A Framework for Multimedia Streaming over WiMAX in an Emerging Market Scenario

Transcription

1 A Framework for Multimedia Streaming over WiMAX in an Emerging Market Scenario A Thesis Submitted in Partial Fulfillment for the Award of M.Tech in Information Technology By Ritesh Kumar K To International Institute of Information Technology Bangalore June 2007

2 Certificate This is to certify that the thesis titled A Framework for Multimedia Streaming over WiMAX in an Emerging Market Scenario being submitted by Mr. Ritesh Kumar K to the International Institute of Information Technology, Bangalore, for the award of the Degree of Master of Technology, is a record of bonafide research work carried out by him under our supervision, and Mr. Ritesh Kumar K fulfills the requirements of the regulations of the degree. The contents of this thesis have not been submitted to any other university or institute for the award of any degree or diploma. Dr. Abhijit Lele Network Systems Research Co-Supervisor, Motorola India Research Labs, Bangalore Dr. Debabrata Das HP Chair and Asst. Professor Supervisor, IIIT-Bangalore, Bangalore Date Date ii

3 Acknowledgement I extend my heartfelt thanks to my supervisors Dr. Debabrata Das and Dr. Abhijit Lele for their guidance and a free academic environment to think and experiment. Their confidence in my abilities and forbearance with my short comings is the very reason that I could complete this thesis. I dedicate my work to my Spiritual Guru Shri P. Rajagopalachari and to my loving parents. I am grateful to Prof. Sadagopan, Director of IIIT-B and Dr. Chandra Kintala, Director Motorola India Research Labs, who have generously supported this endeavor and encouraged me to pursue research. I am indebted to all who are involved in this joint work between IIIT-Bangalore and Motorola India Research Labs, Bangalore. I thank my entire faculty especially Dr. G.N.S Prasanna with whom I started my first research project at IIIT-B. I am indebted to Prof. B. L Desai at B.V.B.C.E.T, Hubli without whose persistent encouragement I wouldn't have pursued my Master's studies. I thank all my friends especially Reetesh Mukul, Mayank Raj, Omanand, Raushan and Ravi Jain whose company I would forever cherish. I also express my sincere thanks to all my classmates who have constantly encouraged me to pursue a career in research. Ritesh Kumar K Date: June 19, 2007 iii

4 Table of contents 1 CHAPTER 1 INTRODUCTION Overview CHAPTER 2 BACKGROUND OF WIMAX IEEE Physical layer MAC Sub Layer In summary CHAPTER 3 LITERATURE SURVEY CHAPTER 4 ARCHITECTURE FOR RURAL BROADBAND Introduction Kiosk based architecture Cost Analysis In summary CHAPTER 5 MULTIMEDIA STREAMING OVER WIMAX: A QOS PERSPECTIVE MPEG video compression WiMAX support for streaming multimedia WiMAX MAC architecture and Quality of Service MAC PDU format Bandwidth request and allocation Quality of Service Performance of Triple-Play over e based networks for rural environments Simulation metrics In summary CHAPTER 6 SPLIT FLOW BASED APPROACH FOR PROVIDING QOS IN TRIPLE PLAY APPLICATIONS Split flow algorithm iv

5 6.2 Proposed node architecture In summary CHAPTER 7 PERFORMANCE EVALUATION OF A PROPOSED SPLIT FLOW BASED APPROACH FOR TRIPLE PLAY APPLICATIONS OVER WIMAX Performance of Triple-Play over e based networks for rural environments Simulation metrics Conclusion BIBLIOGRAPHY v

6 List of figures FIGURE 1.1 TRENDS IN ICT PENETRATION FIGURE 1.2 TYPICAL DEPLOYMENT SCENARIO FOR RURAL AREAS FIGURE 1.3 COMPARISON CHART FOR MP3 FILE DELIVERY THROUGH VARIOUS WIRELESS TECHNOLOGIES FIGURE 4.1 EMERGING MARKET SCENARIO FIGURE 5.1 VBR TRAFFIC CHARACTERISTICS FIGURE 5.2 MAC QOS ARCHITECTURE FIGURE 5.3 PDU FORMAT (A) GENERIC MAC PDU (B) BANDWIDTH REQUEST PDU FIGURE 5.4 WIMAX NODE ARCHITECTURE FIGURE 5.5 AVG. DELAY PER PACKET PER CONNECTION FIGURE 5.6 AVG. JITTER PER PACKET PER CONNECTION FIGURE 5.7 AVG. THROUGHPUT PER CONNECTION FIGURE 5.8 RELATIVE AVG. THROUGHPUT PER CONNECTION FIGURE 5.9 AVG. DELAY PER CONNECTION FIGURE 6.1MPEG GOP-16 FIGURE 6.2 OUTLINE OF DYNAMIC FLOW SCALING ALGORITHM FIGURE 6.3 PROPOSED NODE ARCHITECTURE FIGURE 7.1 AVG. DELAY PER PACKET PER CONNECTION FIGURE 7.2 AVG. JITTER PER PACKET PER CONNECTION FIGURE 7.3 AVG. THROUGHPUT PER CONNECTION List of tables TABLE 5.1 CONVERGENCE SUBLAYER FORMATS SUPPORTED TABLE 5.2 MAC PDU FIELDS TABLE 5.3 MAC LEVEL PARAMETERS USED IN SIMULATIONS TABLE 5.4 VIDEO SOURCE TRAFFIC PARAMETERS USED IN SIMULATIONS TABLE 6.1 SPLIT FLOW STRUCTURE TABLE 6.2 CLASS OF CUSTOMERS TABLE 7.1MAC LEVEL PARAMETERS USED IN SIMULATIONS TABLE 7.2 VIDEO SOURCE TRAFFIC PARAMETERS USED IN SIMULATIONS vi

7 Publications from this work 1. Ritesh Kumar K, Abhijit Lele, Debabrata Das, On Performance of Triple-Play over e based networks for rural environments, Communicated to IEEE-Asia Pacific Conference on Communications-2007, APCC-2007, Thailand, Oct Invention Disclosure 1. Abhijit Lele, Ritesh Kumar K, Debabrata Das, Method and apparatus for QoS provisioning for Triple-Play applications in emerging markets, submitted for review. vii

