University of Nigeria

Transcription

1 Serial No. Author 1 University of Nigeria Virtual Library OSUAGWU, HENRY ONYEMAUCHE PG/M.ENGR/14/68120 Author 2 Author 3 Title: DYNAMIC BANDWIDTH SCHEDULING FOR WCDMA UPLINK TRANSMISSION Keyword: Description: Category: FACULTY OF ENGINEERING DEPARTMENT OF ELECTRONIC ENGINEERING Publisher: Publication Date: Signature: Godwin Valentine Digitally Signed by: Content manager s Name DN : CN = Webmaster s name O= University of Nigeria, Nsukka OU = Innovation Centre 1

2 TITLE DYNAMIC BANDWIDTH SCHEDULING FOR WCDMA UPLINK TRANSMISSION BY OSUAGWU, HENRY ONYEMAUCHE PG/M.ENGR/14/68120 DEPARTMENT OF ELECTRONIC ENGINEERING FACULTY OF ENGINEERING UNIVERSITY OF NIGERIA, NSUKKA MARCH,

3 APPROVAL PAGE DYNAMIC BANDWIDTH SCHEDULING FOR WCDMA UPLINK TRANSMISSION OSUAGWU HENRY ONYEMAUCHE PG/M.ENGR/14/68120 A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUINEERIREMENTS FOR THE AWARD OF MASTER OF ELECTRONIC ENGINEERING (COMMUNICATION) IN THE DEPARTMENT OF ELECTRONIC ENGINEERING, UNIVERSITY OF NIGERIA, NSUKKA OSUAGWU, HENRY ONYEMAUCHE SIGNATURE DATE PROF. COSMAS I. ANI (PROJECT SUPERVISOR) SIGNATURE DATE EXTERNAL EXAMINER SIGNATURE DATE ENGR. DR. M.A. AHANEKU (AG. HEAD OF DEPARTMENT) SIGNATURE DATE PROF. E.S. OBE (CHAIRMAN, FACULTY POSTGRADUATE COMMITTEE) SIGNATURE DATE 3

4 CERTIFICATION This is to certify that OSUAGWU HENRY ONYEMAUCHE, a postgraduate student in the Department of Electronic Engineering with Registration Number PG/M.ENGR/14/68120 have satisfactorily completed the requirements for the course and thesis work for the degree of Master of Engineering (Communications). PROF. COSMAS I. ANI (PROJECT SUPERVISOR) ENGR. DR. M.A AHANEKU (AG. HEAD OF DEPARTMENT) PROF. E.S. OBE (CHAIRMAN, FACULTY POSTGRADUATE COMMITTEE) 4

5 DECLARATION I, Osuagwu Henry Onyemauche, a postgraduate student of the department of Electronic Engineering, University of Nigeria, Nsukka, declare that the work embodied in this dissertation is original and has not been submitted by me in part or in full for any other diploma or degree of this University or any other Universities. OSUAGWU HENRY ONYEMAUCHE PG/M.ENGR/14/68120 DATE 5

6 DEDICATION This work is dedicated to God and to my father Edmond Osuagwu (late) 6

7 ACKNOWLEDGEMENTS I wish to express my profound gratitude to my supervisor, Prof. Cosmas I. Ani for his guidance and attention throughout the duration of this research work. I must acknowledge in a special way all the staff of the Department of Electronic Engineering for making the realization of this research work a success. My gratitude also goes to my mother, Mrs Angela Osuagwu for her prayers and support. I must acknowledge my brothers and sisters for their support and encouragement. I equally want to acknowledge my good friends Agashi Boniface, Ajibo Augustine and Anike Uchenna for their concerns throughout the duration of the program. I thank everyone who has contributed in one way or the other to ensure the successful completion of this research work. 7

8 ABSTRACT Providing quality of service is a challenging issue in UMTS mobile networks for multimedia traffic (video, voice and data). Critical services such as real-time audio, voice and video are given priority over less critical ones, such as file transfer and web surfing. One of the approaches that efficiently provides standard quality of service for multimedia traffic in wireless networks is to dynamically allocate bandwidth to varying traffic load and channel conditions. There are several of such dynamic bandwidth allocation approaches developed in the recent time by researchers. The choice of which one to implement at an instance and for a specific condition is an issue in mobile communication networks. In this work, the popular Code-Division Generalized Processor Sharing (CDGPS) was analyzed. The CDGPS variations priority and non-priority were compared, the two techniques were modelled and simulated using MATLAB Simulink object oriented environment. Simulation results show that priority CDGPS provides the best performance and improvement in the delay and loss rate, while still maintaining a high bandwidth utilization of percentage value of 98.2%. 8

9 TABLE OF CONTENTS Title page i Approval page ii Certification iii Declaration iv Dedication v Acknowledgement vi Abstract vii Table of contents viii List of Figures xi List of Tables xiii List of Acronyms xiv CHAPTER 1: INTRODUCTION Background of the study Statement of problem Aim and Objectives Scope of the work Significance of Study Methodology Thesis outline CHAPTER 2: LITERATURE REVIEW Overview and Third Generation Technology Requirements for Third-Generation system Wideband Code Division Multiple Access Third Generation GSM objectives and capabilities UMTS Multi-radio evolution path UMTS Network Architecture User Equipment (UE) UMTS Terrestrial Radio Access Network (UTRAN) The Core Network UMTS protocol of operation Radio Interface protocol structure User Plane Control Plane

10 2.7 Radio Interface protocol reference layer Physical (PHY) layer Medium Access Control (MAC) layer Radio Link Control (RLC) protocol Packet Data Convergence Protocol (PDCP) Radio Resource Control (RRC) layer Radio Resource Management (RRM) Radio Resource Management (RRM) Function Radio Resource Management (RRM) Function Interaction Scheduling Schemes First-In-First-Out Scheduling Weighted Round Robin Scheduling Priority Scheduling Earliest-Due-Date Scheduling Rate-Controlled Scheduling Requirements of a Scheduler Related Works Conclusion CHAPTER THREE: RESEARCH METHODOLOGY System Model Generalized Processor Sharing (GPS) The Code-Division Generalized Processor Sharing (CDGPS) scheme Traffic Source Model Voice Source Modeling Video Source Modeling Data Source Modeling Model Validation Conclusion CHAPTER FOUR: SIMULATION AND RESULT ANALYSIS Introduction MATLAB Simulation Framework Performance metrics Simulation Results

11 CHAPTER FIVE: CONCLUSION AND RECOMMENDATION Conclusion Recommendation for future work Contribution to knowledge REFERENCE

12 LIST OF FIGURES Figure 2.1: UMTS multi-radio network Figure 2.2: UMTS Network Architecture Figure 2.3: UTRAN architecture Figure 2.4: UMTS protocols Figure 2.5: Radio interface protocol reference architecture Figure 2.6: Protocol termination for a common channel Figure 2.7: Physical layer for transmitting situation Figure 2.8: Frame structure for downlink DPCH Figure 2.9: Frame structure for downlink PDSCH Figure 2.10: Structure of the random-access transmission Figure 2.11: MAC layer architecture Figure 2.12: RLC sub-layer architecture Figure 2.13: Location of RRM functions Figure 2.14: Radio Resource Management Functions Interaction Figure 2.15: FIFO Scheduling Figure 2.16: Weight Round Robin Scheduling Figure 2.17: Priority Queuing Scheduler Figure 3.1: Network Structure Figure 3.2: A queuing model of the CDGPS scheme Figure 3.3: Priority CDGPS flowchart Figure 3.4: Non-priority CDGPS flowchart Figure 3.5: On-Off model Figure 3.6: On-Off voice packetization Figure 3.7: Model validation with CDGPS scheme Figure 4.1: Simulation Framework for WCDMA systems Figure 4.2: Multimedia IP Traffic (voice, video and data) Figure 4.3: Buffer queuing Model

13 Figure 4.4: The system Server Figure 4.5: CDGPS computational model Figure 4.6: A Scope of entities generated Figure 4.7: Throughput as a function of Traffic intensity for multimedia IP traffic 72 Figure 4.8: Throughput per flow as a function of traffic intensity Figure 4.9: Average delay as a function of Traffic intensity Figure 4.10: Loss rate as a function of traffic intensity Figure 4.11: Backlogged flow loss rate as a function of traffic intensity - 76 Figure 4.12: Bandwidth utilization as a function of traffic intensity

14 LIST OF TABLES Table 2.1: Main differences between WCDMA and GSM air interfaces - 7 Table 2.2: Relationship between spreading factor and bit rate Table 2.3: Four UMTS service class Table 3.1: Simulation parameters

15 LIST OF ACRONYMS 1G First Generation 2G Second Generation 3G Third Generation 3GPP Third Generation Partnership Project AC Admission Control ADV Access Delay Variation AM Acknowledgement Mode AS Access Stratum AuC Authentication Center BCCH Broadcast Control Channel BER Bit Error Rate BMC Broadcast/Multicast Control BSC Base Station Controller BTS Base Transceiver Station CCCH Common Control Channel C-CDGPS Credit-based Code-Division Generalized Processor Sharing CCTrCHs Coded Composite Transport Channel CDGPS Code-Division Generalized Processor Sharing CDMA Code Division Multiple Access CM Connection Management CN Core Network CPCH Common Packet Channel CS Circuit Switch CTCH Common Traffic Channel DBA Dynamic Bandwidth Allocation DCA Dynamic Code Assignment DCH Dedicated Channel DFS Delay Fair Scheduling DPA Dynamic Priority Allocation DPCCH Dedicated Physical Control Channel DPCH Downlink Dedicated Physical Channel DRR Deficit Round Robin DRS Dynamic Resource Scheduling DS-CDMA Direct Sequence-Code Division Multiple Access DSCH Downlink Shared Channel DTCH Dedicated Traffic Channel EDD Earliest Due Date EDF Earliest Deadline First EDGE Enhance Data Rates for GSM Evolution EIR Equipment Identity Register ETSI European Telecommunication Standard Institute 15

16 FACH Forward link Access Channel FDD Frequency Division Duplex FIFO First-In-First-Out FLC Fuzzy Logic Controller FTP File Transfer Protocol GERAN GSM/EDGE Radio Access Network GGSN Gateway GPRS Support Network GMM GPRS Mobility Management GMSC Gateway Mobile Switching Center GPRS General Packet Radio Service GPS Generalized Processor Sharing GSM Global System for Mobile Communication HC Handover Control HLR Home Location Register IMT-2000 International Mobile Telecommunication 2000 IP Internet Protocol IPv4 Internet Protocol version 4 IPv6 Internet Protocol version 6 ITU International Telecommunication Union LC Load Control MAC Medium Access Control MDRR Multi-flow Deficit Round Robin ME Mobile Equipment MM Mobility Management MMS Multimedia Message Service MSC Mobile Switching Center MWF 2 Q+ Multi-flow Worst-case Fair Weighted Fair Queuing Plus NAS Non-Access Stratum OVSF Orthogonal Variable Spreading Factor PC Power Control PCCH Paging Control Channel PCH Paging Channel PDCP Packet Data Convergence Protocol PDSCH Physical Downlink Shared Channel PDU Packet Data Unit PRACH Physical Random Access Channel PS Packet Scheduling PS Packet Switch QPSK Quadruped Phase Shift Keying RACH Random Access Channel RLC Radio Link Control RM Resource Manager RNC Radio Network Controller 16

