/. Accessed: 11.05.2017.

[30] The GNU General Public License v3.0 - GNU Project - Free Software Foundation.
https://www.gnu.org/licenses/gpl.html. Accessed: 10.05.2017.

[31] W3C XML Schema. https://www.w3.org/XML/Schema. Accessed:


15.05.2017.

[32] Esteban Zimanyi Chiara Renso, Stefano Spaccapietra. Mobility Data: Modeling,
Management and Understanding. Cambridge, 2013.

[33] Craig Scott Christopher Drane, Malcolm Macnaughtan. Positioning gsm tele-
phones. IEEE Communications Magazine, 36(4):46–59, April 1998.

[34] M. Decina and E. Scace. Ccitt recommendations on the isdn: A review. IEEE
Journal on Selected Areas in Communications, 4(3):320–325, May 1986.

[35] European Telecommunications Standards Institute, 650 Route des Lucioles -


Sophia Antipolis - Valbonne - FRANCE. GSM 03.02: Digital cellular telecom-
munications system (Phase 2+); Network architecture, 5.1.0 edition, May 1996.

[36] European Telecommunications Standards Institute, 650 Route des Lucioles -


Sophia Antipolis - Valbonne - FRANCE. GSM 03.04: Digital cellular telecom-
munications system; Signalling requirements relating to routeing of calls to mobile
subscribers, 5.0.0 edition, November 1996.

69
[37] European Telecommunications Standards Institute, 650 Route des Lucioles -
Sophia Antipolis - Valbonne - FRANCE. GSM 03.12: Digital cellular telecom-
munications system; Location registration procedures, 5.0.0 edition, November
1996.

[38] European Telecommunications Standards Institute, 650 Route des Lucioles -


Sophia Antipolis - Valbonne - FRANCE. GSM 08.02: Digital cellular telecommu-
nications system; Base Station System - Mobile-services Switching Centre (BSS -
MSC) interface; Interface principles, 5.0.0 edition, December 1996.

[39] European Telecommunications Standards Institute, 650 Route des Lucioles -


Sophia Antipolis - Valbonne - FRANCE. GSM 03.08: Digital cellular telecommu-
nications system (Phase 2+); Organization of subscriber data, 5.1.0 edition, April
1997.

[40] European Telecommunications Standards Institute, 650 Route des Lucioles -


Sophia Antipolis - Valbonne - FRANCE. GSM 03.09: Digital cellular telecommu-
nications system (Phase 2+); Handover procedures, 5.1.0 edition, August 1997.

[41] European Telecommunications Standards Institute, 650 Route des Lucioles -


Sophia Antipolis - Valbonne - FRANCE. GPRS 01.60: Digital cellular telecom-
munications system (Phase 2+); General Packet Radio Service (GPRS) Require-
ments specification of GPRS, 6.0.0 edition, April 1998.

[42] European Telecommunications Standards Institute, 650 Route des Lucioles -


Sophia Antipolis - Valbonne - FRANCE. GSM 03.20: 3rd Generation Partnership
Project; Digital cellular telecommunications system (Phase 2+); Security related
network functions (Release 1999), 8.1.0 edition, October 2000.

[43] European Telecommunications Standards Institute, 650 Route des Lucioles -


Sophia Antipolis - Valbonne - FRANCE. 3GPP 25.401: Universal Mobile
Telecommunications System (UMTS); UTRAN Overall Description, 4.2.0 edition,
September 2001.

[44] European Telecommunications Standards Institute, 650 Route des Lucioles -


Sophia Antipolis - Valbonne - FRANCE. 3GPP 04.04:3rd Generation Partner-

70
ship Project; Technical Specification Group GSM/EDGE Radio Access Network;
Layer 1; General requirements, 8.1.2 edition, May 2002.

[45] European Telecommunications Standards Institute, 650 Route des Lucioles -


Sophia Antipolis - Valbonne - FRANCE. 3GPP 25.304: Universal Mobile
Telecommunications System (UMTS); User Equipment (UE) procedures in idle
mode and procedures for cell reselection in connected mode, 8.7.0 edition,
September 2009.