8 Abbreviations ADSL AMC ARQ ATM BE BS CAC CBR CID CRC CS DFPQ DiffServ DL DOCSIS DSL DVB-H EDF ertps FEC FFT FIFO FTP ICT IFFT IP ITU LL- WFQ LOS MAC MAP Asymmetric Digital Subscriber Loop Advanced Coding and Modulation Automatic Repeat-reQuest Asynchronous Transfer Mode Best Effort Base Station Call Admission Policy Constant Bit Rate Connection Identifier Cyclic Redundancy Code Convergence Sub-Layer Deficit Fair Priority Queue Differentiated Services Downlink Data Over Cable Service Interface Specifications Digital Subscriber Loop Digital Video Broadcast-Handheld Earliest Deadline First Extended rtps Forward Error Correction Fast Fourier Transform First In First Out File Transfer Protocol Information and Communication Technologies Inverse Fast Fourier Transform Internet Protocol International Telecommunication Union Low Latency-weighted fair queuing Line of Sight Medium Access Control Mobile Application Part viii

9 MIMO Multiple Input Multiple Output MPDU MAC protocol data unit MPEG Motion Pictures Experts Group MPLS Multi Protocol Label Switching MS Mobile Station MSDU MAC service data unit NIST National Institute of Science and Technology NLOS Non Line of Sight nrtps Non Real-Time Polling Services NS Network Simulator OFDM Orthogonal Frequency Division Multiplexing. OFDMA Orthogonal Frequency Division Multiple Access PDA Personal Digital Assistant PHS Payload Header Suppression PHY Physical layer PMP Point to multi Point QoS Quality of Service RR Round Robin rtps Real Time Polling Service rtps Real-Time Polling Services SFID Service Flow Identifier SOFDMA Scalable Orthogonal Frequency Division Multiplexing Access SS Subscriber Station TDM Time Division Multiplexing UGS Unsolicited Grant Services UL Uplink VBR Variable Bit Rate VLAN Virtual Local Area Network VoIP Voice over Internet Protocol WAN Wide Area Network WFQ Weighted fair queuing Wi-Fi Wireless Fidelity WiMAX Worldwide Interoperability for Microwave Access WirelessMAN Wireless Metropolitan Area Networks ix

10 WRR WWW Weighted Round Robin World Wide Web x

11 Abstract The ever increasing necessity and desire to provide universal connectivity for mobile computers and communication devices is fuelling a growing interest in broadband wireless access networks. To solve problems related to use of wireless for the needs of wireless broadband data networking, study group was formed under IEEE project 802 to recommend an international standard for Wireless Metropolitan Area Networks (WirelessMAN). WiMAX 1 is defined as Worldwide Interoperability for Microwave Access by the WiMAX Forum, formed in June 2001 to promote conformance and interoperability of the IEEE standard. The Forum describes WiMAX as a standards-based technology enabling the delivery of wireless broadband as an alternative to cable and DSL. It is being seen as a tool to bridge the Digital Divide by providing applications over the air to sparsely populated regions across the world. A key part of the standard are the Medium Access Control (MAC) protocols needed to support asynchronous and time bounded delivery of data frames to subscribers providing the pre-negotiated Quality of Service (QoS). The standards support fixed and nomadic users in Line of Sight (LOS) and Non Line of Sight (NLOS) environments e-2005 is the latest standard of WiMAX. It has been optimized for dynamic mobile radio channels; this version provides support for handoffs and roaming. It uses Scalable Orthogonal Frequency Division Multiplexing Access (S-OFDMA), a multi-carrier modulation technique that uses subchannelization. The applications targeted in this study are mainly e-education, e-governance and infotainment. A simulation study of Triple-Play applications on e is conducted. The performance evaluation for a typical emerging market scenario indicates that 6-8 simultaneous video sessions can be supported for over an e network operating in point-to-multi point mode of operation. The call admission policy in this 1 WiMAX is a Registered Trademark of WiMAX Forum xi

12 simulation is the generic bandwidth allocation based call admission policy, where a new call is admitted only if the instantaneous bandwidth required for the call is lesser than the peak bandwidth available in the network. The problem associated with this policy is that the average delay and jitter in the system increases above the tolerable limits of the multimedia streaming if the number of parallel sessions from a subscriber increases above six for given parameters. Hence, a new policy for call admission based upon dynamic flow scaling is proposed that preferentially drops subflows from connections which themselves consist of a plurality of sub flows based on an algorithm running at the Base Station (BS). By means of simulation it is proved that using prioritized dropping of sub flows which can be reconstructed if required at the receiver end, we can increase the number of parallel rtps sessions per subscriber station in an e network. A modified node architecture and an algorithm are proposed that runs at the BS to determine whether a new session can be established or dropped. It has been observed through means of simulation that the average delay for Triple-Play applications is approximately 64 milliseconds, which is well within the tolerable threshold. xii

13 1 Chapter 1 Introduction It has been long known that access to information opens doors for wider economic and social development opportunities. In 1984, [1] pointed the fact that the lack of telecommunication infrastructure in developing countries impedes economic growth. In 1996, the International Telecommunication Union (ITU) initiated a United Nations (UN) project for the Right to Communicate aimed at providing access to basic Information and Communication Technologies (ICTs) for all, with motivation to reduce information poverty for developing countries. Figure 1.1 Trends in ICT penetration As Figure 1.1 suggests in the past decade, though a lot of progress has been made in the area of ICTs, a big void still stays between the connected and the unconnected populations in the developed and developing nations of the world. One possible solution lies in combining more than one of these digital utilities to increase the value proposition of the product. For illustration, a mobile laptop [2] connected to the World Wide Web (WWW) through a broadband metropolitan area network (such as WiMAX) could revolutionize the way of life of people in the emerging markets [3], [4]. 1

14 Figure 1.2 Typical deployment scenario for rural areas Compared with other wired solution such as ADSL that suffer from economic and geographic constraints, or any other wireless (Wi-Fi, 3GPP) or satellite system, WiMAX based access networks will enable operators and service providers to costeffectively reach millions of new potential customers providing them with broadband ICT access. Figure 1.3 shows an interesting perspective of comparing the various widely used broadband wireless technologies to deliver internet multimedia services to the fixed and mobile users. Figure 1.3 Comparison chart for MP3 file delivery through various wireless technologies As seen in the figure above, while it takes more than 18 mins to download an MP3 file of 5MB size through GPRS, it would take only 10 secs to do so through WiMAX. The rest of the current mobile content delivery technologies would take somewhere between 30 secs to 11 mins. The Figure 1.3 clearly suggests the value proposition of WiMAX. WiMAX thus emerges to be a preferred choice for broadband service providers for deployment in emerging markets, because of its data rate capability as 2