17 RRC RRM RTBS RTCS RTE RTGS SAD SF SGSN SIM SM SMS STFQ TBs TDD TDMA TFCI TFCS TM TPC UE UM UMTS USIM UTRAN VLR WCDMA WFQ WF 2 Q WRR Radio Resource Control Radio Resource Management Real Time Bandwidth Scheduling Real Time Code Assignment Real Time Emulator Real Time Generic Scheduling Service Access Delay Spreading Factor Servicing GPRS Support Network Subscriber Identity Module Session Management Short Message Service Start-Time Fair Queuing Transport Block Time Division Duplex Time Division Multiple Access Transport Format Control Indicator Transport Format Combination Set Transport Mode Transmit Power Control User Equipment Un-acknowledgement Mode Universal Mobile Telecommunication System UMTS Subscriber Identity Module UMTS Terrestrial Radio Access Network Visitor Location Register Wideband Code Division Multiple Access Weighted Fair Queuing Worst-case Fair weighted Fair Queuing Weighted Round Robin 17

18 CHAPTER ONE INTRODUCTION 1.1 Background of the study Today, mobile communications play a central role in the voice/data network arena. From the early analog mobile first generation (1G) to the third generation (3G) the standard has changed. The new mobile generations do not pretend to improve the voice communication experience but try to give the user access to a new global communication reality [1]. The aim is to reach communication universality and to provide users with a new set of services. The cellular networks are evolving through several generations; the first generation (1G) wireless mobile communication network was analog system which was used for public voice service with the speed up to 2.4kbps. The second generation (2G) is based on digital technology and network infrastructure. As compared to the first generation, the second generation can support text messaging [2]. Its success and the growth of demand for online information via the internet prompted the development of cellular wireless system with improved data connectivity, which ultimately leads to the third generation systems (3G). It is now time to explore new demands and to find new ways to extend the mobile concept. The first steps have already been taken by the 2.5G, General Packet Radio Service (GPRS) and Enhanced Data Rates for GSM Evolution (EDGE), which gave users access to a data network (e.g. Internet access, Multimedia Message Service). However, users and applications demanded more communication power. As a response to this demand a new generation with new standards has been developed-third generation (3G). Third generation (3G) networks offer greater security than their 2G predecessors. By allowing the UE (User Equipment) to authenticate the network it is attaching to, the user can be sure the network is the intended one and not an impersonator [3]. With all its enhancements, Global System for Mobile Communication (GSM) will represent the 18

19 mainstream of mobile communication systems for the next several years. However it is obvious due to technical and economic reasons, GSM will be followed by third generation (3G) mobile communication system. Third generation (3G) mobile communication system, called Universal Mobile Telecommunication System (UMTS) within European Telecommunication Standard Institute ETSI/Europe, aim to support a wide range of voice and data services, focusing on mobile packet switched data services based on Internet Protocol (IP) technology [4]. Moreover, UMTS will give the mobile user performance similar to the fixed network and will stimulate the development of new mobile multimedia applications. 1.2 Statement of problem Integrated services networks support multiple services and are faced with problem of resource sharing among applications. Providing quality of service ( ) and resource allocation is a challenging issue especially in mobile networks with applications of multimedia traffic (video, voice and data). 1.3 Aim and objectives The aim of this work is to design a dynamic bandwidth scheduling framework which can improve the overall performance of radio resource management strategy in the UMTS. The specific objectives includes the following: Development of scheduling scheme that would support differentiated quality of service ( ) for Universal Mobile Telecommunication System (UMTS) traffic. Develop a scheduling scheme that would optimized bandwidth utilization. Develop a scheduling scheme that would introduced dynamic bandwidth sharing mechanism for backlogged flows. 1.4 Scope of the work 19

20 The scope of this work is the third generation (3G) mobile communication system, called Universal Mobile Telecommunication System (UMTS). It is specific to the modelling of the UMTS uplink scheduler. 1.5 Significance of Study As the Internet evolves into the global infrastructure, there is a growing need to provide a broad range of quality of service guarantee for different applications, which bring forth the necessity of traffic management. This research will be significant to wireless network providers and researchers in finding an effective means of utilizing the available scarce resource in a heterogeneous traffic environments. Also, it will help service provider in putting into consideration the distribution of residual bandwidth among backlogged (active user) session in an equitable manner. 1.6 Methodology To realize the objectives of this work, the following methodology was adopted: Review of UMTS uplink transmission techniques and resource allocation schemes. Review of existing uplink and downlink scheduling scheme in UMTS. Propose a scheme following the best UMTS scheduling scheme from the review. Development of choice computer models of scheduling scenarios and implementing the proposed scheme. Validation of the results of the analysis with performance of existing schemes. Simulation of choice model and obtain data. Analyze data in terms of performance metrics. Compare the performance of the propose scheme in terms of performance improvement. 1.7 Thesis outline The remainder of the thesis is organized as follow: 20

21 In chapter two, a review of UMTS quality of service (QoS) architecture was carried out and radio resource management service functions that provide the background knowledge for the design of resource allocation scheme for the UMTS system. Chapter three is focused on the general system model requirements for achieving the dynamic radio resources allocation. In chapter four, a MATLAB simulation framework is created for analyzing the proposed scheduling algorithm. The simulation results are presented and analyzed. Chapter five provides the conclusion of the thesis, some recommendations for future work and the contribution to knowledge made by this research work. 21

22 CHAPTER TWO LITERATURE REVIEW 2.1 Overview and Third Generation Technology Many packets scheduling algorithms have been extensively studied in the wired networks, such as weighted fair queuing (WFQ), worst-case fair weighted fair queuing (WF 2 Q), deficit round robin (DRR) and start-time fair queuing (STFQ). These results have also been extended to local wireless networks by several researchers. However, due to several unique features of 3G networks, scheduling algorithms proposed for wired and wireless networks in the literature are not directly applicable to 3G networks. Several scheduling schemes have been proposed in the literature for IP-based radio access networks in WCDMA to efficiently utilize radio resources. The review of these literatures are detailed in (section 2.11). Third Generation Technology was developed in order to face up to the new requirements of services that were coming, as high-quality images and video or to provide access to the Web with higher data rates. Third-generation radio access technologies aim to provide the commercial market with high quality, efficient and easy-to-use wireless mobile multimedia services [5]. All 2G wireless systems are voice-centric, most 2G systems also support some data over their voice paths, but at painfully slow speeds usually 9.6 Kb/s or 14.4 Kb/s. So in the world of 2G, voice remains fundamental while data is already dominant in wire-line communications. And, fixed or wireless, all are affected by the rapid growth of the Internet. Planning for 3G started in the 1980s. Initial plans focused on multimedia applications such as videoconferencing for mobile phones. When it became clear that the real killer application was the Internet, 3G thinking had to evolve [6]. 22

23 Since the third-generation (3G) mobile radio systems will provide us from low to high data rate services with a maximum data rate of 2 Mbps, it can be used in several multimedia applications such as voice, audio/video, graphics, data, Internet access, and . These services, regardless of based on packet switched or circuit switched, have to be supported by the radio interface and the network subsystem [7]. In January 1998, the 3GPP (Third- Generation Partnership Project) has agreed on the UMTS (Universal Mobile Telecommunication System) for 3G mobile radio systems. 2.2 Requirements for Third-Generation system The second generation systems were built mainly to provide speech services in macro cells [8,9]. To understand the background to the differences between second and third generation systems, we need to look at the new requirements of the third generation systems which are listed below: Bit rates up to 2 Mbps; Variable bit rate to offer bandwidth on demand; Multiplexing of services with different quality of service requirements on a single connection, e.g. speech, video and packet data; Delay requirements from delay-sensitive real-time traffic to flexible best-effort packet; Quality requirements from 10% frame error rate to 10-6 bit error rate; Coexistence of second-generation and third-generation systems and inter-system handovers for coverage enhancements and load balancing; Support of asymmetric uplink and downlink traffic, e.g. web browsing causes more loading to downlink than to uplink; High spectrum efficiency. Co-existence of FDD and TDD modes The table 2.1 lists the main differences between WCDMA and GSM. In this comparison only the air interface is considered. 23

24 Table 2.1 Main differences between WCDMA and GSM air interfaces [9] WCDMA The differences in the air interface reflect the new requirements of the third-generation systems. For example, the larger bandwidth of 5 MHz is needed to support higher bit rates Wideband Code Division Multiple Access (WCDMA) GSM Carrier spacing 5 MHz 200 khz Frequency reuse factor Power control frequency 1500 Hz 2 Hz or lower Quality control Radio resource management Network planning (frequency algorithms planning) Frequency diversity 5 MHz bandwidth gives Frequency hopping multipath diversity with Rake receiver Packet data Load-based packet scheduling Time slot based scheduling with GPRS Downlink transmit diversity Supported for improving Not supported by the standard, but downlink capacity can be applied This section introduces the principles of the WCDMA air interface. Brief explanations for most of the main system design parameters of WCDMA were presented [8]. Special attention is drawn to those features by which WCDMA differs from GSM. WCDMA is a wideband Direct-Sequence Code Division Multiple Access (DS- CDMA) system, i.e. user information bits are spread over a wide bandwidth by multiplying the user data with quasi-random bits (called chips) derived from CDMA spreading codes. In order to support very high bit rates (up to 2 Mbps), the use of a variable spreading factor and multi-code connections is supported. 24

25 WCDMA supports highly variable user data rates, in other words the concept of obtaining Bandwidth on Demand (BoD) is well supported. The user data rate is kept constant during each 10 frame. However, the data capacity among the users can change from the frame to frame. This fast radio capacity allocation will typically be controlled by the network to achieve optimum throughput for packet data services. WCDMA supports two basics modes of operation: Frequency Division Duplex (FDD) and Time Division Duplex (TDD). In the FDD mode, separate 5 MHz carrier frequencies are used for the uplink and downlink respectively, whereas in TDD only one 5 MHz is timeshared between the uplink and downlink. Uplink is the connection from the mobile to the base station, and downlink is that from the base station to the mobile. The WCDMA air interface has been crafted in such a way that advanced CDMA receiver concepts, such as multiuser detection and smart adaptive antennas can be deployed by the network operator as a system option to increase capacity and/or coverage. In most second generation systems, no provision has been made for such receiver concepts and as a result they are either not applicable or can be applied only under severe constraints with limited increases in performance. WCDMA is designed to be deployed in conjunction with GSM. Therefore, handovers between GSM and WCDMA are supported in order to be able to leverage the GSM coverage for the introduction of WCDMA. 2.3 Third Generation GSM objectives and capabilities 3G GSM (UMTS) is an upgrade from GSM via GPRS or EDGE. IMT-2000 is an ITU s umbrella name for 3G which stands for International Mobile Telecommunications Kbps data rate available to users in high-speed motor vehicles over large areas. 384 Kbps available to pedestrians standing or moving slowly over small areas. Support (to be phased in) for Mbps for office use. Support for both packet-switched and circuit-switched data services. 25

26 More efficient use of the available spectrum in general. Support for a wide variety of mobile equipment. Flexibility to allow the introduction of new services and technologies. 2.4 UMTS Multi-radio evolution path 3GPP (Third Generation Partnership Project) is a global project aiming to develop open standards for the UMTS third Generation Mobile System based on evolved GSM core networks. This multi-radio mobile system comprises two different third generation (3G) radio access networks, GERAN (GSM/EDGE Radio Access Network) and UTRAN (UMTS Terrestrial Radio Access Network), which are based on different radio access technologies, GSM/EDGE and WCDMA (Wideband Code Division Multiple Access), respectively [5]. Initially, Universal Mobile Telecommunication System (UMTS) defined, within the scope of UMTS Terrestrial Radio Access Network (UTRAN) standardization, new radio access network architecture, with protocols, interfaces and quality of service architectures specifically designed for the efficient provision of third generation multimedia services. GSM/EDGE Radio Access Network (GERAN) has adopted all these, hence becoming an integral part of the UMTS third generation frame. Furthermore, 3GPP standards ensure that an efficient integration between UTRAN and GERAN can be accomplished so that they can be merged under a single UMTS multi-radio network. This concept is illustrated in the Figure 2.1. Integrated radio resource management base on QoS management Circuit switch core USIM ME GERAN UTRAN UMTS 3G multi-radio 3G Core Network Packet core Network UMTS 3G multi-radio access network 26

