Linkoping Studies in Science and Technology. ThesesNo. 1454

Coverage Planning and Resource Allocation

in Broadband Cellular Access

- Optimization Models and Algorithms

Lei Chen

Department of Science and Technology

Linkoping University, SE-601 74 Norrkoping, Sweden

Norrkoping 2010

Coverage Planning and Resource Allocation in Broadband Cellular Access - Op-timization Models and Algorithmsc©Lei Chen, 2010leich@itn.liu.se

Thesis number: LIU-TEK-LIC-2010:25ISBN 978-91-7393-279-0ISSN 0280-7971

Linkoping UniversityDepartment of Science and TechnologySE-601 74 NorrkopingTel: +46 11 36 34 96Fax: +46 11 36 32 70

Printed by LiU-Tryck, Linkoping, Sweden, 2010

Abstract

The last two decades have witnessed a booming in the use of cellular com-munication technologies. Billions of people are now enjoying the benefits ofmobile communications. This thesis deals with planning and optimization ofbroadband cellular access network design and operation. The problem typesconsidered include coverage planning, power optimization, and channel as-signment. Mathematical modeling and optimization methods have been usedto approach the problems.

Coverage planning is a classical problem in cellular network deployment. Aminimum-power covering problem with overlap constraints between cell pairsis considered. The objective is to minimize the total power consumption forcoverage, while maintaining a necessary level of overlap to facilitate handover.For this coverage planning problem, the thesis develops two integer program-ming models and compares the models’ strength in approaching global opti-mality. In addition, a tabu search algorithm has been developed for solving theproblem in large-scale networks.

For High Speed Downlink Packet Access (HSDPA) networks, transmissionpower is a crucial factor to performance. Minimizing the power allocated forcoverage enables significant power saving that can be used for HSDPA datatransmission, thus enhancing the HSDPA performance. Exploring this poten-tial power saving, a mathematical model targeting cell-edge HSDPA perfor-mance has been developed. In determining the optimal coverage pattern formaximizing power saving, the model also allows for controlling the degree ofsoft handover for Universal Mobile Telecommunications System (UMTS) Re-lease 99 services. In addition to the mathematical model, heuristic algorithmsbased on local search and repeated local search are developed.

For Orthogonal Frequency Division Multiple Access (OFDMA), which is usedin Long Term Evolution (LTE) networks, inter-cell interference control is akey performance engineering issue. The aspect is of particular importanceto cell-edge throughput. Frequency reuse schemes for mitigating inter-cellinterference at cell-edge areas have received an increasing amount of researchattention. In the thesis, a generalization of the standard Fractional FrequencyReuse (FFR) scheme is introduced. The generalization addresses OFDMAnetworks with irregular cell layout. Optimization algorithms using local searchhave been proposed to find the frequency reuse pattern of generalized FFR that

iii

maximizes the cell-edge area performance.

For the problems considered in the thesis, computational experiments of theoptimization models and algorithms using data sets representing realistic plan-ning scenarios have been carried out. The experimental results demonstratethe effectiveness of the proposed solution approaches.

iv

Acknowledgement

This thesis is the result of research carried out within the Department of Sci-ence and Technology (ITN), Linkoping University.

I would like to thank my supervisor, professor Di Yuan, whose excellent guid-ance and continuous help play a very important role in my research work. Iwould like to say that it has been a fantastic experience to work with such anexcellent researcher, who is always encouraging and willing to help.

I would also like to thank all my colleagues at the Division of Communicationand Transport Systems (KTS), especially my roommate Sara, for providingme with a very friendly and relaxed environment, and all the other friends inSweden for helping me to adapt quickly to the Swedish culture. Those yearsof living and study with them will be a very meaningful and unforgettableexperience in my life.

I also want to show gratitude to the financial support from CENIIT, LinkopingInstitute of Technology, Swedish research council (Vetenskapsradet), as wellas Marie Curie project IAPP@Ranplan within the European FP7 research pro-gramme.

Finally, I would like to express my thanks to my family for their love andunderstanding for me to study far away from them, and to my friends for theirencouragement and support all the time.