15 well as its relatively simpler deployment considerations [11]. A broadband internet service delivery framework such as that of WiMAX effectively can deliver e- governance [5], e-education [6] and infotainment services [7] to a yet untapped audience. It can also provide novel and unique employment opportunities to the rural youth [8]. Since the market projections indicate a steep rise in the number of broadband and internet telephony users in developing countries of the world, such as, India in the next few years, it serves as a tremendous source of motivation to explore the possibility of using WiMAX as a broadband service delivery technology in these scenarios. Thus the focus of this work has been to explore the possibilities of using WiMAX for delivery of Triple-Play services (Voice, Video and Data) in these emerging market scenarios. 1.1 Overview Considering the terrain conditions in rural areas, it is not always possible to lay optical and DSL cables, and hence it is more advantageous to use wireless networks both at the backhaul and last mile access. IEEE e offers high data rates and support for Quality of Service (QoS) and hence is suitable for Triple-Play applications. In addition WiMAX based networks are scalable and cost effective. Specifically, the focus of this work is to analyze the capability of e network operating in Point-to-Multi Point (PMP) mode for Triple-Play applications. The rest of the thesis is organized as follows: Chapter 2 presents an architecture overview for delivery of broadband wireless internet through WiMAX in rural areas. A perspective of emerging market for such applications is also presented with a focus on the Indian scenario. Chapter 3 presents related work that discusses various approaches adopted by the research community, and identifies different issues being addressed by various research groups to provide Quality of Service for multimedia applications over WiMAX networks. Chapter 4 describes the Physical layer (PHY) and the MAC layer (MAC) of the WiMAX standard. It also gives an introductory overview of Quality of Service (QoS) over WiMAX. 3

16 Chapter 5 describes the details of MPEG (Motion Pictures Expert Group) encoding and its statistical characteristics. An overview of the mechanisms involved in guaranteeing the QoS support available in WiMAX is also explained in detail. Performance evaluations of the same along with the results are presented in this section. Chapter 6 presents a novel Call Admission Control algorithm based on Dynamic Flow Scaling for e based WiMAX networks. This chapter also describes the modification in the node architecture proposed to support this new algorithm. Chapter 7 describes the results of performance evaluation conducted using this proposed algorithm and node architecture. We present a discussion of these results obtained and conclude with the outcomes of this research work. 4

17 2 Chapter 2 Background of WiMAX 2.1 IEEE WiMAX is defined as Worldwide Interoperability for Microwave Access by the WiMAX Forum. It was formed in June 2001 to promote conformance and interoperability of the IEEE standard. The Forum describes WiMAX as a standards-based technology enabling the delivery of wireless broadband as an alternative to cable and DSL. The WiMAX technology, based on the IEEE Air Interface Standard has proved itself as a technology that can play a key role in fixed broadband wireless access networks. Fixed WiMAX, based on the IEEE [9] Air Interface Standard has proven to be a cost-effective fixed wireless alternative to cable and DSL services. In December, 2005 the IEEE ratified the e amendment [10] to the standard. This amendment adds the features and attributes to the standard necessary to support mobility. Mobile WiMAX is a broadband wireless solution that enables convergence of mobile and fixed broadband networks through a common wide area broadband radio access technology and flexible network architecture. The Mobile WiMAX Air Interface adopts Orthogonal Frequency Division Multiple Access (OFDMA) for improved multi-path performance in non-line-of-sight environments. Scalable OFDMA (S-OFDMA) has been introduced in the IEEE e Amendment to support scalable channel bandwidths from 1.25 MHz to 20 MHz. Some of the salient features supported by Mobile WiMAX are: High Data Rates: The inclusion of MIMO antenna techniques along with flexible sub-channelization schemes, advanced coding and modulation all enable the Mobile WiMAX technology to support peak downlink (DL) data rates up to 63 Mbps per sector and peak uplink (UL) data rates up to 28 Mbps per sector in a 10 MHz channel [12]. 5

18 Quality of Service (QoS): The fundamental premise of the IEEE MAC architecture is QoS. It defines Service Flows which can map to DiffServ code points or MPLS flow labels that enable end-to-end IP based QoS. Additionally, sub channelization and MAP-based signalling schemes provide a flexible mechanism for optimal scheduling of space, frequency and time resources over the air interface on a frame-by-frame basis. Scalability: Mobile WiMAX technology is designed to be able to scale to work in different channelization from 1.25 to 20 MHz to comply with varied worldwide requirements Physical layer The WiMAX physical layer is based on orthogonal frequency division multiplexing (OFDM). OFDM is the transmission scheme of choice to enable high-speed data, video, and multimedia communications and is used by a variety of commercial broadband systems, including DSL, Wi-Fi, Digital Video Broadcast-Handheld (DVB- H), and MediaFLO, besides WiMAX. OFDM is an elegant and efficient scheme for high data rate transmission in a non-line-of-sight or multipath radio environment. OFDM enjoys several advantages over other solutions for high-speed transmission Reduced computational complexity: OFDM can be easily implemented using Fast Fourier Transform (FFT) / Inverse Fast Fourier Transform (IFFT), and the computational complexity of OFDM can be shown to be O(BlogBT m ), where B is the bandwidth and T m is the delay spread. This complexity is much lower than that of a standard equalizer-based system, which has a complexity O(B 2 T m ). Graceful degradation of performance under excess delay: The performance of an OFDM system degrades gracefully as the delay spread exceeds the value designed for. Greater coding and low constellation sizes can be used to provide fall-back rates that are significantly more robust against delay spread. Adaptive modulation and coding, which allows the system to make the best of the available channel conditions. 6