27 Figure 2.1 UMTS multi-radio network [5] UMTS is based on Wideband Code Division Multiple Access (WCDMA) radio technology, which offers higher throughput, and better real-time services. The UMTS radio access network offers multimedia applications like simultaneous transfer of speech, data, text, pictures and audio a maximum data rate of 2Mbps, which is a result of using 5MHz bandwidth of the radio channels in UMTS instead of 200 khz in GSM [10]. The 3G WCDMA air interface has been designed to provide a packet based wireless service, by which different computing and telephone devices all share the same wireless network and may be connected to the Internet anytime and anywhere. 2.5 UMTS Network Architecture Universal Mobile Telecommunications System (UMTS) is a 3G cellular telecommunication system. It will be the successor of GSM. UMTS is designed to cope with the growing demand of mobile and internet applications with required quality of service parameters [11]. WCDMA is used for the radio interface of UMTS. The UMTS network has three subsystems to address different operations [12]. They are UMTS terrestrial random access network (UTRAN), core network (CN) and user equipment (UE). Figure 2.2, is a UMTS network architecture with its basic domains and this figure also show its external reference points and interfaces with the UTRAN. UTRAN is connected the core network (CN) via Iu interface. Between the radio networks controller (RNC) and Core Network, there is Iu UTRAN interface. The UTRAN interface that is between the CN and the radio network controller (RNC) is called Iu-PS and also UTRAN interface between the RNC and circuit switched domain of CN is known as Iu-CS. Radio interface between User equipment UE and UTRAN is known as Uu interface. These interfaces are also known as reference. 27

28 Mobile station Base Station Subsystem Network Subsystem Other Networks SIM M E USIM ME Uu BTS RNS Node B BSC RNC Iu MSC/ VLR GMSC EIR HLR AuC SGSN GGSN PSTN PLMN Internet UE UTRAN Core Network Figure 2.2 UMTS Network Architecture [13] User Equipment (UE) The user equipment is the physical device which enables the user to have access to network services. The UE consists of ME (Mobile Equipment) and USIM (UMTS Subscriber Identity Module). The ME is a radio terminal used for communication over Uu interface. The ME consists of the Mobile Termination (MT), which performs the radio transmission, and Terminal Equipment (TE), that enables end-to-end application, e.g., a laptop that is connected to a mobile phone [14]. The USIM is a smartcard that holds the subscriber identity, performs authentication algorithms, and stores authentication and encryption keys and some subscriber information that is needed at the terminal UMTS Terrestrial Radio Access Network (UTRAN) UTRAN consists of one or more Radio Network Sub-systems (RNSs). An RNS is a subnetwork within UTRAN and consists of one RNC and one or more Node Bs. RNCs may 28

29 be connected to each other via an Iur interface. RNCs and Node Bs are connected with an Iub Interface. During Release 7, work study on the support of small RNSs was done, meaning the use of collocated RNC and Node B functionalities in a flat architecture, and that was found feasible without mandatory specification changes [8]. The Node B The Node B converts the data flow between the Iub and Uu interfaces. It also participates in radio resource management. It logically corresponds to GSM Base Station but the term Node B was initially adopted as a temporary term during the standardization process and then never changed. The Radio Network Controller (RNC) The RNC is the network element responsible for the control of the radio resources of UTRAN. It interfaces the CN (normally to one MSC and one SGSN) and also terminates the Radio Resource Control (RRC) protocol that defines the messages and procedures between the mobile and UTRAN. It logically corresponds to the GSM BSC. UTRAN Interfaces The UTRAN interfaces are as follows: Iub interface - The Iub connects a Node B and a RNC. Iur interface - The open Iur interface allows soft handover between RNCs from different manufacturers and, therefore, complements the open Iu interface. Iu interface - This connects UTRAN to the CN and is similar to the corresponding interfaces in GSM, the open Iu interface gives UMTS operators the possibility of acquiring UTRAN and CN from different manufacturers. The enabled competition in this area has been one of the success factors of GSM. UTRAN interfaces are shown in figure

30 U u lub UE Node B RNC l ub l ur lu MSC Node B SGSN l ub RNC lu Node B Figure 2.3 UTRAN architecture [8] The Core Network HLR (Home Location Register):This is a database located in the user s home system that stores the master copy of the user s service profile [15]. The HLR also stores the UE location on the level of MSC and SGSN. MSC/VLC (Mobile Switching Center/Visitor Location Register): The MSC function is used to switch the CS (Circuit Switch) transactions, and VLR function holds a copy of the visiting user s service profile, as well as more precise information on the UE s location within the serving system. GMSC (Gateway MSC): The Switch at the point where UMTS is connected to external CS networks. All incoming and outgoing CS connections go through GMSC. SGSN (Serving GPRS Support Node): Similar to that of MSC / VLR but is used for Packet Switched (PS) services. The part of the network that is accessed via the SGSN is often referred to as the PS domain. It is an upgraded version of serving GPRS support node. 30

31 GGSN (Gateway GPRS Support Node): Functionality is close to that of GMSC but is in the relation to PS services. It is an upgraded version of gateway GPRS support Node 2.6 UMTS Protocol of operation The communication among the different entities of the UMTS architecture involves several protocol stacks that are defined for each interface and are depicted in figure 2.4. A protocol stack defines a set of layers that specify the communication procedures between two network entities [16]. Each layer in a network entity (e.g. the UE) communicates with the same layer of the network entity (e.g. the node B) by means of a specific protocol that includes a set of procedures involving a number of messages transferred between both entities. From a vertical perspective, a given layer provides the means for the transfer of the messages originated at the above layers. In turn, from a horizontal perspective, the concatenation of several protocol stacks allows the communication between non-adjacent entities (e.g. between the User Equipment (UE) and the Core Network). The unifying principle in the UTRAN development work has been to keep the mobility management (MM) and connection management (CM) layers independent of the air interface radio technology [17]. This idea has been realized as the access stratum (AS) and non-access stratum (NAS) concepts (Figure 2.4). The access stratum (AS) is a functional entity that includes radio access protocols between the UE and the UTRAN. These protocols terminate in the UTRAN. The NAS includes core network (CN) protocols between the UE and the CN itself. These protocols are not terminated in the UTRAN, but in the CN; the UTRAN is transparent to the NAS. The Mobility Management (MM) and Connection Management (CM) protocols are GSM Core Network protocols; GPRS Mobility Management (GMM) and Session Management (SM) are GPRS Core Network protocols. 31

32 Figure 2.4 UMTS Protocols [16] Just as the NAS tries to be independent of the underlying radio techniques, so also have the MM, CM, GMM, and SM protocols tried to remain independent of their underlying radio technologies. The Connection Management (CM) and Session Management (SM) protocols responsible for the establishment and release of connections or sessions for an UE, respectively or the Mobility Management (MM) and GPRS Mobility Management (GMM) protocols, responsible for dealing with mobility functions at the network layer (e.g. location area updating, routing area updating, paging, etc.). In turn, in the user plane, the main NAS protocol at the network layer for packet switched services is the IP protocol, while for circuit services information comes directly from the source without the need for a network protocol. In the UMTS architecture, the access stratum (AS) includes three different protocol stacks, namely the radio interface protocols, the Iub interface protocols and the Iu interface protocols. In particular, the radio interface protocol stack allows communication between the UE and the UMTS access network (UTRAN). Note that the protocols at the upper layers terminate in the UE and RNC, while the lower layers terminate in the UE and Node 32

33 B. With respect to the Iub interface protocols, they involve the communication of the lower layers of the RNC and the Node B. Finally, the Iu interface protocols allow communication between the RNC and the CN, distinguishing between the Iu-CS for communication between RNC and MSC and the Iu-PS for communication between RNC and SGSN Radio Interface Protocol structure In this thesis, this work is focused on the management of the resources at the radio interface, whose scarcity constitutes in most cases the bottleneck for a proper communication to be carried out. In Figure 2.5 the UMTS radio interface protocol stack is shown. The termination of each protocol can be seen in Figure 2.6. The radio interface protocol is comprised of three layers: The physical layer PHY (layer 1) The data link layer (layer 2) Network layer (layer 3). Figure 2.5 Radio interface protocol reference architecture [16] 33

34 Figure 2.6 Protocol termination for a common channel [16] Layer 1 The layer 1 is the physical layer which is based on WCDMA technology with a chip rate of 3.84 h/. It offers data transport services to the MAC layer via transport channels. Transport channels are characterized by how the information is transferred to the radio interface. They are divided into common and dedicated transport channels. Layer 2 Layer 2 is split into the following sub layers: Medium Access Control (MAC), Radio Link Control (RLC), Packet Data Convergence Protocol (PDCP) and the Broadcast/Multicast Control (BMC) layer. The BMC layer is responsible for the management of broadcast and multicast messages like the SMS Cell Broadcast Service. The MAC layer offers logical channels to the RLC layer. A logical channel is defined by what type of information is transferred. A general classification of logical channels is into control channels and traffic channels. Control channels are used to transfer higher layer signaling messages. Traffic channels are used for transfer of user information. There is a Dedicated Traffic Channel (DTCH) and a Common Traffic Channel. The CTCH is used for broadcasting messages. The DTCH is a point-to-point channel, dedicated to one user. 34

35 Layer 3 Layer 3 consists of one protocol in the control plane. This protocol is the radio resource Control (RRC). It handles all messages required to set up, modify and release layer 1 and layer 2 entities. The radio interface protocol is also divided into control and user plane. The control plane provides services for transmitting signaling messages. The user plane is responsible for user data transmission [14] User Plane Radio Link Control (RLC): Presents a reliable channel to higher layers by retransmitting erroneous packets Medium Access Control (MAC): Channel access, multiplexing traffic streams, scheduling priority flows Physical Layer (PHY): Measurements, power control algorithms Control Plane Radio Resource Control (RRC): Connection and management. Radio Resource Management (RRM): Algorithms for admission control, handovers Radio Interface protocol reference layer Physical (PHY) layer This section presents the characterization of the physical layer, whose mission is to transform the flow of information coming from the different transport channels into physical radio signals transmitted by the antenna [18]. The physical layer at the transmitter side receives Transport Blocks (TBs) from the MAC layer. These transport blocks may belong either to one or to several transport channels that are simultaneously multiplexed. Then, the physical layer executes a set of procedures over the received transport blocks to 35

36 generate the radio signal that is sent to the antenna. At the receiver side, the reverse procedures are carried out to recover the transport blocks from the received physical signal at the antenna and to deliver them to the MAC. Figure 2.7 shows an overview of the activities in the physical layer for a transmitting situation. Figure 2.7 Physical layer for transmitting situation [14] The first baseband signal processing entity includes channel coding, rate matching, interleaving and others. Channel coding can be applied as either convolutional coding or turbo coding with rate 1/2 or 1/3. It's also possible that no coding is done. Further the data stream is segmented into 10ms blocks which are multiplexed with blocks of other transport channels. The transport channel multiplexing entity combines all coded transport channels into one special channel, called Coded Composite Transport Channel (CCTrCH), which is split up into one or several physical channels later on. These physical channels are separately spreaded and scrambled. Spreading is done with an Orthogonal Variable Spreading Factor (OVSF) code which ensures orthogonality between the different physical channels of a user and which enables different data rates for them. In addition to spreading, scrambling is provided in order to separate terminals or base stations from each other. The symbol rate is not affected by the scrambling operation anymore. After scrambling all the 36