Norrkoping, December 2010

Lei Chen

v

vi

Abbreviations

1G First Generation

2G Second Generation

3G Third Generation

3GPP 3rd Generation Partnership Project

4G Fourth Generation

AMC Adaptive Modulation and Coding

AMPS Advanced Mobile Phone Service

BS Base Station

CAPEX Capital Expenditure

CDMA Code Division Multiple Access

CoMP Coordinated Multi-Pointing Transmission/Reception

D-AMPS Digital AMPS

DC-HSPA Dual-carrier High Speed Packet Access

DCA Dynamic Channel Allocation

DCH Dedicated Channel

E-DCH Enhanced Dedicated Channel

EDGE Enhanced Data rate for GSM Evolution

EUL Enhanced Uplink

FAP Frequency Assignment Problem

FDMA Frequency Division Multiple Access

FFR Fractional Frequency Reuse

vii

FH Frequency Hopping

GPRS General Packet Radio Service

GSM Global System for Mobile Communications

HARQ Hybrid Automatic Repeat Request

HS-DSCH High Speed Downlink Shared Channel

HSCSD High-Speed Circuit Switched Data

HSDPA High Speed Downlink Packet Access

HSUPA High Speed Uplink Packet Access

ITU International Telecommunications Union

LS Local Search

LTE Long Term Evolution

LTE-A LTE Advanced

MI-FAP Minimum Interference Frequency Planning Problem

MIMO Multiple Input and Multiple Output

Min-GBR Minimum Guaranteed Bit Rate

NMT Nordic Mobile Telephone

NP-hard Non-deterministic Polynomial-time hard

OFDMA Orthogonal Frequency Division Multiple Access

OPEX Operational Expenditure

OR Operational Research

PC Power Control

PDC Personal Digital Cellular

PF Proportional Fair

QAM Quadrature Amplitude Modulation

QoS Quality of Service

RAN Radio Access Network

viii

RB Resource Block

RNC Radio Network Controller

RRM Radio Resource Management

SAE System Architecture Evolution

SC-FDMA Single Carrier Frequency Division Multiple Access

SFR Soft Frequency Reuse

SHO Soft Handover

SINR Signal-to-Interference-plus-Noise Ratio

SIR Signal Interference Ratio

TACS Total Access Communication System

TD-SCDMA Time Division Synchronous Code Division Multiple Access

TDMA Time Division Multiple Access

TS Tabu Search

UE User Equipment

UMTS Universal Mobile Telecommunication Systems

WCDMA Wideband Code Division Multiple Access

WiMAX Worldwide Interoperability for Microwave Access

ix

x

Part I

Introduction and Overview

INTRODUCTION

1 Evolution of Cellular Networks

The concept of cellular communications was introduced by Bell Laborato-ries in 1947 to increase the communication capacity of mobile phones. Inthe 1970s, commercial mobile communications based on cellular technologybegan to appear all over the world. Systems deployed in this period are re-ferred to as the first generation (1G) cellular networks. NTT launched the firstcommercial cellular network in the metropolitan area of Tokyo in 1979. Soonafter, commercial cellular systems became available in Europe and America.Representative systems include the nordic mobile telephone system (NMT) inScandinavia, the total access communication system (TACS) in UK, and theadvanced mobile phone service system (AMPS) in America. Being analog-based, 1G networks suffered from many limitations such as low frequencyutilization, limited service, poor communication quality, high equipment cost,and low security.

To overcome the limitations and meet the demand of mobile communications,the second generation (2G) cellular systems were developed. 2G cellular sys-tems adopted digital transmission that is fundamentally different from the 1Gcellular technology. Besides frequency division multiple access (FDMA) ,which is also used in 1G networks, 2G networks use more advanced accesstechnologies such as time division multiple access (TDMA) and code divisionmultiple access (CDMA) , as well as hierarchical cell structure.

Representative 2G standards include the global system for mobile communica-tions (GSM), interim standard 136 (IS-136, aka Digital AMPS, or D-AMPS),IS-95 (aka CDMAone), and the personal digital cellular (PDC) system. BothGSM and D-AMPS were based on TDMA. GSM was firstly introduced in Eu-rope to provide a single and unified 2G standard in European countries. Later,GSM was implemented in many countries in the rest of the world. D-AMPSevolved from AMPS and was used in America, Israel and some of the Asiancountries. IS-95 systems were based on CDMA, which allowed users to si-multaneously access the same frequency band using different codes. As theonly commercial CDMA standard of that time, IS-95 had been mostly imple-mented in North America and Asia. PDC was a TDMA-based standard andwas available in Japan only.

The primary service in 2G networks was voice communication. To meet theemerging demand for data communication, a number of upgrades were in-troduced to 2G systems as low-cost, intermediate solutions for data services,

3

INTRODUCTION

while developing the third generation (3G) systems. The intermediate steps inthe transition from 2G to 3G were called 2.5G and 2.75G. One such step washigh-speed circuit switched data (HSCSD) for GSM. HSCSD was easy to im-plement and required low cost in deployment. However, since it was still basedon circuit switching, it suffered from inefficient frequency utilization. An-other technology was the general packet radio service (GPRS) . Using packetswitching, GPRS had better frequency utilization in handling best-effort datacommunication. GPRS utilizes dynamically free TDMA channels in GSM sys-tems, and can provide a speed of up to 114 kbps. Later, with the introductionof 8PSK encoding, GPRS evolved to enhanced data rate for GSM evolution(EDGE) , aka enhanced GPRS. EDGE increased the data rate to 384kbps andwas backward-compatible. At the same time, within the CDMA family of stan-dards, IS-95 evolved to the CDMA2000 1 times radio transmission technology(CDMA2000 1xRTT), which supported a data rate of 153kpbs.

Soon after the commercialization of 2G networks, development of 3G cel-lular communication systems started. The goal of 3G was to provide highdata rate and multimedia data services for mobile Internet. Three major fami-lies of specifications, namely CDMA2000, wideband CDMA (WCDMA), andtime division synchronous CDMA (TD-SCDMA) were accepted as 3G stan-dards by the international telecommunications union (ITU). CDMA2000 wasdeveloped from IS-95 and mostly used in North America and South Korea.WCDMA, aka universal mobile telecommunication systems (UMTS) was de-veloped from GSM, with efforts in Europe and Japan. Since GSM dominatedthe 2G commercial markets, the evolution path GSM-GPRS-EDGE-WCDMAwas very natural. At present, WCDMA is the most widely implemented 3Gtechnology. TD-SCDMA was China’s 3G standard, proposed by Datang Tele-com and was mostly used in China.