19 Use as a multiaccess scheme: OFDM can be used as a multiaccess scheme, where different tones are partitioned among multiple users. This scheme is referred to as OFDMA and is exploited in mobile WiMAX. This scheme also offers the ability to provide fine granularity in channel allocation. In relatively slow time-varying channels, it is possible to significantly enhance the capacity by adapting the data rate per subscriber according to the signal-to-noise ratio of that particular subcarrier e PHY In Mobile WiMAX, the FFT size is scalable from 128 to 2,048. Here, when the available bandwidth increases, the FFT size is also increased such that the subcarrier spacing is always khz. This keeps the OFDM symbol duration, which is the basic resource unit, fixed and therefore makes scaling have minimal impact on higher layers. A scalable design also keeps the costs low. The subcarrier spacing of khz was chosen as a good balance between satisfying the delay spread and Doppler spread requirements for operating in mixed fixed and mobile environments. This subcarrier spacing can support delay-spread values up to 20µs and vehicular mobility up to 125 kmph when operating in 3.5GHz. A subcarrier spacing of khz implies that 128, 512, 1,024, and 2,048 FFT are used when the channel bandwidth is 1.25MHz, 5MHz, 10MHz, and 20MHz, respectively MAC Sub Layer The primary task of the WiMAX MAC layer is to provide an interface between the higher transport layers and the physical layer. The MAC layer takes packets from the upper layer. These packets are called MAC service data units (MSDUs) and organize them into MAC protocol data units (MPDUs) for transmission over the air. The IEEE and IEEE e-2005 MAC design includes a convergence sublayer that can interface with a variety of higher-layer protocols, such as ATM, TDM Voice, Ethernet, IP, and any future protocol. The convergence sublayer also supports MSDU header suppression to reduce the higher layer overheads on each packet. The WiMAX MAC is designed from the ground up to support very high peak bit rates while delivering quality of service similar to that of ATM and DOCSIS. The WiMAX MAC uses a variable-length MPDU and offers a lot of flexibility to allow 7

20 for their efficient transmission. For example, multiple MPDUs of same or different lengths may be aggregated into a single burst to save PHY overhead. Similarly, multiple MSDUs from the same higher-layer service may be concatenated into a single MPDU to save MAC header overhead Quality of Service Reliable QoS control is achieved by using a connection-oriented MAC architecture, where all downlink and uplink connections are controlled by the serving BS. Before any data transmission happens, the BS and the mobile station (MS) establish a unidirectional logical link, called a connection, between the two MAC-layer peers. Each connection is identified by a connection identifier (CID), which serves as a temporary address for data transmissions over the particular link. WiMAX also defines a concept of a service flow. A service flow is a unidirectional flow of packets with a particular set of QoS parameters and is identified by a service flow identifier (SFID). By means of these and other auxiliary techniques of differentiating between the application traffic requirements WiMAX provides provisions to deliver Quality of Service (QoS). 2.2 In summary In this chapter a description of WiMAX is provided, that highlights its advantages as compared to the other current broadband wireless networks. WiMAX differentiates itself by providing a high peak bit rate over a large geographical area as well as by providing flexibility in terms of scale of deployment and Quality of Service (QoS) guarantees to its subscribers. The chapter also describes the physical layer and the medium access control layer of WiMAX in brief describing the provisions provided in the IEEE standard for fixed as well as mobile configurations. 8

21 3 Chapter 3 Literature Survey The literature survey conducted can be broadly divided into the following categories: Streaming Multimedia through wireless networks Delivery of Triple-Play with strict latency requirements WiMAX QoS provisioning for multimedia applications Prioritized Video streaming over broadband networks WiMAX performance evaluation for real time traffic Rural broadband through wireless media An important contribution to the area of streaming multimedia content through broadband wireless access networks is that presented by [14].In this work, the authors describe a three pronged approach to tackle the problems related to streaming real time content over the wireless networks. They suggest use of scalable video coding, network-aware adaptation of end systems and adaptive QoS support from networks. This adaptive framework consists of three basic components: 1) Scalable video representations through layered coding 2) Network aware end systems 3) Adaptive services. Under this dynamically adjustable framework, as wireless channel conditions change, mobile terminals and network elements can scale the video streams and transport the scaled video streams to receivers with a smooth change of perceptual quality. Triple- Play (converged voice, video and data) services are expected to be the next big thing to happen in wireless access networks. But Triple-Play services themselves are associated with challenges dealing with scheduling and QoS provisioning. In [15] performance evaluation of a scenario where voice, video and data content are streamed through a Diff. Serve router based network is conducted. The scheduling algorithms under consideration are Priority queuing, weighted fair queuing and Low latency weighted fair queuing (LL-WFQ). The simulations indicate that LL-WFQ performs the best under scenarios where voice traffic has to meet strict delay requirements. The simulation results illustrate the fact that for the video service, delay is conversely proportional to packet loss, for each scheduling scheme and buffer size. 9

22 As the buffer size increases, the packet's presence in the queue is extended, decreasing the number of discarded packets. In case of LL-WFQ the buffer requirements though are larger than in other scheduling schemes. In the present broadband wireless access networks WiMAX seems to be an attractive proposition to deliver Triple-Play services to rural communities. Most of the current work conducted over WiMAX network architecture for QoS provisioning has been limited to scheduling algorithms [16] is one such work where the author has built a simulation platform for MAC performance evaluation. In [16] the authors have described their scheduling architecture and its effectiveness through QualNet [17] simulations. This scheduling architecture is based on GPSS grant mode with min-max fair allocation for uplink scheduling and WFQ for downlink scheduling. In SS uplink scheduler of their architecture, SSs send bandwidth request packets just after UGS data transmission to BS in unicast uplink slots allotted to this SS. The authors do not mention anything in the context of to how these bandwidth request packets are different from the packets sent in bandwidth contention period. In [18] the authors have described a modified QoS architecture using WFQ scheduler both at the SS and BS. They claim of simpler QoS architecture and strict guarantees. The authors however do not provide an analysis of the overhead on the good data throughput because of the additional MAC mechanism to communicate the queue size to the BS before the UL transmission. In [19], the authors describe a Weighted Round Robin (WRR) scheduler [20] with GPSS grant mode. The duration of contention slots and uplink data slots are dynamically distributed according to bandwidth requirements. The authors chose five priority queues with dynamic priority competitive ratio parameter assignment. The authors use WFQ scheduling for higher priority service, WRR scheduling for middle priority service and FIFO scheduling for lower priority service. The authors also employ traffic policing and traffic shaping methods to stop SSs using more then allocated bandwidth. In [21] the authors suggest QoS enhancement of IEEE standard based on cross layer optimizations. These optimizations include traffic classifications and packet mapping strategies of DiffServ services. A hierarchical (two-layer) scheduling algorithm is deployed at BS. Six queues are defined according to their direction (uplink and downlink) and service classes (rtps, nrtps and BE). Fixed bandwidth allocated to UGS flows (and deducted the same from total available bandwidth before 10