37 data sequences are summed up and modulated. In the downlink direction, Quadruped Phase Shift Keying (QPSK) modulation is applied. Physical Channel Physical channels are defined by a certain carrier frequency, scrambling code and spreading code. Spreading of the low-bandwidth data signal to produce the wideband CDMA signal consists of two steps: Channelization or spreading code to reach channel rate of 3.84 h/; Scrambling to provide separation of transmissions. UMTS uses variable spreading and power levels to provide different user data rates. In FDD mode 10 frames are used. The number of chips per bits is called the Spreading Factor (SF) and it defines the data service required for the user: For UMTS: =! (2.1) $%& '(&) =3.84 *+,%-./. (/h 012) (2.2) The Spreading Factor (SF) can change in every 10 frame Table 2.2 Relationship between spreading factor and bit rate [15] Service Bearer Data Rate (Kbps) SF Modulation Rate (*+,%-./.) Speech Packet 64 Kbps Packet 384 Kbps Furthermore, in this section the structure of two physical channels in the downlink and random access channel in the uplink will be explained, the downlink dedicated physical channel (DPCH) and the physical downlink shared channel (PDSCH). Physical channels are typically structured into radio frames. A radio frame is a unit which consists of 15 slots. 37

38 The length of a radio frame corresponds to chips, which equals 10ms. Therefore one slot takes 2560 chips. Downlink Dedicated Physical Channel (DPCH) DPDCH DPCCH DPDCH DPCCH DATA 1 TPC TFCI DATA 2 PILOT 2560 chips, SF = Slot 0 Slot 1 Slot i Slot 14 1 radio frame = 10 ms Figure 2.8 Frame structure for downlink DPCH [14] Figure 2.8 shows the frame structure of the DPCH. The DPCH transmits Layer 2 user data and physical layer control information in a time multiplexed manner. The control information transmitted on the downlink Dedicated Physical Control Channel (DPCCH) consists of TPC, TFCI, and pilot bits. This control information is generated at Layer 1 and provides Transmit Power Control (TPC) and Transport Format Indication (TFCI). The pilot is used for channel estimation. The exact number of bits of each field may vary and it is fixed when the connection is established. The spreading factor ranges from 512 down to 4. Physical Downlink Shared Channel (PDSCH) Figure 2.9 is the frame structure of the PDSCH. A PDSCH is always associated with a downlink DPCH. So each user which shares a PDSCH requires an active DPCH. The PDSCH doesn't carry Layer 1 information, all this information is transmitted on the DPCCH part of the associated DPCH. Since the PDSCH is shared among several users, a user has to be informed that it should listen to the PDSCH. This is done by the TFCI field 38

39 in the associated DPCCH. Power control for both DPCH and PDSCH is performed by the TPC field. The spreading factor for this channel ranges from 256 to 4. PDSCH Data 2560 chips, SF=4 256 Slot 0 Slot 1 Slot i Slot 14 1 radio frame = 10 ms Figure 2.9 Frame structure for downlink PDSCH [14] Random Access Channel (RACH) The RACH is an uplink transport channel which is always received from the entire cell. RACH is characterized by a limited size data field, collision risk and the use of open loop power control. The Physical Random Access Channel (PRACH) is used to carry the RACH. Its operation is based on a Slotted ALOHA approach with fast acquisition indication [19]. The user equipment (UE) can start the transmission at a number of welldefined time offsets, which are denoted as access slots. There are 15 access slots per two frames, and they are spaced 5120 chips apart, i.e ms. Information on what access slots are available in the current cell is given by higher layers. The structure of the randomaccess transmission is shown in figure The random access transmission consists of one or several preambles of length 4096 chips, which is 1 ms, and a message of length 10 ms. The preamble part of the random-access burst consists of 256 repetitions of a signature. There are a total of 16 different signatures, based on the Hadamard code set of length 16. The 10 ms message is split into 15 slots, each of length 345 = 2560 chips. Each slot consists of two parts, a data part that carries Layer 2 information, and a control part that 39

40 carries Layer 1 control information. The data and control parts are usually transmitted in parallel. Data Data N data bit Control Pilot N pilot bits TFCI N TFCI bits T slot = ms, 10 2 k bits (0..3) Slot 0 Slot 1 Slot i Slot 14 T RACH = 10 ms Preamble 1 Preamble n Message sent T Preamble = 1 T Message = 10 ms ms Figure 2.10 Structure of the random-access transmission [19] The data part consists of 10*2 k bits, where k=0,1,2,3. This corresponds to a spreading factor of 256, 128, 64 and 32, respectively, for the message data part. The control part consists of 8 known pilot bits to support channel estimation for coherent detection and 2 TFCI bits. This corresponds to a spreading factor of 256 for the message control part. The total number of TFCI bits in the random-access message is 15*2 = 30. The TFCI value corresponds to a certain transport format of the current random-access message. 40

41 2.7.2 Medium Access Control (MAC) layer The main role of the transmission network is to transport MAC frames between RNCs and Node Bs [20]. The UTRAN MAC is not the same protocol as the GPRS MAC, even though they both have similar names and handle similar tasks in similar ways. The UTRAN MAC can even contain different functionalities depending on whether it supports FDD, TDD, or both modes. The Medium Access Control (MAC) layer is responsible for the handling of the logic channels and most of the priority and multiplexing issues [21]. The MAC layer is also responsible for selecting an appropriate transport format (TF) for each transport channel depending on the instantaneous source rate(s) of the logical channels. The transport format is selected with respect to the transport format combination set (TFCS) which is defined by the admission control for each connection. Figure 2.11 shows the MAC layer architecture. Figure 2.11 MAC layer architecture [9] The MAC layer consists of three logical entities: MAC-b which handles the broadcast channel (BCH). There is one MAC-b entity in each UE and one MAC-b in the UTRAN (located in Node B) for each cell. MAC-c/sh which handles the common channels and shared channels paging channel (PCH), forward link access channel (FACH), random access channel 41

42 (RACH), uplink Common Packet Channel (CPCH) and Downlink Shared Channel (DSCH). There is one MAC-c/sh entity in each UE that is using shared channel(s) and one MAC-c/sh in the UTRAN (located in the controlling RNC) for each cell. MAC-d is responsible for handling dedicated channels (DCH) allocated to a UE in connected mode. There is one MAC-d entity in the UE and one MAC-d entity in the UTRAN (in the serving RNC) for each UE. MAC Logical Channels Control channels: Broadcast control channel (BCCH) Paging control channel (PCCH) Dedicated control channel (DCCH) Common control channel (CCCH) Traffic channels: Dedicated traffic channel (DTCH) Common traffic channel (CTCH) MAC Services The services MAC provides to the upper layers include the following: Data transfer; Reallocation of radio resources and MAC parameters; Reporting of measurements to RRC. MAC Functions MAC functions include the following: Mapping between logical channels and transport channels; Selection of the appropriate transport format for each transport channel depending on the instantaneous source rate; 42

43 Priority handling between data flows of one UE; Priority handling between UEs by means of dynamic scheduling; Identification of UEs on common transport channels; Multiplexing/demultiplexing of higher-layer PDUs into/from transport blocks delivered to/from the physical layer on common transport channels; Multiplexing/demultiplexing of higher-layer PDUs into/from transport block sets delivered to/from the physical layer on dedicated transport channels; Traffic-volume monitoring; Transport-channel type switching; Ciphering for transparent RLC; Access service class selection for RACH and CPCH transmission Radio Link Control (RLC) protocol The Radio Link Control sub-layer is located in both the UE and the RNC immediately above the MAC sub-layer according to the radio interface protocol architecture. In the control plane, it provides services directly to layer 3, while in the user plane it may also provide services to the PDCP and BMC sub-layers [22]. This layer provides three different transfer modes to higher layer data flows: transparent mode (TM), unacknowledged mode (UM) and acknowledged mode (AM). Each mode is associated with a different Service Access Point (SAP) for upper layers, denoted as TM-SAP, UM-SAP and AM-SAP, respectively, and with different RLC entities, as shown in Figure All these three modes provide buffering of higher layer messages. The TR and the UM entity have separated transmitting and receiving entities. The AM entity is realized as one combined transmitting and receiving entity due to retransmission management. 43

44 Figure 2.12 RLC sub-layer architecture [16] In general, the RLC layer is in charge of the actual data packet (containing either control or user data) transmission over the air interface. It makes sure that the data to be sent over the radio interface is packed into suitably sized packets. The RLC task maintains a retransmission buffer, performs ciphering, and routes the incoming data packets to the right destination task (RRC, BMC, PDCP, or voice codec) Packet Data Convergence Protocol (PDCP) Packet Data Convergence Protocol (PDCP) sub layer is standardized in [23]. The Packet Data Convergence Protocol only exists in the user plane and is specifically for Packet Switched services. Its main functionality is to improve the efficiency in the radio transmission by means of executing header compression of the IP data packets coming from upper layers. UMTS supports several network layer protocols providing protocol transparency for the users of the service. At the moment, IPv4 and IPv6 are supported. Introduction of new network layer protocols to be transferred over UTRAN shall be possible without any changes to UTRAN protocols. Therefore all functions related to transfer packets from higher layers shall be carried out in a transparent way by the UTRAN 44

45 network entities. One task of the PDCP layer is this transparent transmission. Therefore, the functions the PDCP shall perform include the following: Header compression and decompression of IP data streams; Transfer of user data; Maintenance of PDCP sequence numbering Radio Resource Control (RRC) layer The Radio Resource Control (RRC) layer standardized within [24], handles the control plane signaling of layer 3 between the UE's and UTRAN. As shown in Figure 2.5, Radio Resource Control is attached to all logical channels that transfer control information. Further the RRC layer is connected to all entities within the UTRAN in order to exchange signaling information. The RRC protocol handles a large number of signaling tasks. The functions of RRC are as follows: Broadcast of information related to the non-access stratum (Core Network) Broadcast of information related to the access stratum Establishment, maintenance and release of an RRC connection between the UE and UTRAN Establishment, reconfiguration and release of Radio Bearers Assignment, reconfiguration and release of radio resources for the RRC connection RRC connection mobility functions Control of requested Quality of Service ( ) User Equipment measurement reporting and control of the reporting Outer loop power control Control of ciphering Paging Initial cell selection and cell re-selection Arbitration of radio resources on uplink Dedicated Channel (DCH) Timing advance (Time Division Duplex mode) 45

46 2.8 Radio Resource Management (RRM) Radio Resource Management (RRM) techniques are used to improve the utilization of radio resources of the wireless network [25]. RRM operations include essential functions like admission control, congestion control, power control, handover management, radio resource allocation and transmission parameters management [26]. The main theme behind the UMTS is to deliver the multimedia services characterized by stringent real time requirements, great sensitivity to delivery delay and packet loss and the need for considerable wireless resources. There are four basic classes of service in UMTS for quality of service ( ) provisioning. These classes are: Conversational Class: This class is for the most delay sensitive traffic. This class is used for voice over IP, video conference or any type of real-time interactive traffics. The transfer delay and delay variation are very strict. However, there are loose requirements on error tolerance. Streaming Class: This class is used for real-time voice and video streaming applications. Because it is unidirectional, it does not have stringent transfer delay compared with the Conversational Class. However, a maximum bound on delay variation is given to this class. There is no strict upper limit for the packet loss rate. Interactive Class: This class is used for web browsing, database retrieval and any kind of human interaction with remote equipment s applications. A short response time is expected for interactivity thus the round trip delay time is important in this class. This class requires low bit error rate transport. Background Class: This class is reserved for most delay insensitive applications. This is because the destination does not have to accept data within a certain time limit. The class is mainly used for and database download. It requires low bit error rate transport. Table 2.3 summarizes the UMTS classes defined by 3GPP. 46