Since the first release by 3rd generation partnership project (3GPP), referredto as release 99 (R99), WCDMA has evolved rapidly. In 2002, Release 5(R5) was finalized to include high speed downlink packet access (HSDPA). In2005, Release 6 (R6) became available, supporting high speed data serviceson the uplink by introducing enhanced uplink (EUL), aka high speed uplinkpacket access (HSUPA). In 2007, Release 7 (R7) introduced evolved high-speed packet access, aka HSPA+, to support higher order modulation, e.g.,64QAM and multiple input and multiple output (MIMO) technology. In Re-lease 8 (R8), which was finalized in 2008, developments of two standards werespecified, HSPA+ and long term evolution (LTE). HSPA+ was enhanced by adual-carrier HSPA (DC-HSPA) which doubled the throughput by combining

4

INTRODUCTION

two WCDMA radio channels. HSPA+ was a backward-compatible standard,allowing operators that have heavily invested in WCDMA to offer new fea-tures to their subscribers with legacy UMTS terminals. At the same time,the first release of LTE was specified in R8. LTE represented a new evolu-tion path. The LTE standards adopted orthogonal frequency division multipleaccess (OFDMA) for downlink and single carrier FDMA (SC-FDMA) for up-link, to provide higher peak data rate and flexible bandwidth usage. Additionalfeatures, such as femto-cell, were introduced in Release 9 (R9) in 2009.

The ultimate goal of LTE is to pave the way for the fourth generation (4G)systems. Peak data rate of LTE targets 100 Mbps at downlink and 50 Mbpsat uplink with a 20 MHz bandwidth and 2×1 MIMO antenna. In comparisonto R6, average user throughput of LTE is expected to be 3-4 times higher atdownlink, and 2-3 times higher at uplink. In LTE, spectrum utilization is scal-able over blocks of 5, 10, 15, and 20 MHz. Blocks smaller than 5 MHz arealso supported. As LTE is not supposed to be backward-compatible, it enjoysthe freedom of implementing the latest transmission technologies. Besides thenew air interface, advanced antenna solutions such as MIMO, beam-formingare also utilized. A flat all IP core network, namely system architecture evo-lution (SAE) , is designed to replace the GPRS core network and to supporthigh throughput with low latency in the radio access network (RAN). Despitethe economic crisis in 2008 and 2009, trials of LTE networks have not beeninterrupted. The first commercial LTE service was provided by TeliaSonera inStockholm and Oslo in 2009. The network delivers a downlink speed between20 Mbps and 80 Mbps.

To meet the future demand of mobile broadband, the research on LTE-advanced(LTE-A), which represents a 4G standard, is also underway. LTE-A specifica-tions will be part of Release 10 (R10). LTE-A will include more advancedtechnologies like clustering SC-FDMA, relaying, coordinated multi-pointingtransmission/reception (CoMP), higher order MIMO transmission, etc., to takethe peak data rate up to 1 Gbps for low mobility scenarios.

The above discussion applies to the development of standards in 3GPP. Be-sides LTE, worldwide interoperability for microwave access (WiMAX) , spec-ified in IEEE 802.16 series of standards, is also evolving towards fulfillingthe requirement of 4G systems. An illustration of the evolution of cellu-lar networks is given in Fig. 1. More detailed discussions can be found in[1, 15, 26, 27, 28, 42, 58].

5

INTRODUCTION

19911979

speech

9.6kbps

1Mbps

UMTS R99, WCDMA FDD/TDD,CDMA2000, TD-SCDMA,…

2Mbps

HSDPA(R5),HSUPA(R6),HSPA+(R7,R8,…),LTE(R8,…),…

326Mbps

LTE-Advanced(R10),WiMax,…

1Gbps

1999 2000 2002 2010

AMPS,NMT,…

GSM,IS-95,D-AMPS,… 2G

GPRS, EDGE, CDMA 1xRTT,…

2.5/2.75G

3G

3.5/3.75/3.9G

4G

1G

Throu

ghpu

t

Figure 1: Evolution of cellular technology.

2 Cellular Network Planning and Optimization

Network planning and optimization play a key role in reducing the capital ex-penditure (CAPEX) and operational expenditure (OPEX) for deploying andexpanding cellular systems. Typically, radio network planning begins with adefinition and dimensioning stage, which includes traffic estimation, servicedefinition, coverage and capacity requirements, etc. It is traditionally consid-ered as a static process. Some of the main tasks are site selection for basestation location, location area and routing area planning, and radio resourcemanagement (RRM) strategies. Besides fulfilling the initial requirements suchas coverage and capacity, radio resource has to be acquired in a way that thecost is minimized. Optimization is a long-term process before and after thelaunch of a network. The process applies various methods to maximize thesystem performance by optimally configuring the network and utilizing its re-sources. Traditionally, a large amount of manual tuning has been used in theoptimization process. Nowadays, advanced optimization tools have been de-veloped to automatically optimize the parameters for maximizing system per-formance, making the optimization process much more efficient.

6

INTRODUCTION

Analog-based 1G cellular networks did not pose much requirement to plan-ning and optimization. Having low capacity requirement, the key problem isto provide a satisfied degree of coverage. With the success of 2G networksand the increasing amount of voice and data demand, cellular systems havebecome more and more complex. New radio access technologies also intro-duce new challenges to the planning and optimization processes. In the fol-lowing, we outline some representative planning and optimization problemsin cellular networks. More detailed discussions can be found in, for example,[40, 47, 48, 51].

2.1 Base Station Location

A very fundamental planning task in cellular networks is thebase station (BS)location. To deal with the increasing user demand, more and more base sta-tions are needed to provide satisfactory services. A resulting optimizationproblem is to determine how many and where base stations should be locatedin order to meet coverage and capacity requirements. An illustration of theproblem is shown in Fig. 2. In the figure, the red spots require BS installedwhile the blue spots do not.