23 two-layer scheduling). Deficit Fair Priority Queue (DFPQ) [22] is used as first layer scheduling algorithm. In the context of scheduling algorithms for different flows (except UGS), Early Deadline First (EDF) used rtps, WFQ for nrtps and Round Robin (RR) for BE flows. The authors present simulation results to show the effectiveness of their cross layer QoS architecture. Most of the works described are based on the Scheduler architecture. Nevertheless, it has been shown that suboptimal Call Admission Policies have severe impact on the network performance of the wireless networks [23]. In view of the same few works have been reported on Call Admission policies for the WiMAX networks. One of the significant contributions is [24]. It is however not clear in the way the scheme can be adapted into the framework. The Call admission negotiation based upon strict QoS guarantees of minimum assured rate does provide an upper limit for delay and jitter in packet transmission, but the entire performance analysis is more towards the number of parallel sessions that can be accepted. Since, it has been shown earlier [25] that MPEG traffic is inherently non stationary in nature; it is not convincing how strict bandwidth guarantees can be booked for fast changing traffic characteristics. Though the authors have proposed a bandwidth estimator mechanism, it is not evident form the simulations how this mechanism comes into practice since the intervals over which negotiations are conducted are of the order of one second. One method suggested by the research community while using layered multimedia codecs over time varying network conditions is to drop non-essential frames for traffic shaping [26]. A very significant contribution to this area is from [27]. The author reports graceful degradation of service over the broadband ATM network using prioritized and selective dropping of MPEG frames. Though WiMAX has the best support for QoS in the current generation of wireless networks, very few practical performance evaluations have been conducted to this direction like [28]. The findings with a practically deployed system from Alvarion are reported here. The major findings in this report are that the performance deteriorates sharply as the network load increases above a certain threshold. Also the peak throughput available in the network is about 30-40% of the advertised data rate capability of WiMAX. The growing demand of bandwidth intensive web based services such as Triple-Play in urban as well as rural areas are motivating the use of broadband wireless technologies such as WiMAX. To this direction the work reported in [29] is 11

24 significant. The major contribution of this work is the implementation of WiMAX simulator on NCTUns-3.0 [30] simulator. The performance analysis conducted by the authors is very credible. The results show that the system can support up to 9/10 users, streaming typical cinematic video content in CIF resolution, 24 frames/sec, at average rates of around 750 kbps, when the users are distributed uniformly around the base station. What comes out as an additional observation here again is that there is substantial overhead in terms of contention and MAC messaging in the WiMAX architecture. The observed collective peak throughput in this case is 7.5 Mbps where as the channel bandwidth of 20 MHz is capable of about 70 Mbps peak throughput as per the e standard. The learning that comes out of these research contributions over a period of a decade is that to design and deploy a reliable wireless broadband network to deliver real time streaming video to rural subscribers requires a well defined System Design approach wherein all the layers of a communication network must interact to provide QoS guarantees. 12

25 4 Chapter 4 Architecture for rural broadband 4.1 Introduction The landscape of broadcast services like digital and cable television is continuing to evolve around the world, emerging markets, is looking at this as a delivery platform for e-education, e-governance and Infotainment applications. Taking advantage of this wireless broadband evolution, different services such as telephony, video, and internet are being integrated into a single service called as Triple-Play service. There are multiple approaches to deliver Triple-Play services to the end-user. Using optical networks [32] as a backbone network and xdsl [33] for the last mile is one of the options that have been tested. However, considering the rural conditions, it is not always possible to have wired connectivity, and hence it is beneficial to use wireless networks both at the back haul and customer premise access. IEEE e offers high throughput and supports for Quality of Service (QoS) guarantees and hence is more suitable for Triple-Play applications. WiMAX based networks are known to be economically viable under such scenarios. 4.2 Kiosk based architecture Figure 4.1 Emerging market scenario 13

26 A typical scenario for broadband access in a rural environment is given in Figure 4.1. The last mile access is provided using a WiMAX operating in PMP mode. The WiMAX Base Station (BS) is connected to the internet using a core access network. Triple- Play application servers such as e-education servers reside in the internet. Each house is a Subscriber Station (SS), and multiple end devices such as PC, telephone, remote heath monitoring devices are connected to the SS. Rural environments are characterized by flat terrains, high foliage losses and relatively medium user density. When considering a technology for the emerging markets, affordability determines its economic sustainability [34]. One of the ways to increase economic sustainability is to enhance the capacity of network, so as to maximize the number of simultaneous users supported by the network infrastructure without compromising the QoS requirements. Therefore it is of interest to analyze and maximize the capacity in these deployment scenarios. 4.3 Cost Analysis With respect to rural connectivity in India, the government's objective is to reach about 50 million rural connections, or one phone per three rural households, by end of 2007 and about 80 million rural connections, or one phone per two rural households, by 2010 [39]. While low broadband penetration is a clear opportunity for BWA/WiMAX, the market take off will require sufficient spectrum, very low cost CPE and affordable end-to-end connectivity, including the computing platform. A country where broadband's average revenue per user (ARPU) is estimated at US$8-10 requires very low equipment cost. The innovative net-enabled community info-kiosk is an ideal way to reach the many who are not yet connected. The pricing of the access equipment though, must clearly be competitive to those available currently for DSL access. In fact, DSL modems today are made available to customers at rates below $15. For the scenario depicted in Figure 4.1, a cost analysis for the WiMAX equipment and service is performed below: A subscriber station costs approximately $1200. The installation and cabling charges added together would ramp up the cost of a info-kiosk to around $

27 A WiMAX base station would cost around $30,000. The user equipment ( a PC or equivalent) would cost another $1000. The recovery period for these set up costs would be around two financial years. The subscriber expectations from such a broadband wireless technbology also serves as a major source of motivation for innovation. According to [43], most of the users in India, expect a hand-held subscriber terminal that is feature rich and multifunctional also within a $80 to $100 in cost. At the sametime the operators expect a service model that can guarantee a ARPU comparable to that from DSL currently. This makes it very important for the entrepreneur or the service provider to maximize the system capacity while ensuring the user satisfaction. This calls for an in depth analysis and performance evaluation of a typical rural info-kiosk model for WiMAX. 4.4 In summary This chapter provides a brief description of the proposed kiosk based architecture for delivering broadband services to the rural masses. The chapter also provides an insight into the cost performance issues and revenue model for the service provider under such deployment scenarios. 15