47 Table 2.3 Four UMTS service class [27] Service Class Class Description Example Application 67 requirement Conversational Preserve time relation between entities Conversation pattern Real time Voice over IP Video conferencing Interactive game Low jitter Low delay Streaming Interactive Background Preserve time relation between entities Unidirectional continuous stream Bounded response time Preserve the payload content Preserve the payload content Real time video FTP Still image Web browsing E-commerce Fax Low jitter Round trip delay time Low BER Low BER As it is clearly seen from the table above, all service class impose different quality of service requirements. So to maintain these requirements during communication, management of radio resources of network is necessary [25]. The main objectives of radio resource management are to: Maximize the performance of all users with coverage and capacity; Guarantee the quality of service for different applications; Maintained the planned coverage; Optimized the system capacity. Radio resource management (RRM) is divided into two phases as follows: Radio resource configuration: This is responsible for allocating the proper resources to new requests coming into the system as a result it will not cause network to become overloaded thus compromising stability of network. However the congestion might occur, thus affecting quality of service ( ) due to the mobility of users. 47

48 Radio resource re-configuration: This is responsible for re-allocating the resources within the network when load is building up or congestion starts to occur to maintain for different applications throughout the network. It should change overloaded system back to target system by rearranging the resource between various applications on the same network. Thus Radio Resource Reconfiguration is also very essential part of RRM and infect of UMTS Radio Resource Management (RRM) Function When taking into account the constraints imposed by the radio interface, Radio Resource Management functions are responsible for taking decisions regarding the setting of the different parameters influencing the air interface behavior [16]. The following elements have been identified to be responsible for taking decisions in RRM: The number of active users. The number of simultaneous users transmitting The corresponding transmission rates for each user. The transmitted power levels corresponding to every simultaneous user. Radio Resource Management schemes can also include a set of service control functions, which are categorized into network based functions and connection based functions [28]. Network based functions include admission control (AC), load control (LC), packet scheduler (PS) and resource manager (RM); whereas connection based functions include power control (PC) and handover control (HC). Network based functions: Admission control (AC) - Handles all new incoming traffic. Check whether new connection can be admitted to the system and generates parameters for it. Occurs when new connection is set up as well during handovers and bearer modification. 48

49 Load control (LC) - Manages situation when system load exceeds the threshold and some counter measures have to be taken to get system back to a feasible load. Packet scheduler (PS) - Handles all non-real time traffic, (packet data users). It decides when a packet transmission is initiated and the bit rate to be used. Resource Manager (RM) - Controller over logical resources in Base Transceiver Station (BTS) and Radio Network Controller (RNC) and reserves resources in terrestrial network. Connection based functions: Handover Control (HC) - Handles and makes the handover decisions. Controls the active set of Base Station (BS) of Mobile Station (MS). Power Control (PC) - Maintains radio link quality. Minimize and control the power used in radio interface. Figure 2.13 shows the radio resource management functions implementation on different areas of a UMTS network. Power Control UE Uu Power Control Load Control Node B lub Power Control Handover Control Admission Control Load Control Packet Scheduling RNC Figure 2.13 Location of RRM functions [29] 49

50 2.8.2 Radio Resource Management (RRM) Functions Interaction Radio Resource Management functions are highly interrelated and coupled as long as they are all influencing the air interface. Since the objectives of the Radio Resource Management scheme are to achieve acceptable levels for the user application traffic and to design an efficient radio resource utilization. Connection establish request Resource Manager Admission Control Packet Scheduling Power Control Handover Control Load Control Figure 2.14 Radio Resource Management Functions Interaction [27] In order to achieve an efficient utilization of radio resource, it is very important to clearly identify the requirements of services and the characteristics of user traffics, (table 2.3). Based on these, the overall performance can be improved by efficiently combining different Radio Resource Management functions. Admission Control (AC) is the function that handles all new incoming traffics and checks whether new connection requests can be admitted to the system subject to a set of admission criteria such as the requirements and the subscriber profile. The AC sends the most up to date system load information to the Load Control (LC) function, which monitors the load condition of the system. LC also provides system load information that will enable AC to decide whether to admit a connection request without violating the system load limit. When the system load exceeds 50

51 the threshold, LC may decide to release existing connections to other lightly loaded coverage areas or to divert the connection request to another lightly loaded coverage region or to borrow resource from other resource pools in order to accommodate new connections. Packet Scheduler (PS) decides when a packet is to be transmitted and the bit rate that is to be used based on the connections quality of service ( ) parameters provided by Admission Control. Resource Manager (RM) is responsible for the logical radio resource configuration and status, such as the available resources and codes. It reserves the proportion of the available radio resources according to the resource request from Admission Control for each connection. Power Control (PC) maintains the radio link quality to minimize and control the power used and to satisfy the target bit error rate (BER) and Signal to Interference Ratio (SIR) specified by the Admission Control. Lastly is the Handover Control (HC), whose function is to handle and make handover decisions. 2.9 Scheduling Schemes Packet scheduling is a very important aspect of radio resource management in packet switched wireless networks. It interacts with other RRM control functions in order to ensure that the user quality of service ( ) requirements are respected. The nature of a scheduling framework can greatly impacts the levels that can be provided in the system.based on dynamic changes in the network topology and different types of heterogeneous access networks, next-generation wireless networks must be able to support the multimedia communications of multiple requirements, and simultaneously ensure high system throughput and low transmission delay [30]. These require a scheduling technology of wireless networks with very high specific performance. There will be different kinds of users in wireless networks, which do have distinct quality of service demands. Some applications require certain characteristics from the assigned radio resources in order to work, while others are more insensitive.this arises the need for assigning resources in a smart way, to meet the requirements of the users and also to utilize 51

52 the available resources most efficiently. Scheduling schemes can be classified into two groups based on the type of applications they can support. They are: Best-effort applications: These applications don't require certain performance in order to work, they accept whatever resources the network assigns to them. For example, a file or a web page download of course would prefer high bandwidth and low end-to-end delay, but it also works with little resources. Scheduling schemes for serving best-effort applications are: First-In-First-Out scheduling scheme (FIFO), Weighted Round Robin (WRR) scheduling scheme, Priority scheduling scheme, Queue Length Dependent scheduling scheme and Channel State Dependent scheduling scheme. Guaranteed-service applications: These applications need a certain amount of resources in order to work well, e.g., interactive multimedia requires a certain bandwidth as well as a small round-trip delay. Scheduling scheme for serving guaranteed-service applications are: Earliest-Due-Date scheduling, Rate-Controlled scheduling First-In-First-Out Scheduling First-in-first-out (FIFO) is the simplest type of scheduling scheme. The incoming packets are placed in a single queue and are served in the order as they were received. This scheduling scheme shown in figure 2.15, requires very little computation and its behavior is very predictable, i.e. packet delay is a direct function of the queue size [31]. There are many undesirable properties related to this queuing policy, due to the simplistic nature. It is impossible to offer different services for different packet classes since all packets are inserted into the same queue. If an incoming flow suddenly becomes bursty, then it is possible for the entire buffer space to be filled by this single flow and other flows will not be serviced until the buffer is emptied. 52

53 Flow 1 S 2 c h FIFO queue Flow 2 3 e d u l Flow 3 e 1 r Figure 2.15 FIFO Scheduling [31] Weighted Round Robin Scheduling The simplest form of fair scheduling scheme is Round Robin. In a Round Robin scheduling scheme packets are stored in different classes. Packets of each class have the same chance to be transmitted in a scheduling period [27]. Therefore every user requires its own queue. The scheduler assigns the same amount of resources to all users successively in a cyclic manner, empty queues are skipped. If a user provides less data, the remaining part is shared among all the others. Simple Round Robin only can serve equal users. If users have different requirements, Weighted Round Robin (WRR) shall be applied. In WRR, packets are first classified into various service classes and then assigned to a queue that is specifically dedicated to that service class. Each of the queues is then serviced in a round robin (RR) order. The weight indicates how many packets have to be sent in each cycle from each queue [32]. The WRR scheduler doesn t take the size of the transmitted packets into account. As a result, it is difficult to predict the actual bandwidth that each queue obtains, but it ensures that all service classes have access to at least some configured amount of network bandwidth 53

54 Figure 2.16 Weight Round Robin Scheduling [33] Priority Scheduling In priority scheduling, packets are slotted into different queues according to their quality of service requirements. These queues have different priorities and packets in the higher priority queues have a higher chance to be transmitted than the packets in the lower priority queues. Priorities can either be assigned statically to services, or dynamically to single packets according to their delay and rate requirements. Even though this queuing strategy is a good way of providing differentiated service, it also has some shortcomings, like large continuous flow of high priority traffic into the queue, equals excessive delay, and perhaps even service starvation for lower priority packets [31]. Figure 2.17 Priority Queuing Scheduler [33] 54

55 2.9.4 Earliest-Due-Date Scheduling The Earliest Due Date (EDD) scheduling, also known as Earliest Deadline First (EDF) scheduling, is a classic example of a deadline-based scheme where packets are scheduled based on the earliest-deadline-first principle [34]. EDD was originally designed for serving individual flows, but it can also be applied to class based differentiation. Working with the assumption that the traffic arriving in each class is periodic and using 8 to denote the period for class, the EDD algorithm works simply as follows: upon arrival of the 9 packet of class at the router at time 1 :, the packet is stamped with a deadline i.e. the sum of its arrival time and period. < : = 1 : +8 (2.3) The packets are then served in the numerical order of their deadlines. Notice that, in reality, the arriving traffic is not periodic; the purpose of the period is only to describe the expected inter arrival time of packets Rate-Controlled Scheduling A rate-controlled scheduler consists of a regulator and a scheduler. The regulator is responsible for shaping the traffic of each service in order to guarantee conformance with the desired traffic pattern. Hence, the scheduler receives packets with a predefined rate. The scheduler then controls the transmission order of packets belonging to different services. With this approach it is possible to assign to one service simultaneously a lower bandwidth with higher delay requirements Requirements of a Scheduler There are some general desirable properties common to all scheduling disciplines. A scheduling discipline must satisfy four, sometimes contradictory, requirements: performance bounds, fairness and protection, flexibility and ease of implementation. 55

56 Performance bounds: This can be expressed deterministically or statistically. A deterministic bound holds for every packet. Whereas a statistical bound is either expressed as a mean value or a 95-percentile. This percentile expresses that the bound is met by 95% of all packets [14]. Deterministic bounds of course require more network resources to be reserved than statistical bounds. General performance parameters are bandwidth, delay, delay jitter, and packet loss due to transmission errors or full queues. Fairness and protection: Some notion of fairness is incorporated in many network mechanisms used today [35]. Fairness is a desirable property of scheduling algorithms serving equal services. If there are several classes of services, fairness should be provided within every class. Fairness means that resources are shared equally among all services which are ready to send. Since real network environment is not static [34], scheduling discipline should be able to protect and satisfy the performance requirements of well-behaved users, also in the presence of different sources of variability, such as best effort traffic, badly behaved users and network load fluctuations. Flexibility: The scheduling discipline must not optimize performance from a single application s point of view but should rather be able to accommodate applications with varying traffic characteristics and performance requirements. Ease of implementation: The scheduling algorithm has to do its decisions in real time, so the complexity of the algorithm determines the hardware requirements. Therefore it is necessary to compose fast and easy implementable algorithms in order to keep hardware requirements and time for computation low. Schedulers are capable of improving transmission for certain services. However, this improvement is at the expense of worse performance for other services. This fact is revealed by the conservation law. Consider a set of N connections at a scheduler. Traffic at arrives at a mean rate > and the mean service time for a packet from connection is?. 56