Figure 2: An illustration of base station location.

In general, the base station location problem is NP-hard [7, 46] in problemcomplexity. For 2G networks, to a large extend, the task can be regarded as atype of set covering problem, the performance is mainly coverage-driven andmostly depends on the propagation model. In this case, BSs are located amongthe candidate locations so that signal strength is high enough for the areas to

7

INTRODUCTION

be covered. For 3G networks with WCDMA, this is however not enough as thesingle-to-interference ratio (SIR) needs to be taken into account. Besides thelocation of BS, coverage itself is heavily influenced by the traffic distribution.Thus, BS location in UMTS networks has to consider power control which inits turn is tightly connected to traffic distribution. In the past years, BS locationhas been extensively studied for both 2G and 3G networks [5, 6, 20, 35, 46,59, 61, 64]. Both mathematical modeling and heuristic algorithms have beenproposed for problem solution.

2.2 Frequency Planning

Along with BS location, another vital issue in 2G GSM networks is to dealwith the frequency assignment problem (FAP) . To avoid interference in GSMnetworks, neighboring cells should use different frequencies, see Fig. 3 for anillustration.

Figure 3: An illustration of frequency assignment.

Due to the high site density and the scarcity of the spectrum resource, fre-quency allocation has to be carefully planned so that high spectral efficiencyand low interference can be achieved. Many optimization methods and algo-rithms have been proposed for FAP in GSM networks [3, 4, 10, 17, 18, 21,25, 37, 50]. Planning strategies for more advanced frequency use, such as fre-quency hopping (FH) [2, 52] and dynamic channel allocation (DCA) [55, 56],have also been proposed.

In WCDMA networks, the frequency reuse factor is one, so FAP is not present.For the new radio interface OFDMA adopted in LTE networks, frequency op-

8

INTRODUCTION

timization becomes again a key issue. LTE is designed to scale well in thespectrum availability. The spectrum is divided into a large number of sub-carriers which are orthogonal to each other. This enables a higher flexibility inrespect to resource allocation. With OFDMA, intra-cell interference does notexist because of the orthogonality, but inter-cell interference may become theperformance-limiting factor, especially for cell-edge users with poor radio sig-nal condition. Sub-carrier allocation has to be done carefully to mitigate inter-cell interference. To this end, fractional frequency reuse (FFR) [19, 43, 60] andsoft frequency reuse (SFR) [31] schemes have been proposed. With the evo-lution from 3G to 4G networks, inter-cell interference mitigation is attractingmore and more research attention.

2.3 Radio Resource Management (RRM)

RRM is fundamental in cellular networks. Many optimization problems in-volving allocation strategies of various types of resource appear in RRM. Theaforementioned FAP can be regarded as part of RRM, since frequency is ahighly valuable resource in cellular networks. Another key resource type istransmission power. Transmission power plays an important role in interfer-ence management, energy consumption, as well as service quality. Thuspowercontrol (PC) mechanisms have been considered. PC can be located at userequipment (UE), BS, or radio network controller (RNC). It is a key compo-nent in IS-95 CDMA systems because of the near-far effect. GSM networksalso has PC mechanism, although the need for PC is not as high as IS-95.

For 3G and 4G systems, more advanced PC mechanisms have been explored.Rather than considering a fixed SIR as in 2G systems, the achievable SIRvaries, and is usually jointly controlled and optimized in the PC mechanismin 3G systems. Furthermore, PC algorithms also need to closely cooperatewith other RRM procedures (e.g. scheduling) to maximize the network perfor-mance. Due to the rapid growth of cellular networks, extensive research hasbeen done on PC algorithms with fixed SIR and variable SIR, opportunisticPC, PC based on game theory, joint PC and beamforming, joint PC and BSlocation, etc. A survey can be found in [14].

In addition to PC, scheduling is an important component of RRM since packetswitching became supported in cellular networks. Scheduling means usuallyto allocate resource among users along the time dimension, to maximize tar-get metrics (e.g. throughput, quality of service (QoS), fairness, etc). Typical

9

INTRODUCTION

scheduling strategies include round robin (RR), maximum throughput, propor-tional fair (PF), minimum guaranteed bit rate scheduling (min-GBR), mini-mum bit rate scheduling with proportional fairness (min-GBR+PF), as well asmaximum delay scheduling. For UMTS R99, which is based on a dedicatedchannel (DCH), scheduling is done at RNC. HSDPA introduces a high speeddownlink shared channel (HS-DSCH). Instead of placing the packet schedulerat RNC, HSDPA utilizes a fast scheduler at Node B, giving Node B the abil-ity to rapidly adapt to the channel conditions of the users. Similar to HSDPA,HSUPA introduces an enhanced DCH (E-DCH) for uplink which also supportsfast scheduling. At present, HSPA is the most widely used cellular technologyfor mobile broadband, and a vast amount of literature has been devoted to itsscheduling algorithms and RRM strategies [8, 9, 12, 23, 24, 29, 30, 33, 36, 53,63].

With OFDMA, the flexibility in radio resource allocation grows. The down-link resource grid in LTE consists of resource blocks (RBs), each consists of12 sub-carriers and 7 OFDM symbols. Depending on the available bandwidth,up to 100 RBs (with a bandwidth of 20 MHz) can be allocated. Power isjointly allocated with RBs. Moreover, information can be transmitted andcombined by multiple antennas, which add one more dimension of freedom.Thus resource allocation in LTE involves several dimensions. Because of theflexibility, the performance of scheduling algorithms is crucial in OFDMA-based networks. Numerous articles investigated resource allocation strategiesfor OFDMA [13, 22, 32, 34, 45, 57, 62, 67, 68], and more specifically for LTE[11, 16, 38, 39, 41, 44, 49, 54, 65, 66].