[46] European Telecommunications Standards Institute, 650 Route des Lucioles -


Sophia Antipolis - Valbonne - FRANCE. 3GPP 43.051: Digital cellular telecom-
munications system (Phase 2+); GSM/EDGE Radio Access Network (GERAN)
overall description; Stage 2, 8.0.0 edition, January 2009.

[47] European Telecommunications Standards Institute, 650 Route des Lucioles -


Sophia Antipolis - Valbonne - FRANCE. 3GPP 44.069: Digital cellular telecom-
munications system (Phase 2+); Broadcast Call Control (BCC) protocol (3GPP
TS 44.069 version 9.0.0 Release 9), 9.0.0 edition, January 2010.

[48] European Telecommunications Standards Institute, 650 Route des Lucioles -


Sophia Antipolis - Valbonne - FRANCE. 3GPP 23.009: Digital cellular telecom-
munications system (Phase 2+); Universal Mobile Telecommunications System
(UMTS); Handover procedures, 10.0.0 edition, April 2011.

[49] European Telecommunications Standards Institute, 650 Route des Lucioles -


Sophia Antipolis - Valbonne - FRANCE. 3GPP 24.007: Digital cellular telecom-
munications system (Phase 2+); Universal Mobile Telecommunications System
(UMTS); LTE; Mobile radio interface signalling layer 3; General Aspects, 10.0.0
edition, March 2011.

[50] European Telecommunications Standards Institute, 650 Route des Lucioles -


Sophia Antipolis - Valbonne - FRANCE. 3GPP 44.018: Digital cellular telecom-
munications system (Phase 2+); Mobile radio interface layer 3 specification; Ra-
dio Resource Control (RRC) protocol, 10.3.0 edition, June 2011.

[51] European Telecommunications Standards Institute, 650 Route des Lucioles -


Sophia Antipolis - Valbonne - FRANCE. 3rd Generation Partnership Project;

71
Technical Specification Group GSM/EDGE Radio Access Network; Mobile Station
- Base Station System (MS - BSS) interface; Data Link (DL) layer specification,
8.4.0 edition, January 2012.

[52] European Telecommunications Standards Institute, 650 Route des Lucioles -


Sophia Antipolis - Valbonne - FRANCE. 3GPP 23.002: Digital cellular telecom-
munications system (Phase 2+); Universal Mobile Telecommunications System
(UMTS); LTE; Network architecture, 12.5.0 edition, October 2014.

[53] European Telecommunications Standards Institute, 650 Route des Lucioles -


Sophia Antipolis - Valbonne - FRANCE. 3GPP 44.018: Digital cellular telecom-
munications system (Phase 2+); Mobile radio interface layer 3 specification; Ra-
dio Resource Control (RRC) protocol; Iu mode, 13.0.0 edition, January 2016.

[54] European Telecommunications Standards Institute, 650 Route des Lucioles -


Sophia Antipolis - Valbonne - FRANCE. 3GPP 24.008: Digital cellular telecom-
munications system (Phase 2+) (GSM); Universal Mobile Telecommunications
System (UMTS); LTE; Mobile radio interface Layer 3 specification; Core network
protocols; Stage 3, 13.8.0 edition, January 2017.

[55] Flarion. OFDM for Mobile Data Communications. https://www.scribd.


com/document/7352160/OFDM-Mobile-Data-Communications,
March 2003.

[56] Ki Won Sung Slimane Ben Slimane Guowang Miao, Jens Zander. Fundamentals
of Mobile Data Networks. Cambridge, March 2016.

[57] Antonin Guttman. R-trees: a dynamic index structure for spatial searching. SIG-
MOD ’84 Proceedings of the 1984 ACM SIGMOD international conference on
Management of data, pages 47–57, 1984.

[58] Ray Horak. Telecommunications and Data Communications Handbook, 2nd Edi-
tion. Wiley, August 2008.