28 5 Chapter 5 Multimedia streaming over WiMAX: A QoS perspective In this chapter a broad description of the building blocks required to support multimedia streaming over WiMAX is presented. This chapter includes a description of the widely used MPEG compression scheme for multimedia. In section 5.1 the characteristics of a variable bit rate MPEG codec is presented. Section 5.2 discusses the various mechanisms in practice for streaming data over WiMAX. In the following section 5.3 QoS architecture supported in WiMAX networks at MAC layer is discussed. Section 5.4 describes a performance evaluation conducted on NS-2 simulation platform for WiMAX in point to multi point (PMP) mode. In Section 5.5 the outcomes of this evaluation along with the conclusions derived are described. 5.1 MPEG video compression Digital video data is encoded as a series of code words in a manner such that the average length of the code words is much smaller than the equivalent bit representation of the original video frames. The MPEG [40] standard allows for the encoding of video over a range of resolutions including HDTV. In this system, encoded pictures are made up of pixels. A source picture is a contiguous rectangular array of pixels. A picture may be a complete frame of video ( frame picture ) or one of the interlaced fields from an interlaced source ( field picture ). A field picture does not have any blank lines between its active lines of pixel. In a progressive sequence, all pictures are frame pictures, and all macroblocks are frame DCT coded. Each 8 x 8 array of pixels is known as a block. A block is an 8 x 8 array of pixels. It is the fundamental unit of discrete cosine transformation (DCT). A 2 x 2 array of blocks is termed as a macroblock. Each macroblock consists of a 16 x16 array of luma (Y) pixels ( 4 blocks). The number of chroma pixels (Cr, Cb) will vary depending upon the chroma pixel structure indicated in the sequence header. The macroblock is the fundamental unit for motion compensation and will have motion vectors associated with it if it is predictively coded. The anchor pictures are previously encoded and decoded pictures that are available at the encoder because the encoder contains a decoder. Backward prediction is done by storing pictures until the 16

29 desired anchor picture is available before encoding the current stored frames. The encoder can decide on a macroblock basis to use forward prediction from a previous picture, backward prediction from a following picture, or interpolated prediction to minimize the prediction error. The encoder must transmit the pictures (I, P and B pictures) out of order so that the decoder has the anchor pictures before decoding the predicted pictures. In motion compensated prediction, the encoder searches for a portion of a previous frame which is similar to the part of the new frame to be transmitted. It then sends a motion vector telling the decoder what portion of the previous frame it will use to predict the new frame. It also sends the prediction error so that the exact new frame may be reconstituted. Compression is achieved using the prediction (motion estimation in the encoder, motion compression in the decoder), 2 dimensional discrete cosine transform (DCT) performed on 8 x 8 blocks of pixels, quantization of DCT coefficients, and Huffman and run/level coding. Pictures called I pictures are encoded without prediction. In MPEG, Huffman coding in combination with Run-Level coding and zig zag scanning is applied to quantized DCT coefficients. Run-Level refers to a run length of zeroes followed by a non zero level. A Huffman code is an entropy code that is optimum in the sense that it achieves the shortest average possible code word length for a source. This average code word length is greater than or equal to the entropy of the source. The zig zag scanning pattern for run length coding of the quantized DCT coefficients was established in the original MPEG standard. The same pattern is used for luminance and chrominance. Additions have been made over years for coding of interlaced picture blocks. Pictures termed P pictures may be encoded with prediction from the previous pictures. B pictures may be encoded using prediction from both previous and subsequent pictures. A Group of Pictures (GOP) begins with a start code and a header. The header carries -Time code information -Editing information -Optional user data First encoded picture in a GOP is always an I picture. Typical structure of a GOP is as shown IBBPBBPBBPBBPBBI. This apparently provides an I picture with sufficient frequency to allow a decoder to quickly begin correct decoding. Over the past several years, since the advent of MPEG standard a lot of work has been devoted towards the modelling and analysis of MPEG video traffic. It has been 17

30 convincingly showed that the MPEG traffic not only exhibits the long range dependency (LRD) characteristics but also is statistically non-stationary [25] due to its inherent variable bit rate (VBR) nature of data bursts. The bandwidth requirement of three typical Triple-Play applications viz, e-education video [35], IPTV clip, and advertisement video, encoded in MPEG format is given in Figure 5.1. e-education applications are characterized by low bandwidth requirement as they have less motion content. IPTV content have a high activity factor and exhibit a high bandwidth requirement. Figure 5.1 VBR traffic characteristics Due to this property exhibited by most of the modern multimedia compression schemes, it becomes important that adequate resource provisioning be undertaken to ensure that the video quality stays within the tolerable limits of human perception. It follows that most of the modern broadband networks provide architectural provisions for Quality of Service (QoS) at the design stage for streaming multimedia. In the following section a discussion of such provisions for WiMAX networks is presented. 18

31 5.2 WiMAX support for streaming multimedia In what follows, a concise description of available support in WiMAX for streaming multimedia is provided. Any latency sensitive application such as streaming multimedia calls for strict guarantees to be provided apriori. This can be assured if the network is designed for an end to end QoS control architecture. This includes support in network routing as well as MAC level resource provisioning [41]. As an under lying carrier network, WiMAX is used for transmission of a wide mixture of traffic types such as voice, multimedia, games and data traffic such as Web browsing, messaging and file transfers. Services can be created, changed, or deleted through the issue of Dynamic Service Addition (DSA), Dynamic Service Deletion (DSD) messages. The architecture provides support for SS or BS initiated actions to be taken through a two or three way hand shake procedure. IP network service is based on a connectionless and best effort model, which is not adequate for many applications that normally require assurances on QoS performance metrics. A number of enhancements including the IntServ and DiffServ architectures have been proposed to address this issue. IntServ is implemented by four components: the signalling protocol (e.g. RSVP), the admission control, the classifier and the packet scheduler. Further, rules are prescribed to classify DiffServ IP packets into different priority queues based on the QoS indication bits in the IP header. Thus, IEEE can support both IntServ and DiffServ architectures. For an effective mechanism to operate over WiMAX network to provide QoS guarantees, a cross layer design is essential. This effectively translates into mapping of IntServ/DiffServ priority fields into the MAC level parameters of WiMAX. Since the MAC of WiMAX does allow traffic to be classified on the basis of priority into four traffic classes, it is possible to define a mapping function in the convergence layer between the Layer 2 and Layer 3 that passes on this information. The Secondary Connection Identifier (CID) can be used to transfer the information between the SS and the BS along with the other MAC layer control messages. Once the RSVP session is established at the IP layer, an equivalent resource reservation can be established at the MAC level using the Dynamic Service Addition (DSA) mechanism. In this manner, adequate resources can be booked for streaming 19