Additional functionalities, such as admission control and load control, are alsoneeded in RRM. It is worth mentioning the RRM components are not indepen-dent of each other, and the overall performance is a joint effect of the variousRRM functionalities.

3 The Present Thesis

This thesis consists of three research papers in coverage planning in cellularnetworks, power optimization in HSPA networks, and sub-carrier allocationin LTE networks, respectively. The objectives and contributions of the thesiswork along with a summary of future research are give below.

10

INTRODUCTION

3.1 Objectives

The main objective of the thesis is to investigate some optimization problemsarising in coverage planning and resource allocation of cellular networks. Twotypes of resource parameters are considered. The first is the transmissionpower that determines coverage, where coverage is defined in terms of re-ceiving sufficiently well the signal announcing network presence. The secondresource parameter is the frequency band to be allocated in LTE networks.The planning objectives in the optimization problems range from the powerconsumption for coverage, the power requirement for achieving performancethreshold for HSDPA service, to LTE network capacity in terms of cell-edgethroughput.

For the planning problems, the research targets developing optimization mod-els and time-efficient methods that are capable of delivering high-quality solu-tions. The research is also aimed to use realistic planning scenarios to demon-strate the benefit of the optimization approaches.

3.2 Thesis Summary and Contribution

Three papers are included in this thesis. Paper I deals with a coverage planningproblem with overlap constraint for cellular networks. Paper II develops op-timization approaches for enhancing HSDPA performance for networks withco-existing HSDPA and R99 services. Paper III generalizes FFR and devel-ops optimization algorithms for its application in large scale OFDMA-basednetworks with irregular cell structures.

The papers are co-authored with Di Yuan. The author of this thesis has con-tributed to the papers by major involvement in the research planning, the mod-eling and simulation work, the analysis of the results, along with taking a sig-nificant part in authoring the papers.

Paper I: Solving a minimum-power covering problem with overlap con-straint for cellular network design

This paper deals with coverage planning in cellular network design. Giventhe locations of BSs, the problem amounts to determining cell coverage atminimum cost in terms of power usage. Minimum-power coverage tends to

11

INTRODUCTION

make cell size as small as possible; such a solution may have negative impacton handover performance. To tackle this issue, the objective is to performpower minimization with the requirement of having sufficient overlap betweencells.

Two integer linear models are presented. The strength of their respective con-tinuous relaxations as well as the numerical performance of the two modelsare thoroughly investigated. For large-scale networks, a tabu search (TS) al-gorithm is developed. The algorithm is able to obtain close-to-optimal resultswithin short computation time. Simulation results are reported for both syn-thesized networks and networks originating from real planning scenarios.

The paper has been published inEuropean Journal of Operational Research.

Paper II: Coverage planning for optimizing HSDPA performance andcontrolling R99 soft handover

This paper deals with a coverage planning problem for the currently commonnetwork deployment scenario of having co-existing HSDPA and R99 services.By utilizing the power saving resulted from power minimization in coverageplanning, the throughput of HSDPA can be improved. The focus is to bringup the HSDPA performance at cell edge, for which improved data throughputis more perceived in comparison to cell center. At the same time, the level ofsoft handover (SHO) for R99 is taken into account in determining the coveragepattern.

An integer linear model is developed for the problem. The model can be usedto find global optimum for small-sized networks. For large-scale planning sce-narios, a heuristic algorithm is developed. The algorithm is based on localsearch and repeated local search. Performance benchmarking on small testnetworks shows that the algorithm gives close-to-optimality results. In addi-tion, simulation results from the model and the use of the heuristic algorithmdemonstrate significant power saving and HSDPA performance improvementin comparison to the reference solution.

The paper has been accepted for publication inTelecommunication Systems.Parts of the paper have been published in the following conferences:

• L. Chen and D. Yuan. Automated planning of CPICH power for en-hancing HSDPA performance at cell edges with preserved control of

12

INTRODUCTION

R99 soft handover. InProc. of IEEE International Conference on Com-munications (ICC ’08), pages 2936-2940, 2008.

• L. Chen and D. Yuan. Achieving higher HSDPA performance and pre-serving R99 soft handover control by large scale optimization in CPICHcoverage planning. InProc. of IEEE Wireless Telecommunications Sym-posium (WTS ’09), 2009.

Paper III: Generalizing and optimizing fractional frequency reuse in broad-band cellular radio access networks

This paper deals with inter-cell interference mitigation in OFDMA-based net-works. With OFDMA, inter-cell interference is the main performance-limitingfactor for cell-edge users due to their high level of sensitivity to interference.FFR is one of the schemes used for inter-cell interference mitigation by fre-quency separation. Standard FFR uses a fixed number of sub-bands and afixed reuse factor. This works well for a regular, hexagonal cell layout. How-ever, for networks with irregular cell patterns, standard FFR can not be directlyapplied, and the performance is far from being optimal.

In the paper, the standard FFR is generalized to enable a high flexibility in thetotal number of OFDMA sub-bands and the number of sub-bands allocatedin the cells’ edge areas. Two power assignment strategies are considered insub-band allocation. Solution algorithms using local search are developed tooptimize the sub-band allocation for generalized FFR. Performance evalua-tions are conducted for large-scale networks with realistic radio propagationconditions. The results demonstrate the applicability and benefit of applyingand optimizing generalized FFR in performance engineering of OFDMA net-works.