57 Then = >? is the mean utilization of the link due to connection. If the mean waiting time for packets of connection is denoted as A, then the conservation law becomes: C B- % F % =+7G.& %DE (H.I) For schedulers which are only idle if all queues are empty. Since this equation is independent of the scheduling discipline, reduction of the delay for a certain connection results in a higher delay of other connections. In WCDMA, packet scheduling algorithms can be done in two ways, in a time or code division manner. Time Division Scheduling One user is allocated a channel at a time (10 frame); All available capacity can be allocated to that user; High data rate for a short period of time; Increase more users, each user has to wait longer. Advantages of Time Division Scheduling High bit rate required less energy per bit; Less interference; Shorter delay due to high bit rate. Disadvantages of Time Division Scheduling High unused physical resources due to short transmission time and relatively long set up and release time; High variations in the interference levels due to high bit rate and bursty traffic; Limited uplink range of high bit rate due to mobile s limited transmission power. 57

58 Code Division Scheduling Many users are allocated the channels simultaneously; The capacity is shared with all users; Low data rate for a long period of time; Increase more users, each user s bit rate is decreased. Advantages of Code Division Scheduling Resources are in full usage due to longer transmission time; Small variation in interference level; Longer uplink range due to lower bit rate. Disadvantages of Code Division Scheduling Longer transmission delay due to low bit rate; High interference due to high energy per bit; Low total throughput Related works IP-based network entities integrated voice and data on unified IP backbone, which can increase the resource utilization over existing mobile networks. WCDMA radio access network must manipulate the delay-sensitive real-time packets to provide IP multimedia service. Several scheduling schemes have been proposed in the literature for IP-based radio access networks in WCDMA to efficiently utilize radio resources. In [36], a credit based scheduling algorithm is proposed for use in the forward link of a CDMA system. This algorithm dynamically assigns an OVSF code to a mobile user on a timeslot-by-timeslot basis. On connection set-up the mobile user negotiates with the network management module a guaranteed bandwidth, denoted as GBW. Throughout the duration of a connection, the base station keeps track of a priority variable called the credit C: / =221J KL MNO 221J_RST_U_VW8_192 U1V (2.5) 58

59 In this scheduling algorithm connections with more credits are scheduled to receive more packets. This type of credit-based prioritization does not provide low packet delays and differentiation. In [37], Delay Fair Scheduling (DFS) scheme is proposed. This scheme has low computational complexity and provides fair distribution of the available shared capacity to the connections according to their delay requirements. According to DFS each connection at the start of each frame n, is characterized by its priority X. Where, X = < (R 3 ) 0, R=0,1,2, (2.6) = threshold for the acceptable data packets delays, defined during connection setup, < = head-of-line packet delay for queue i and 3 = 10ms, is the scheduling period of the DFS scheme. Subsequently, the connections are sorted and served in decreasing order of their priorities. This scheduling scheme fail to utilize efficiently the scarce radio resource. The Delay Fair Scheduler with prediction (DFS_PRED) was proposed in [38]. The main idea of DFS_PRED is to prioritize the connections, not only according to their delay requirements but also according to their predicted error probability during the next frame. As with DFS each connection at the start of each frame n, is characterized by its priority X. X = < (R 3 ) \1 X LK ()] 0, R=0,1,2, (2.7) However, in this case the priority of a connection is proportional to its probability of successful transmission (1 X LK ()). Consequently, between two connections with the same head-of-line packet delay and delay threshold, the one with the higher probability for successful transmission will also have higher priority. Thus, DFS_PRED is able to encounter the variable capacity of the wireless interface better than DFS. DFS_PRED serves the connections not only according to their delay sensitivity, but also according to the predicted state of their wireless channel. The simulation results shows the efficiency of the scheme in terms of average packet delay and bandwidth utilization. 59

60 The scheduling schemes presented in [39,40] are also developed for WCDMA systems. These schemes have a scheduling discipline that resembles to the Wireless Fair Queuing as they directly assign to each connection a guaranteed bandwidth 0`, instead of a weight that corresponds to a fraction of the available capacity. Furthermore, they use a priority variable called the credit /, which is the difference between the error-free service (221J_2 0`, ) and the actual number of packets a, that each flow has received so far. Therefore, the credit / of flow is defined as follows: / =\221J_2 0`, ] a, (2.8) Connections with more credits are scheduled to receive more packets. Assuming perfect power control, credit based prioritization provides, in the long run, a data rate guarantee to each accepted connection. The above mention credit based schemes supporting multimedia traffic either do not consider channel condition or fail to address the exact code position in the code tree, which may result in inefficiency in resource utilization. A credit-based scheduler which considers channel condition and explores the concept of compensation codes is proposed in [41]. With this channel-sensitive scheduling algorithm, a user with more credits will have more chance to transmit without compromising to the transmission quality. Simulation results justify that this scheme work as claim. Many of the proposed schemes take into account the throughput and fairness. However, the environment, which were considered in those schemes, is that, at each frame, the base station can transmit to at most one mobile node on a separate spectrum. This environment is very different from the UMTS network in which all the mobiles share the same spectrum. Sallents et al. [42] proposed a packet scheduling algorithm, the real time emulator (RTE) with transmission power constraint for UMTS. This scheduler serves the packet based on the priority value, but the fairness property was not considered. Thus, the amount of service time allocated to the ill-behaved users can be more than that of the well-behaved users. In [43], a scheduling scheme that select packet based on the value of the service credit is proposed. This scheme allocates the data rate based on the time scheduling strategy or the 60

61 code scheduling strategy. Using this scheduling algorithm, the soft can be guaranteed. However, this scheduling algorithm has an undesirable effect that the backlogged flow can be starved for an arbitrary period of time as a result of excess bandwidth it received from the server when other flows are idle. Min-Xiou Chen, Ren-Hung Hwang in [44], proposed two scheduling algorithms, Multiflows Worst-case fair Weighted Fair Queuing plus (MWF 2 Q+) and Multi-flow Deficit Round-Robin (MDRR), for multiple classes of service over the same spectrum in the forward link of the UMTS network. MWF 2 Q+ and MDRR, are based on the WF 2 Q+ and DRR algorithms, respectively. These two algorithms are well studied in wired broadband networks. The WF 2 Q+ is known for its excellent fairness properties and DRR is known for its low computational complexity while maintaining reasonable fairness properties. The algorithm of MWF 2 Q+ In MWF 2 Q+ scheduler, each time a packet arrives or a packet of backlogged flows gets served, the virtual time will be updated. The MWF 2 Q+ scheduler is based on the Generalized Processor Sharing scheduling (GPS) discipline [45]. In MWF 2 Q+ scheduler, if all flows in U are all in backlogged mode (have data to send in their sending queues), the data rate allocated to flow is, b / Kcd (2.9) f h b f Where / is the system capacity. As long as b 1, f h flow can be guaranteed with a minimum rate of, V =b / (2.10) V is the minimum guaranteed rate for flow, and can be any positive number. In MWF 2 Q+, j (2) is determine based on the virtual time of flow. j (2) denotes the amount of data served at frame 2 for flow. For the UMTS system, a new constraint should be added, that is, j (2) {special rates provided in UMTS}. The virtual time of flow can be derived from the GPS discipline. If : is the 9 packet of flow. The arrival time and 61

62 packet length of the packet are denoted as k( : : ) and j respectively. If \X : ] and l\x : ] also represent the virtual start tag (time) and virtual finish tag (time) of the packet, respectively. Then, m(2) becomes the system virtual time. \X : ], l(x : ) and m(2) are defined as follows: \X : l\x :op ], U [k(x : ) o ] 0 ]=n max {l\x :op ],m[k\x : ]]}, U [k(x : ) o ]=0 ( 2.11) l\x : ]= \X : ]+ j : V (2.12) m(2+y)=max {m(2)+o(2,2+y), min () [ \X () ]]} (2.13) Where [k( : ) o ] and [(2 1) o ] are the queue length of flow at the time just before k( : ) and the (2 1)2h frame, respectively. O(2,2+y) is the total amount of service during the period [2,2+y], N(2) is the set of backlogged flow at time 2, and X () is the packet at the head of flow queue at time2. At initial, m(2)=0. In WF 2 Q+, only one flow at a time transmits a packet and that packet must be transmitted without interruption. The system serves those packets based on the increasing order of their virtual time. In UMTS, multiple flows can transmit their packets simultaneously on the same spectrum. Furthermore, the definition of the virtual time for each flow in MWF 2 Q+ is very different to that in WF2Q+. The virtual time of flow,m (2) is further define as follows: m (2)=~ m (2 1)+ (op), U [(2 1) o ] 0 \X : ], U [(2 1) o ]=0 1R8 [(2) o ] 0 (2.14) 62

63 Assuming ƒ denotes the data rate of layer in the OVSF code tree, and.ƒ % denotes the minimum data rate for flow % in the UMTS, and.ƒ % can be defined as.ƒ % = { ƒ % ƒ % {Special rates provided in UMTS}, and ƒ % ƒ % }. As long as % ƒ % Šand % $(&).ƒ % <Š. The residual bandwidth (RB), should be distributed to the backlogged flows in a fair and equitable manner. Where, 0N=1?S Œ2 112Œ B () V (2.15) In [46], an experimental study of a new scheduling policy for achieving fairness,, and optimal use of resources is proposed. The proposed scheduling policy Code Division Multiple Access based on Generalized Processor Sharing with Dynamic Weights (CDMA/GPS-DW) is an improvement of a previous GPS policy. This scheme utilizes dynamic weights for bandwidth assignment, the weights are calculated as a function of the number of active Mobil Terminals (MTs). Simulation results show that the proposed policy achieves fairness of the specified and makes efficient use of the network resources. This scheme is not flexible in the traffic management, that is, it is restricted only for three traffic types. In addition, admission control is not considered and it is designed without considering the effect of multipath fading in the cell. Mendez, Panduro, Covarrubias and Romero [47] proposed a rate scheduling scheme which is based on GPS. This scheme treats different traffic type according to their quality of service requirements in the uplink of CDMA cellular network. Multiple traffic with variable traffic rate can be served simultaneously. This is consider a drawback of GPS because the classical packet-based systems are based in TDMA. With this scheme, the flexibility in bandwidth allocation of CDMA system is exploited. The analysis and simulation results shows that this scheme is an improvement of the CDMA-GPS in [46]. The improvement is based on the dynamic weight assignment of bandwidth allocation for each type of service provided (e.g., video, voice and WWW-data), as compared to the static weight assignment of bandwidth allocation utilized by CDMA-GPS scheme. 63