[59] Jean Bolot Hui Zang, Francois Baccelli. Bayesian inference for localization in
cellular networks. IEEE INFOCOM’10 Proceedings of the 29th conference on
Information communications, pages 1963–1971, 2010.

72
[60] Simone Frattasi João Figueiras. Mobile Positioning and Tracking: From Conven-
tional to Cooperative Techniques. Wiley, June 2010.

[61] Alexander Agathos Kai Hormann. The point in polygon problem for arbitrary
polygons. Computational Geometry, 20(3):131–144, November 2001.

[62] Daniel Krajzewicz. Traffic Simulation with SUMO – Simulation of Urban Mobility,
pages 269–293. Springer New York, New York, NY, 2010.

[63] Daniel Krajzewicz, Jakob Erdmann, Michael Behrisch, and Laura Bieker. Recent
development and applications of SUMO - Simulation of Urban MObility. Inter-
national Journal On Advances in Systems and Measurements, 5(3&4):128–138,
December 2012.

[64] Stefan Krauß. Microscopic Modelling of Traffic Flow: Investigation of Collision


Free Vehicle Dynamics. PhD thesis, Deutsches Zentrum für Luft- und Raumfahrt
e.V. (Hauptabteilung Mobilität und Systemtechnik), 1998.

[65] Bell Laboratories. High-capacity mobile telephone system. Technical Report,


December 1971.

[66] W. C. Y. Lee. Elements of cellular mobile radio systems. IEEE Transactions on


Vehicular Technology, 35(2):48–56, May 1986.

[67] V. H. MacDonald. The cellular concept. Bell System Technical Journal, 1(58):15–
42, January 1979.

[68] Jakob Erdmann Daniel Krajzewicz Michael Behrisch, Laura Bieker.

[69] Yannis Theodoridis Niklos Pelekis. Mobility Data Management and Exploration.
Springer, 2014.

[70] Jeffrey H. Reed Nishith Tripathi. Cellular Communications: A Comprehensive


and Practical Guide. Wiley, October 2014. 1032 pages.

[71] Christoph Sommer, Zheng Yao, Reinhard German, and Falko Dressler. Simulating
the Influence of IVC on Road Traffic using Bidirectionally Coupled Simulators. In
27th IEEE Conference on Computer Communications (INFOCOM 2008): IEEE

73
Workshop on Mobile Networking for Vehicular Environments (MOVE 2008), pages
1–6, Phoenix, AZ, April 2008. IEEE.

[72] V. Pelishok T. Maksymyuk, M. Brych. Stochastic geometry models for 5g hetero-


geneous mobile networks. Smart Computing Review, 5(2):89–101, April 2015.

[73] 5G Initiative Team. 5G White Paper. https://www.ngmn.org/uploads/


media/NGMN_5G_White_Paper_V1_0.pdf, February 2015.

[74] Chai K. Toh. Ad Hoc Wireless Networks: Protocols and Systems. Prentice Hall
PTR, 2001.

[75] F. Calabrese F. Morlot V.A. Traag, A. Browet. Social event detection in massive
mobile phone data using probabilistic location inference. 2011.

74
Appendix

I. Licence

Non-exclusive licence to reproduce thesis and make thesis public


I, Siim-Toomas Marran,

1. herewith grant the University of Tartu a free permit (non-exclusive licence) to:

1.1 reproduce, for the purpose of preservation and making available to the pub-
lic, including for addition to the DSpace digital archives until expiry of the
term of validity of the copyright, and
1.2 make available to the public via the web environment of the University of
Tartu, including via the DSpace digital archives until expiry of the term of
validity of the copyright,

of my thesis
Simulating Cellular Network Behaviour in the Context of Urban Environ-
ment
supervised by Amnir Hadachi and Artjom Lind.

2. I am aware of the fact that the author retains these rights.

3. I certify that granting the non-exclusive licence does not infringe the intellectual
property rights or rights arising from the Personal Data Protection Act.

Tartu, 18.05.2017

75

You might also like