Parts of the paper have been published in the following conferences:

• L. Chen and D. Yuan. Soft frequency reuse in large networks with ir-regular cell pattern: how much gain to expect? InProc. of the IEEE20th International Personal, Indoor and Mobile Radio CommunicationsSymposium (PIMRC ’09), pages 1467-1471, 2009.

• L. Chen and D. Yuan. Generalized frequency reuse schemes for OFDMAnetworks: optimization and comparison. InProc. of IEEE VehicularTechnology Conference (VTC ’10-Spring), 2010.

13

INTRODUCTION

• L. Chen and D. Yuan. Generalizing FFR by flexible sub-band allocationin OFDMA networks with irregular cell layout. InProc. of IEEE Wire-less Communications and Networking Conference (WCNC ’10) Work-shops, 2010.

3.3 Future Research

Cellular network planning and optimization form a complex engineering pro-cess, involving many resource types, network elements, and decision variables.The constant evolution of radio access technologies offers new opportunitiesto efficient radio resource utilization, and at the same time poses challenges tothe planning and optimization process. To master the complexity, the planningand optimization algorithms have to become more advanced, efficient, andautomatic. The latter means that the optimization process should become anintegrated part of the network, and the work of manual tuning must be reducedas much as possible.

The work presented in the thesis contributes to dealing with some planning andoptimization problems in cellular networks. The development of optimizationtools integrating the models and algorithms for the problems will be a naturalfollow-up to the thesis work. There are several additional potential lines for fu-ture research. First, there are many problems that are beyond the scope of thethesis, but highly relevant to the research field. One example is modeling andsolving the problem of optimally locating BSs of OFDMA networks. Second,network optimization problems accounting for recent technological advancessuch as MIMO, relay stations, and heterogeneous radio networks, warrant thor-ough research. Third, approaches that effectively integrate optimization andsystem-level simulation give rise to a very interesting topic for automated net-work planning. Fourth, within the network self-optimization paradigm, thedevelopment and implementation of distributed algorithms for resource allo-cation in broadband mobile access open up new research horizons.

14

Bibliography

[1] 3GPP. http://3gpp.org.

[2] 3GPP. Multiplexing and multiple access on the radio path. TS 45.002,v7.7.0, 2008.

[3] K. Aardal, S. Van Hoesel, A. M. C. A. Koster, C. Mannino, and A. Sas-sano. Models and solution techniques for frequency assignment prob-lems.4OR, 1:261–317, 2003.

[4] M. Almgren, S. M. Magnusson, and K. Wallstedt. Slow adpative channelalocation for automatic frequency planning. InProc. of the 5th IEEEInternational Conference on Universal Personal Communications, pages260–264, 1996.

[5] E. Amaldi, A. Capone, and F. Malucelli. Discrete models and algorithmsfor the capacitated location problems arising in UMTS network plan-ning. InProc. of the 5th International Workshop on Discrete Algorithmsand Methods for Mobile Computing and Communcations (DIAL-M ’01).ACM Press, 2001.

[6] E. Amaldi, A. Capone, and F. Malucelli. Optimizing base station sitingin UMTS networks. InProc. of IEEE Vehicular Technology Conference(VTC ’01-Spring), 2001.

[7] E. Amaldi, A. Capone, and F. Malucelli. Planning UMTS base stationlocation: optimisation models with power control and algorithms.IEEETransactions on Wireless Communications, 2:939–952, 2003.

[8] M. Andrews, K. Kumaran, K. Ramanan, A. Stolyar, and P. Whiting. Pro-viding quality of service over a shared wireless link.IEEE Communica-tions Magazine, 39:150–154, 2001.

15

INTRODUCTION

[9] G. Barriac and J. Holtzman. Introducing delay sensitivityinto the propor-tional fair algorithm for CDMA downlink scheduling. InProc. of Inter-national Symposium on Spread Spectrum Techniques and Applications,pages 652–656, 2002.

[10] P. Bjorklund, P. Varbrand, and D. Yuan. Optimal frequency planning inmobile networks with frequency hopping.Computers and OperationsResearch, 32:169–186, 2005.

[11] F. D. Calabrese, C. Rosa, M. Anas, P. H. Michaelsen, K. I. Pedersen, andP. E. Mogensen. Adaptive transmission bandwidth based packet schedul-ing for LTE uplink. In Proc. of IEEE Vehicular Technology Conference(VTC ’08-Fall), 2008.

[12] M. Cardei, L. Cardei, and D. Z. Du.Resource Management in WirelessNetworking. Springer, 2005.

[13] T. Chee, C. Lim, and J. Choi. Adaptive power allocation with user pri-oritization for downlink orthogonal frequency division multiple accesssystems. InProc. of IEEE International Conference on CommunicationsSystems (ICCS ’04), pages 210–214, 2004.

[14] M. Chiang, P. Hande, T. Lan, and C. W. Tan.Power Control in WirelessCellular Networks. Now Publishers, 2008.

[15] E. Dahlman, S. Parkvall, J. Skold, and P. Beming.3G Evolution: HSPAand LTE for Mobile Broadband. Wiley, 2007.

[16] L. A. M. R. de Teminno, G. Berardinelli, S. Frattasi, and P. Mogensen.Channel-aware scheduling algorithms for SC-FDMA in LTE uplink. InProc. of the IEEE 19th International Symposium on Personal, Indoor andMobile Radio Communications (PIMRC ’08), 2008.