32 multimedia sessions. The Dynamic Service Change (DSC) mechanism in the WiMAX MAC does allow incremental additions and deletions in the bandwidth allotment to individual subscriber stations (SSs). Further, an effective mechanism is necessary for the latency and jitter requirements to be met even after the bandwidth provisions are made. These effectively translate into the MAC QoS architecture of WiMAX. In the following section an overview of the MAC architecture in WiMAX and provisions for real time multimedia are provided. 5.3 WiMAX MAC architecture and Quality of Service Some of the important functions of the MAC layer in WiMAX [31] are: Segment or concatenate the service data units (SDUs) received from higher layers into the MAC PDU (protocol data units), the basic building block of MAC-layer payload Select the appropriate burst profile and power level to be used for the transmission of MAC PDUs Retransmission of MAC PDUs that were received erroneously by the receiver when automated repeat request (ARQ) is used Provide QoS control and priority handling of MAC PDUs belonging to different data and signaling bearers Schedule MAC PDUs over the PHY resources Provide support to the higher layers for mobility management Provide security and key management Provide power-saving mode and idle-mode operation 20

33 Figure 5.2 MAC QoS architecture The MAC layer of WiMAX, as shown in Figure 5.2, is divided into three distinct components: the service-specific convergence sublayer (CS), the common-part sublayer, and the security sublayer. The CS, which is the interface between the MAC layer and layer 3 of the network, receives data packets from the higher layer. These higher-layer packets are also known as MAC service data units (SDU). The CS is responsible for performing all operations that are dependent on the nature of the higher-layer protocol, such as header compression and address mapping. The CS can be viewed as an adaptation layer that masks the higher-layer protocol and its requirements from the rest of the MAC and PHY layers of a WiMAX network. The common part sublayer of the MAC layer performs all the packet operations that are independent of the higher layers, such as fragmentation and concatenation of SDUs into MAC PDUs, transmission of MAC PDUs, QoS control, and ARQ. The security sublayer is responsible for encryption, authorization, and proper exchange of encryption keys between the BS and the MS. The CS is also responsible for mapping higher-layer addresses, such as IP addresses, of the SDUs onto the identity of the PHY and MAC connections to be used for its transmission. The WiMAX MAC layer is connection oriented and identifies a logical connection between the BS and the MS by a unidirectional connection identifier (CID). The CIDs for UL and DL connections 21

34 are different. The CID can be viewed as a temporary and dynamic layer two address assigned by the BS to identify a unidirectional connection between the peer MAC/PHY entities and is used for carrying data and control plane traffic. In order to map the higher-layer address to the CID, the CS needs to keep track of the mapping between the destination address and the respective CID. Convergence sublayer types supported by : IPv4 IPv /Ethernet 802.1Q/VLAN IPv4 over 802.3/Ethernet IPv6 over 802.3/Ethernet IPv4 over 802.1Q/VLAN IPv6 over 802.1Q/VLAN ATM Table 5.1 Convergence sublayer formats supported MAC PDU format 1. The generic MAC PDU is used for carrying data and MAC layer signaling messages. A generic MAC PDU starts with a generic header whose structure is shown in Figure 5.3 as followed by a payload and a CRC. The various information elements in the header of a generic MAC PDU are shown in Table The bandwidth request PDU is used by the MS to indicate to the BS that more bandwidth is required in the UL, due to pending data transmission. A bandwidth request PDU consists only of a bandwidth-request header, with no payload or CRC. The various information elements of a bandwidth request header are provided in Table

35 Figure 5.3 PDU format (a) Generic MAC PDU (b) Bandwidth Request PDU Once a MAC PDU is constructed, it is handed over to the scheduler, which schedules the MAC PDU over the PHY resources available. The scheduler checks the service flow ID and the CID of the MAC PDU, which allows it to gauge its QoS requirements. Based on the QoS requirements of the MAC PDUs belonging to different CIDs and service flow IDs, the scheduler determines the optimum PHY resource allocation for all the MAC PDUs, on a frame-by-frame basis. 23

36 Table 5.2 MAC PDU fields 24

37 5.3.2 Bandwidth request and allocation Figure 5.4 WiMAX node architecture 25

38 In the downlink as in Figure 5.4, all decisions related to the allocation of bandwidth to various MSs are made by the BS on a per CID basis, which does not require the involvement of the MS. As MAC PDUs arrive for each CID, the BS schedules them for the PHY resources, based on their QoS requirements. Once dedicated PHY resources have been allocated for the transmission of the MAC PDU, the BS indicates this allocation to the MS, using the DL-MAP message. In the uplink, the MS requests resources by either using a stand-alone bandwidthrequest MAC PDU or piggybacking bandwidth requests on a generic MAC PDU, in which case it uses a grant-management subheader. Since the burst profile associated with a CID can change dynamically, all resource requests are made in terms of bytes of information, rather than PHY-layer resources, such as number of subchannels and/or number of OFDM symbols. Bandwidth requests in the UL can be incremental or aggregate requests. When it receives an incremental bandwidth request for a particular CID, the BS adds the quantity of bandwidth requested to its current perception of the bandwidth need. Similarly, when it receives an aggregate bandwidth request for a particular CID, the BS replaces its perception of the bandwidth needs of the connection with the amount of bandwidth requested. The Type field in the bandwidth-request header indicates whether the request is incremental or aggregate. Bandwidth requested by piggybacking on a MAC PDU can be only incremental. When multiple CIDs are associated with a particular MS, the BS-allocated UL aggregate resources for the MS rather than individual CIDs. When the resource granted by the BS is less than the aggregate resources requested by the MS, the UL scheduler at the MS determines that allocation and distribution of the granted resource among the various CIDs, based on the amount of pending traffic and their QoS requirements. In WiMAX, polling refers to the process whereby dedicated or shared UL resources are provided to the MS to make bandwidth requests. These allocations can be for an individual MS or a group of MSs. When an MS is polled individually, the polling is called unicast, and the dedicated resources in the UL are allocated for the MS to send a bandwidth-request PDU. The BS indicates to the MS the UL allocations for unicast polling opportunities by the UL MAP message of the DL sub frame. Since the resources are allocated on a per MS basis, the UL MAP uses the primary CID of the MS to indicate the allocation. The primary CID is allocated to 26