64 Skoutas and Rouskas [48] proposed a Dynamic Priority Allocation scheduling algorithm, which is designed to operate within a cross-layer framework that provides Dynamic Priority Allocation (DPA) with the necessary information in order to take into account the variations of the wireless channel. The proposed scheme is designed for provisioning in the Downlink-Shared Channel (DSCH) in WCDMA 3G systems. The differentiation between connections is based on their delay sensitivity and head-of-line (HOL) packet delay. The DPA scheme has low computational complexity and provides fair distribution of the available DSCH capacity to the connections. By providing a guaranteed rate per traffic flow at each scheduling period, DPA is able to offer a deterministic delay bound to each session when the transmission is constantly reliable and a stochastic delay bound for identical DSCH connections with certain constraints. Simulation results demonstrate Dynamic Priority Allocation (DPA) fairness property and its efficiency. Wigard, Madsen and Gutiérrez [49] proposed a packet scheduling algorithm, that can differentiate the among user and service classes in WCDMA. At the same time a parameter is introduced, which gives the possibility of adjusting the packet scheduling algorithm from signal to interference ratio (C/I) based scheduling to inverse signal to interference ratio (C/I). The algorithm can be tuned from signal to interference ratio (C/I) based scheduling to Round Robin and beyond. The algorithm can upgrade the priority of the users in the queue, in order to avoid unacceptable delays for low priority users. This part can be used to cut the tails of the packet call delay distributions. Retransmissions are given a relatively high priority in order to avoid timeout and unnecessary retransmissions caused by higher layer protocols. Wan, Shih and Chang [50] proposed three real-time scheduling algorithms to support quality-of-service at IP-based radio access networks for the UMTS. The real-time generic scheduling (RTGS) algorithm applies the functionalities of the radio management framework to establish new data sessions for real-time service requests. The real-time bandwidth scheduling (RTBS) algorithm implements the early-deadline-first (EDF) 64

65 scheme to do the schedulability analysis and to schedule the data sessions to reduce power consumption. The real-time code scheduling (RTCS) algorithm (RTCS) applies Dynamic Code Assignment (DCA) scheme to improve radio resource utilization. Experimental results show that, under various traffic loads, RTCS performs best in terms of power consumption, session drop rate and bandwidth utilization. It also shows that RTBS outperforms RTGS. Chandramathi, Raghuram, Srinivas and Singh [51] proposed a fuzzy logic (FL)-based dynamic bandwidth allocation algorithm for multimedia services with multiple quality of service : Probability of blocking (PB), Service access delay (SAD), Access delay variation (ADV) and the arrival rate requirements. In this algorithm, each service can declare a range of acceptable quality of service levels (e.g. high, medium, and low). As quality of service demand varies, the proposed algorithm allocates the best possible bandwidth to each of the services. This maximizes the utilization and fair distribution of resources. Simulation results show that the required quality of service can be obtained by appropriately tuning the fuzzy logic controller (FLC). Xu, Shen and Mark [52] proposed a code-division generalized processor sharing (CDGPS) fair scheduling dynamic bandwidth allocation (DBA) scheme for WCDMA systems. The scheme exploits the capability of the WCDMA physical layer by allowing channel rates to be dynamically and fairly scheduled by varying the spreading factor and/or using multiple code channels, rather than allocating service time to each packet. Analysis and simulation results of the model shows that bounded delay can be provisioned for real-time application by using generalized processor sharing (GPS) service discipline, while high utilization of system resources is achieved. Gürbüz and Owen [53] proposed Dynamic Resource Scheduling (DRS) Scheme as a framework that will provide quality of service provisioning for multimedia traffic in W- CDMA systems. This scheme is an extended DRS family that is aimed at examining the temporal quality of service in terms of delays. The DRS framework monitors the traffic 65

66 variations and adjusts the transmission powers of users in an optimal manner to accommodate different service classes efficiently. Variable and optimal power allocation is also suggested to provide error requirements and maximize capacity, while prioritized queuing is introduced to provision delay bounds. Simulations of this scheme shows that the delay performance can be provisioned for guaranteed services by multiple queues. Xu, Shen and Mark [54] proposed a credit-based CDGPS (C-CDGPS) scheme to further improve the utilization of the soft capacity by trading off the short-term fairness. With the C-CDGPS scheme, the soft uplink capacity is allocated by using a combination of creditbased scheduling and CDGPS fair scheduling. The model considered a frequency division duplex (FDD) Wideband DS-CDMA network supporting a large number of multimedia users. Packetized multimedia traffic is considered. Simulation results shows that bounded delays, increased throughput, and long-term fairness can be achieved for both homogeneous and heterogeneous traffic. Xu, Shen and Mark [55] proposed a dynamic fair resource allocation scheme to efficiently support real-time and non-real-time multimedia traffic with guaranteed statistical quality of service (QoS) in the uplink of a wideband code-division multiple access (CDMA) cellular network. The scheme provides a trade-off between the Generalized Processor Sharing (GPS) fairness and efficiency in resource allocation to maximize the radio resource utilization under the fairness and quality of service constraints. Analysis and simulation results show that, in a multipath fading environment, the proposed scheme can reduce the inter-cell interference, increase the network capacity, guarantee a statistical delay bound for real-time traffic and a statistical fairness bound for non-real-time users. Salman [56] proposed a Multi-operators Code Division Generalized Processor sharing (M- CDGPS) scheme for supporting Multiservice in the uplink of WCDMA cellular networks with multi-operators. The scheme employs both adaptive rate allocation to maximize the resource utilization and Generalized Processor Sharing (GPS) techniques to provide fair services for each operator. The simulation results show that the proposed scheme improve 66

67 both system utilization and average delays. The proposed scheme allows for a flexible trade-off between the GPS fairness and efficiency in resource allocation and is an effective way to maximize the radio resource utilization under the fairness and QoS constraints Conclusion Universal Mobile Telecommunication System (UMTS) quality of service (QoS) architecture has been summarized in this chapter. This QoS architecture is used for designing the Radio Resource Management (RRM) scheme for the UMTS system under consideration in this research work. Radio resource management service functions, such as power control, admission control, load control, handover control and packet scheduling are also presented. Various algorithms, strategies and classification for the RRM frameworks have been reviewed in order to provide the background knowledge for the design of resource allocation scheme for the UMTS compatible system. From the related works above on resource allocation techniques, the classes of techniques based on the Generalized Processor Sharing (GPS) provide more flexibility in bandwidth allocation as they ensure fairness while dynamically allocating resources. But, in all the techniques outline in section 2.11, the issue of backlogged flow loss rate has not been well addressed. In the next chapter, the Code-Division Generalized Processor Sharing (CDGPS) scheme will be discussed in detail. 67

68 CHAPTER THREE RESEARCH METHODOLOGY 3.1 System Model This section gives the design of the scheduler functional architecture for providing quality of service requirements, while also achieving efficient utilization of radio resources. A code-division generalized processor sharing (CDGPS) is proposed for WCDMA systems, to support differentiated quality of service (QoS) with a central controller that can dynamically allocate bandwidth to mobile users according to the variation of channel condition and traffic load. The CDGPS scheduler makes use of both the traffic characteristics in the link layer and the adaptivity of the WCDMA physical layer to achieve efficient utilization of radio resources. It adjusts only the channel rate (service rate) of each traffic flow in the WCDMA system by varying the spreading factor and/or using a multiple of orthogonal code channels, rather than allocating service time to each packet. This results in lower implementation complexity of the CDGPS scheme than for a conventional GPSbased time scheduling scheme. The system model considered in this work, is the frequency division duplex UMTS cellular network (UMTS-FDD) where user equipment (UE) are interconnected with the Internet through Node B, Radio Network Controller (RNC) and core network, as shown in figure 3.1. The radio link in the UMTS-FDD system can be characterized by orthogonal channels in the downlink (from Node B to UE) and multiple access channels in the uplink (from UE to Node B). A pair of bandwidth schedulers are assumed to reside in each Node B. The schedulers allocate the power and respective rate of the channels in the downlink and uplink to all UEs in the same cell covered by Node B. Although the capacity of the downlink is equal to the uplink capacity, the discussion in this chapter only focuses on the uplink. 68

69 UE Wireless backbone (Internet) UE RNC Core Network Node B UE RNC Downlink Scheduler Node B Uplink Scheduler Node B Radio Link UE UE UE UE Figure 3.1 Network Structure In this work, the network architecture illustrated in Figure. 3.1 was employed. The physical data channels in the uplink comprises a small number of random access channels and a large number of dedicated channels. Each mobile user is assigned to at least one dedicated data channel, and shares the random access channels with other users in the same cell. While signaling and short messages may be transmitted freely through the random channels, most IP data flows are scheduled for transmission on the dedicated channels. Multimedia IP traffic (e.g., voice, video, and data) are supported by this network. In a multimedia IP traffic, the quality of service requirements general consist of two parts: Delay Loss rate 69

70 3.2 Generalized Processor Sharing (GPS) A Generalized Processor sharing (GPS) server is work conserving server, which implies, the server must be busy if there are packets waiting in the system. GPS sever also operates at a fixed rate V, and is characterized by positive real number b p,b,.b. These numbers denote the relative amount of service in each session, that is, if (y,2) is defined as the amount of session traffic served during an interval (y,2), a session is backlogged at a time 2 if a positive amount of that session s traffic is queued at that time 2. Then a GPS server is defined as one for which the following inequality holds for any session that is continuously backlogged in any interval (y,2) [45], equation 3.1 will holds with equalities, and the allocated bandwidth of each user is exactly proportional to its weight. (y, 2) f(y,2) b b f, =1,2,. (3.1) A basic principle of GPS is to assign a fixed positive real number (namely weight), instead of a fixed bandwidth, to each flow, and to dynamically allocate bandwidth for all flows according to their weights and traffic load. Due to the burstiness of packet traffic, sometimes a user may not have packets to transmit and gives up its bandwidth for a while. The excess bandwidth can be distributed among all backlogged sessions at each instant in proportion to their individual weight b. This makes the GPS server efficient and fair in bandwidth allocation. 3.3 The Code-Division Generalized Processor Sharing (CDGPS) scheme The proposed scheme in this research work is the code-division generalized processor sharing (CDGPS) fair scheduling scheme. This scheme uses the GPS fair scheduling discipline to dynamically allocate channel rates. The model of figure 3.2 comprise of a server of capacity, C Mbps. The input traffic is from varied sources comprising of voice, video, and data traffic, which are bundled into flow classes. Each flow maintains a 70

71 connection with link rate / (9) during the 9 time slot. The sum of / (9) over all users should not exceed C. Flow 1 Flow 2 Flow N S c h e d u l e r Connection 1 (/ p (9)) Connection 2 (/ (9)) Connection N (/ (9)) C Radio Link (Capacity = C) Figure 3.2 A queuing model of the CDGPS scheme For each slot, the scheduler allocates adequate service rates to the N flows, using the following scheduling procedure: Let the pre-assigned weight for flow be b, =1,2,, and (9) to denote the amount of session traffic that would be served during time slot k. According to the GPS scheduling discipline, Eq. (3.1) should hold for any flow that is continuously backlogged in the time slot k. Then, the proposed CDGPS server allocates each / (9) using the following steps: Step 1: Let N (9) be the total amount of backlogged traffic of flow during time slot k. Estimate N (9), =1,2,., as follows: Where y : is the end time of slot (9 1), (y : )= Backlogged traffic at time y :, N (9)= (y : )+V (9) (3.2) V (9)= Estimated traffic arrival rate of flow during time slot k 71