[17] M. Duque-Anto and B. Ruber. Channel assignment for cellular radiousing simulated annealing.IEEE Transactions on Vehicular Technology,42:14–21, 1993.

[18] A. Eisenblatter. Frequency assignment in GSM networks: Models,heuristics, and lower bounds. PhD thesis, Technische Universitat Berlin,2001.

[19] Ericsson. Inter-cell interference handling for E-UTRA. Technical re-port, 3GPP TSG RAN WG1 Meeting Document, Reference number R1-050764, 2005.

16

INTRODUCTION

[20] L. C. P. Floriani and G. R. Mateus. Optimization models foreffectivecell planning design. InProc. of the 1st Interntaional Workshop on Dis-crete Algorithms and Mehods for Mobile Computing and Communica-tions, 1997.

[21] M. Frullone, G. Riva, P. Grazioso, and G. Falciasecca. Advanced plan-ning criteria for cellular system.IEEE Personal Communications, 3:10–15, 1996.

[22] L. Gao and S. Cui. Efficient subcarrier, power, and rate allocation withfairness consideration for OFDMA uplink.IEEE Transactions on Wire-less Commnunications, 7:1507–1511, 2008.

[23] P. J. A. Gutierrez.Packet scheduling and quality of service in HSDPA.PhD thesis, Aalborg University, Denmark, 2003.

[24] W. Hai, W. Lei, and L. Min. HSDPA link adaptation based on novelquality model. InProc. of IEEE Vehicular Technology Conference (VTC’05-Spring), pages 334–338, 2005.

[25] W. K. Hale. Frequency assignment: Theory and applications.Proc. ofthe IEEE, 68:1497–1514, 1980.

[26] H. Holma and A. Toskala.HSDPA/HSUPA for UMTS: High Speed RadioAccess for Mobile Communications. Wiley, 2006.

[27] H. Holma and A. Toskala.WCDMA for UMTS: HSPA Evolution andLTE. Wiley, 2007.

[28] H. Holma and A. Toskala.LTE for UMTS: OFDMA and SC-FDMABased Radio Access. Wiley, 2009.

[29] J. M. Holtzman. CDMA forward link water filling power control. InProc. of IEEE Vehicular Technology Conference (VTC ’00-Spring), pages1663–1667, 2000.

[30] J. M. Holtzman. Asymptotic analysis of proportional fair algorithm. InProc. of the 12th IEEE Personal Indoor and Mobile Radio Communica-tions (PIMRC ’01), pages F33–F37, 2001.

[31] Huawei. Soft frequency reuse scheme for UTRAN LTE. Technical re-port, 3GPP TSG RAN WG1 Meeting Document, Reference number R1-050507, 2005.

17

INTRODUCTION

[32] W. Jiao, L. Cai, and M. Tao. Competitive scheduling for OFDMA sys-tems with guaranteed transmission rate.Elsevier Computer Communica-tions, 32:501–510, 2009.

[33] F. Kelly. Charging and rate control for elastic traffic.European Transac-tions on Telecommunications, 8:33–37, 1997.

[34] K. Kim, Y. Han, and S. Kim. Joint subcarrier and power allocation in up-link OFDMA systems.IEEE Communication Letters, 9:526–528, 2005.

[35] I. Kocsis, I. L. Farkas, and L. Nagy. 3G base station positioning usingsimulated annealing. InProc. of IEEE International Symposium on Per-sonal Indoor and Mobile Radio Communications (PIMRC ’02), pages330–334, 2002.

[36] T. E. Kolding. Link and system performance aspects of proportionalfair packet scheduling in WCDMA/HSDPA. InProc. of IEEE VehicularTechnology Conference (VTC ’03-Fall), pages 1717–1722, 2003.

[37] A. M. C. A. Koster. Frequency Assignment - Models and Algorithms.PhD thesis, Maastricht University, 1999.

[38] R. Kwan, C. Leung, and J. Zhang. Multiuser scheduling on the down-link of an LTE cellular system.Research Letters in Communications,2008:3:1–3:4, 2008.

[39] R. Kwan, C. Leung, and J. Zhang. Proportional fair multiuser schedulingin LTE. IEEE Signal Processing Letters, 16:461–464, 2009.

[40] J. Laiho, A. Wacker, and T. Novosad.Radio Network Planning and Op-timisation for UMTS. Wiley, 2006.

[41] S. B. Lee, I. Pefkianakis, A. Meyerson, S. Xu, and S. Lu. Proportional fairfrequency-domain packet scheduling for 3GPP LTE uplink. InProc.ofIEEE Conference on Computer Communications (INFOCOM ’09), pages2611–2615, 2009.

[42] P. Lescuyer and T. Lucidarme.Evolved Packet System (EPS): The LTEand SAE Evolution of 3G UMTS. Wiley, 2008.

[43] LG Electronics. Interference mitigation in evolved UTRA/UTRAN.Technical report, 3GPP TSG RAN WG1 Meeting Documents, Referencenumber R1-050833, 2005.

18

INTRODUCTION

[44] S. Lu, Y. Cai, L. Zhang, J. Li, P. Skov, C. Wang, and Z. He. Channel-aware frequency domain packet scheduling for MBMS in LTE. InProc.of IEEE Vehicular Technology Conference (VTC ’09-Spring), 2009.

[45] Y. Ma. Constrained rate-maximization scheduling for uplink OFDMA.In Proc. of IEEE Military Communications Conference (MILCOM ’07),2007.

[46] R. Mathar and T. Niessen. Optimum positioning of base stations forcellular radio networks.Wireless Networks, 6:421–428, 2000.

[47] A. R. Mishra. Fundamentals of Cellular Network Planning and Optimi-sation: 2G/2.5G/3G... Evolution to 4G. Wiley, 2004.

[48] A. R. Mishra. Advanced Cellular Network Planning and Optimisation:2G/2.5G/3G... Evolution to 4G. Wiley, 2007.

[49] G. Monghal, K. I. Pedersen, I. Z. Kovacs, and P. E. Mogensen. Qosoriented time and frequency domain packet schedulers for the UTRANlong term evolution. InProc. of IEEE Vehicular Technology Conference(VTC ’08-Spring), pages 2532–2536, 2008.

[50] M. Naghshineh and I. Katzela. Channel assignment schemes for cellu-lar mobile telecommnication systems: A comprehensive survey.IEEEPersonal Communications, 3:10–31, 1996.

[51] M. J. Nawrocki, M. Dohler, and A. H. Aghvami.Understanding UMTSRadio Network: Modelling, Planning, and Automated Optimisation. Wi-ley, 2006.

[52] H. Olofsson, J. Naslund, and J. Skold. Interference diversity gain infrequency hopping GSM. InProc. of IEEE Vehicular Technology Con-ference (VTC ’95), pages 102–106, 1995.

[53] S. Parkvall, J. Peisa, J. Torsner, M. Sagfors, and P. Malm. WCDMA en-hanced uplink - principles and basic operation. InProc. of IEEE Vehicu-lar Technology Conference (VTC ’05-Spring), pages 1411–1415, 2005.

[54] A. Pokhariyal, G. Monghal, K. I. Pedersen, P. E. Mogensen, I. Z. Kovacs,C. Rosa, and T. E. Kolding. Frequency domain packet scheduling underfractional load for the UTRAN LTE downlink. InProc. of IEEE Vehicu-lar Technology Conference (VTC ’07-Spring), pages 699–703, 2007.

19

INTRODUCTION

[55] M. Salmenkaita, J. Gimenez, and P. Tapia. A practical DCA implementa-tion for GSM networks: dynamic frequency and channel assignment. InProc. of IEEE Vehicular Technology Conference (VTC ’01-Spring), pages2529–2533, 2001.

[56] M. Salmenkaita, J. Gimenez, P. Tapia, and M. Fernandez-Navarro. Opti-mizing the GSM/EDGE air interface for multiple services with dynamicfrequency and channel assignment. InProc. of IEEE Vehicular Technol-ogy Conference (VTC ’02-Fall), pages 2215–2219, 2002.

[57] K. Seong, M. Mohseni, and J. M. Cioffi. Optimal resource allocation forOFDMA downlink systems. InProc. of IEEE International Symposiumon Information Theory, pages 1394–1398, 2006.

[58] S. Sesia, I. Toufik, and M. Baker.LTE: From Theory to Practice. Wiley,2009.

[59] H. D. Sherali, M. Pendyala, and T. Rappaport. Optimal location of trans-mitters for micro-cellular radio communications system design.IEEEJournal on Selected Areas in Communications, 14:662–673, 1996.

[60] Siemens. Interference mitigation - considerations and results on fre-quency reuse. Technical report, 3GPP TSG RAN WG1 Meeting Doc-ument, Reference number R1-050738, 2005.

[61] Iana Siomina, Peter Varbrand, and Di Yuan. Automated optimization ofservice coverage and base station antenna configuration in UMTS net-works. IEEE Wireless Communications Magazine, 13:16–25, 2006.

[62] T. Thanabalasingham, S. Hanly, L. Andrew, and J. Papandriopoulos.Joint allocation of subcarriers and transmit power in a multiuser OFDMcellular network. InProc. of IEEE International Conference on Commu-nications (ICC ’06), pages 269–274, 2007.

[63] N. D. Tripathi, J. H. Reed, and H. F. VanLandingham.Radio ResourceManagement in Cellular Systems. Kluwer Academic Publishers, 2002.

[64] K. Tutschku. Demand-based radio network planning of cellular mobilecommunication systems. InProc. of the 17th Annual Joint Conferenceof the IEEE Computer and Communications Societies (INFOCOM ’98),pages 1054–1061. IEEE, 1998.

20

INTRODUCTION

[65] N. Wei, A. Pokhariyal, T. B. Sorensen, T. E. Kolding, and P.E. Mo-gensen. Performance of MIMO with frequency domain packet schedul-ing in UTRAN LTE downlink. InProc. of IEEE Vehicular TechnologyConference (VTC ’07-Spring), pages 1177–1181, 2007.

[66] M. Wernersson, S. Waanstedt, and P. Synnergren. Effects of QoSscheduling strategies on performance of mixed services over LTE. InProc. of the 18th IEEE International Symposium on Personal, Indoorand Mobile Radio Communications (PIMRC ’07), 2007.

[67] Y. Zhang and K. Letaief. Adaptive resource allocation and scheduling formultiuser packet-based OFDM networks. InProc. of IEEE InternationalConference on Communications (ICC ’04), pages 2949–2953, 2004.

[68] Y. Zhang and K. Letaief. Multiuser adaptive subcarrier-and-bit allocationwith adaptive cell selection for OFDM systems.IEEE Transactions onWireless Communications, 3:1566–1575, 2004.

21