72 The estimated traffic arrival rate V (9) of flow during time slot 9 can be estimated from past traffic measurement using the following two approaches: 1. One-step estimation the estimated traffic arrival rate is thus: V (9)= 1 (9 1) (3.3) Where 1 (9 1) is the total amount of the arrival traffic (in bits) during time slot (9 1). 2. Exponential averaging Let 2 and J be the arrival time and length of the R packet of flow, respectively. The estimated rate of flow,v, is updated every time a new packet arrives: V L =\1 oa ] j +oa 54š V (3.4) Where =2 2 op and is a constant. An approximate value for would be between 100 and 500. = Inter packet arrival time. Step 2: Based on the estimated N (9),=1,2,.,, the expected amount of service (9) received by th user is determine thus: (9)=œ 0, U N (9)=0, U N (9)>0 where is the scheduling period in CDGPS scheme, (9)= b / (3.5) fdp b f is the minimum guaranteed rate of flow and C is the network capacity. If (9)</, Dp then the remaining network resource is distributed proportionally to users who expect more than their guaranteed service rate. The distribution of the remaining network resources should be in proportion to each 72

73 user s weight b according to the GPS service discipline. The allocated channel rate to user can then be determined by equation (3.6). / (9)= (9) (3.6) The CDGPS scheme weights are related as (b p = b Ÿ 2 =b Ÿ 3 ), that is, different priority values, with b b b p, where b corresponds to highest priority and b p corresponds to the lowest priority [52]. These values (1, 1/2, 1/3) do not guarantee the maximum data transmission under UMTS platform (384 Kbps). Therefore, a different set of values (1/5, 1/3, 1/2) that better utilize the available bandwidth in UMTS is presented. Where 1/5 corresponds to the lowest priority and 1/2 corresponds to the highest priority. The priority CDGPS and non-priority CDGPS scheme flowchart are depicted in figure 3.3, and figure 3.4 respectively. Both figures 3.3 and 3.4 illustrate the operation of the proposed scheduling scheme. The only difference between priority and non-priority CDGPS flowcharts is the different ways in which the backlogged flows N (9), are sorted. For priority CDGPS, the backlogged flows are sorted by decreasing order of weight b, while for non-priority CDGPS, the backlogged flows are sorted by first-come-first-serve (FCFS) order of weight b. After the sorting operation, the first flow from the sorting list is removed. If the total amount of traffic of flow is greater than zero and the condition in equation 3.2 satisfied, then the estimated rate of flow is updated every time a new packet arrives, otherwise the sorting list is empty. The expected amount of service (9) by 2h user is received based on the estimated amount of traffic of flow. If the sorting list is empty and the total amount of traffic of flow is equal to zero, then the amount of service received by 2h user becomes zero. 73

74 1. Sort the backlog flow N (9) by decreasing order of weight b Remove the first flow from the sorting list If N (9) > 0, and N (9) = (y 9 )+V (9)? No Yes Sorting list not empty? Yes No Update the estimated rate V (9) of flow, V (9) = 1 (9 1)/ If N (9) = 0? No 1 Yes 3. Assign (9) =, = b / =1b Then, (9) = 0 Flow i is not backlogged If B (9) < /? Yes Repeat step 2 and 3 No Idle Flow generation Active 2 Figure 3.3 Priority CDGPS flowchart 74

75 1. Sort the backlog flow 2 N (9) by first-come-firstserve order of weight b 1 2. Remove the first flow from the sorting list If N (9) > 0, and N (9)= (y 9 )+V (9)? No Yes Sorting list not empty? Yes No Update the estimated rate V (9) of flow, V (9) = 1 (9 1)/ If N (9) = 0? No 1 Yes 3. Assign (9) =, = b/ =1b Then, (9) = 0 Flow is not backlogged If B (9) < /? Yes Repeat 2 and 3 No Idle Flow generation Active 2 Figure 3.4 Non-priority CDGPS flowchart 75

76 3.4 Traffic Source Model In traffic source model, while the characterization of voice users is fairly straight forward, the traffic generated by packet data users is highly dependent on the application and has a high degree of burstiness [57]. Multimedia traffic is very bursty in nature and simple models as the Poisson process do not capture the important characteristics of this sources. To model bursty traffic sources different approaches are available [58], many of them using Markov modulated processes (MMP). These are doubly stochastic processes in which each state of N states of embedded Markov chain originates another stochastic process. If this originated process is a Poisson process the MMP is called Markov Modulated Poisson Process (MMPP), if it is deterministic it is a Markov Modulated Deterministic Process (MMDP) Voice Source Modeling In this work, the voice sources are simulated using the ON OFF model for a single source and aggregating many such sources. This is a Markov Modulated Deterministic Process with only two state as shown in figure 3.5. α OFF ON β m! Figure 3.5 On-Off Model 76

77 In the voice on-off model, the duration times of burst and silent period is exponentially distributed with mean 2 1/ and 2 1/ respectively. Some studies have proposed and [59], setting the transition rates to: R8, The voice source generation follows exponential inter-arrival times t with constant arrival rate of 16 Kbps. The voice activity factor 2 is assumed to be 0.4 sec. In the active state packets are generated with a constant speed m j 2, with j as the packet length and 2 as the packet inter-arrival time. This is depicted in figure =0.4 sec 2 +2 = Figure 3.6 On-Off voice packetization Time Video Source Modeling An accurate traffic model of VBR video is necessary for evaluating the performance of a network design. A major component of multimedia networking is the data compression (source coding) of multimedia data sources i.e. speech, audio, image and video. The process of reducing the amount of data required to represent a digital video signal, prior to transmission or storage is called video compression or video encoding [60]. Once the data is compressed, the bit stream is packetized and sent over the Internet. In this work the variable bit rate (VBR) video sources are simulated by Markov Modulated Poisson Process 77

78 (MMPP). Each video source is governed by an m-state Markov chain with probability transition matrix P = f, where, f number of transitions from state to state number of transitions out of state (3.7) Where, =0,1,2,, 1, when a source is in state, it generates rate,0 K T. The average duration in each state and/or the length of video transmissions is assumed to be exponentially distributed with a mean chosen to be 40 ms, which is equivalent to the length of one frame of the video sequence with a frame rate of 25 frames/s Data Source Modeling The data sources are simulated using the ON OFF model for a single source and aggregating many such sources. The ON-OFF model implies, there will be only 2-state model (source silent period or source producing data at a chosen peak rate). Data source generation too follows exponential inter-arrival times, but unlike the case of voice, where bit rate is constant, for data source it varies randomly. A Poisson process is used to generate the data traffic with packet size 2560 bits and average arrival rate 256 kb/s. 78

79 Table 3.1 Simulation parameters Parameter Value Radio access mode WCDMA (FDD) uplink Chip rate 3.84 Mcps Spread spectrum 5.0 MHz WCDMA channel rate 2.0 Mbps Slot duration ms Frame duration 10 ms Voice source rate 16 Kbps VBR video source rate 16 to 384 Kbps Data source rate 256 Kbps Voice active factor 0.4 Packet arrival Poisson Packet generation type Exponential Queue type FIFO TTI Model Validation The proposed CDGPS model is validated using the work in [54]. Both works are related because, they made use of dynamic bandwidth allocation mechanism to serve each flow or user according to their estimated arrival rate in the uplink direction. Figure 3.7 shows the throughput comparison of the proposed CDGPS and Liang Xu CDGPS. The comparison shows that, the proposed model attained an average performance level of 97.9% to that of Liang Xu CDGPS. 79

80 THROUGHPUT Proposed CDGPS Liang Xu CDGPS TRAFFIC INTENSITY Figure 3.7 Model validation with CDGPS scheme 3.6 Conclusion This chapter has investigated the general system model requirements for achieving the dynamic radio resources allocation. The proposed CDGPS scheduling algorithm, presents a different set of priority values that better utilize the available bandwidth in UMTS. The proposed CDGPS has been implemented with MATLAB and validated. 80

81 CHAPTER FOUR SIMULATION AND RESULT ANALYSIS 4.1 MATLAB Simulation Framework MATLAB modeler version R2010b is chosen as the simulation method to evaluate the performance of the designed dynamic bandwidth allocation algorithms for the WCDMA system. The computer simulation model in figure 4.1 was simulated in the Simulink environment to evaluate the performances of the dynamic scheduling algorithms for multimedia IP traffic (e.g. voice, video and data). Figure 4.1 is the entire system model and is summarized thus; the model simply consist of traffic sources, buffer, UMTS server and the scheduler computational model. Each source block is an aggregation of voice, video and data traffic source with different quality of service requirements. These source blocks are served on a first-in-first-out basis in the buffer block, which queue is serviced by the UMTS server at dynamic rate for all the source distribution. The path-combiner was employed in the simulation model in order to ensure that the arriving traffic has equal probability of being served in a random manner. The system server block uses a service discipline of first-in-first-out (FIFO) to service voice, video, or data traffic source at a service rate based on their individual quality of service requirements. The computational model block does the actual scheduling and bandwidth allocation processes and also ensuring that, the scarce resource is utilized efficiently. 81

82 Figure 4.1 Simulation Framework for WCDMA systems 82

83 Figure 4.2 Multimedia IP traffic (voice, video and data) Markov-modulated Poisson process typically models the ON-OFF traffic pattern for the services supported by a UMTS network, as shown in figure 4.2. Services such as voice, video and data are supported. Each service is model and generates its entities using a Markov-modulated Poisson process, whose rate depends on the state of the Markov chain. The model in figure 4.2 includes three independent modulated Markov sources whose behavior depends on the rate of the Poisson process when the Markov chain is in the ON state. The Path Combiner block aggregates the output of all the On-Off Modulated Markov Source subsystems. 83

84 Figure 4.3 Buffer queuing Model Figure 4.3 showing the buffer queuing model. This model was developed using a FIFO buffer, were entities are served every one second observation period through an entity departure counter that is triggered by a repeating sequence block. These entities are stored in the FIFO Queue block, released and translated back into a signal by the Get Attribute block. A FIFO buffer stores data as a part of data exchange between processors. The FIFO Queue block enables simulation of such buffers in software. Each processor is driven by a clock. Because each clock synchronizes the processor hardware, the clock appears synchronous to that processor. 84

85 Figure 4.4 The system Server The single server block as shown in figure 4.4 serves one entity for a period of time, and then attempts to output the entity through the OUT port. The server uses first-in-first-out service discipline to service voice, video, or data traffic at a service rate that is based on the individual quality of service of each service class. The Embedded MATLAB Function block allows one to add MATLAB functions to Simulink models for deployment to embedded processor, the quality of service of individual class of service is computed and compared before attempting to output the entity and feed it as the new updated service rate into the server. 85

86 Figure 4.5 CDGPS Computational model Figure 4.5 shows the detailed simulation model of the rate scheduling procedure. From the rate scheduling model, the constant block divides the service rate coming from the server for each flow, i.e., voice, video, or data, and are scheduled slot by slot. By the end of each time slot, the used bandwidth capacity is then subtracted from the total UMTS capacity (2 Mbps) and the remaining capacity is stored into the system memory block. Upon receiving bandwidth requests from all backlogged flows (active users), the next service rate computation is done in the form of written MATLAB function using the embedded MATLAB function block. The scheduler then allocates the service rate for the next slot, taking into account the individual quality of service of each flow and the available uplink 86

87 capacity. The total channel rate used by different service class is summed up by the add block and subtract from the total UMTS channel rate by the subtract block to aid the computation of the service utilization. Figure 4.6 A Scope of entities generated Figure 4.6 is a scope showing the number of entities (traffic) generated by the server at any given time. 4.2 Performance metrics When evaluating the quality of service, several metrics are used in this work such as Throughput, Average delay, Packets loss rate. Throughput: This is the amount of successfully transmitted packets for each flow divided by the amount of total sent packets. It is computed as hvs hs2= 92 3L.(1 X 4533 ) 92 3L (4.1) 87