39 the MS during the network entry and initialization stage and is used to transport all MAC level signaling messages Quality of Service One of the key functions of the WiMAX MAC layer is to ensure that QoS requirements for MAC PDUs belonging to different service flows are met as reliably as possible given the loading conditions of the system. This implies that various negotiated performance indicators that are tied to the overall QoS, such as latency, jitter, data rate, packet error rate, and system availability, must be met for each connection. The real-time polling services (rtps) are designed to support real-time services that generate variable-size data packets on a periodic basis, such as MPEG (Motion Pictures Experts Group) video. In this service class, the BS provides unicast polling opportunities for the MS to request bandwidth. The unicast polling opportunities are frequent enough to ensure that latency requirements of real-time services are met. This service requires more request overhead than UGS does but is more efficient for service that generates variable-size data packets or has a duty cycle less than 100 percent. In WiMAX, a service flow is a MAC transport service provided for transmission of uplink and downlink traffic and is a key concept of the QoS architecture. Each service flow is associated with a unique set of QoS parameters, such as latency, jitter throughput, and packet error rate, that the system strives to offer. A service flow has the following components: Service flow ID, a 32-bit identifier for the service flow. Connection ID, a 16-bit identifier of the logical connection to be used for carrying the service flow. The CID is analogous to the identity of an MS at the PHY layer. As mentioned previously, an MS can have more that one CID at a time, that is, a primary CID and multiple secondary CIDs. The MAC management and signaling messages are carried over the primary CID. Provisioned QoS parameter set, the recommended QoS parameters to be used for the service flow, usually provided by a higher-layer entity. Admitted QoS parameter set, the QoS parameters actually allocated for the service flow and for which the BS and the MS reserve their PHY and MAC resources. The admitted QoS parameter set can be a subset of the provisioned 27

40 QoS parameter set when the BS is not able, for a variety of reasons, to admit the service with the provisioned QoS parameter set. Active QoS parameter set, the QoS parameters being provided for the service flow at any given time. Authorization module, logical BS function that approves or denies every change to QoS parameters and classifiers associated with a service flow. The various service flows admitted in a WiMAX network are usually grouped into service flow classes, each identified by a unique set of QoS requirements. This concept of service flow classes allows higher-layer entities at the MS and the BS to request QoS parameters in globally consistent ways. WiMAX does not explicitly specify what the service flow classes are, leaving it to the service provider or the equipment manufacturer to define. As a general practice, services with very different QoS requirements, such as VoIP, Web browsing, , and interactive gaming, are usually associated with different service flow classes. The overall concept of service flow and service flow classes are flexible and powerful and allow the service provider full control over multiple degrees of freedom for managing QoS across all applications. 5.4 Performance of Triple-Play over e based networks for rural environments Performance evaluation [42] is carried out for single cell as shown in Figure 4.1. The system under consideration comprises one BS and twelve SS. The BS and SS maintain separate queues for each class of traffic. All the simulation results presented in this thesis is for rtps traffic. rtps is an upstream flow scheduling service type that is used for mapping Variable bit rate (VBR) to a service flow. At the start of the session, the SS captures QoS requirement in terms of Peak Bandwidth, Delay and Jitter. Based on these QoS parameters, the grant size and grant intervals are negotiated with the BS. The AllocateQoSParam primitive is used to negotiate the QoS, and ActivateQoSparam is used to activate a service flow. Details of the QoS negotiation mechanisms are 28

41 discussed in [10]. The SS and BS maintain a separate queue for the rtps flow. Non conforming packets can be dropped both at the BS and the SS. Simulations were carried out using the Network Simulator- 2 patched with the NIST- WiMAX extensions proposed in [36]. The NIST-WiMAX implements an OFDM physical layer with configurable modulation, time division multiplexing, fragmentation and reassembly of frames and management messages to execute network entry without authentication. In order to make the simulator compatible for Triple-Play applications, the following extension were made to the simulator. Configurable division of total frame length into uplink and downlink subframes according to a given UL: DL ratio. Uplink scheduler extended to support multiple bursts for different nodes in a single UL subframe on a round robin basis. Uplink scheduler dynamically allots different modulation and FEC profiles according to a suitable metric such as the instantaneous queue length at BS or specific connection information. Contention window size made variable to vary the effective UL bandwidth available to rtps. Call admission control is implemented to block calls that exceed the instantaneous threshold bandwidth and also using the split flow approach. Extended the simulator to support variable bit rate, variable packet size traffic generator instead of the default CBR generator to model the multimedia behavior on the WiMAX network. Rician fading channel with shadowing as proposed in [37] is used instead of the default two ray propagation to match with the channel characteristics of rural environments. For the purpose of simulation, the MAC layer was configured as per the parameters given in Table 5.3. DL:UL ratio 3:2 No. of OFDM symbols per frame 450 Frame length 20 msec Modulation Profile 64 QAM, ¾ rate FEC Rician K Factor, Shadowing exponent 1, 2.3 Table 5.3 MAC level parameters used in simulations 29

42 Simulations were carried out using traffic sources from [38], the characteristics of which are given in Table 5.4. On arrival of a new session, one of the five traffic source given in Table 5.4 is randomly selected and associated with it. This random selection is a uniform process. Video1 Video2 Video3 Video4 Video5 Min. size (bytes) Max. size (Kbytes) Avg. size(kbytes) Avg. rate (Mbps) Burstyness ratio Table 5.4 Video source traffic parameters used in simulations Simulation metrics Simulations were carried out for a scenario comprising a single BS. The number of active sessions is varied from 2 to 12 in steps of 2. The average duration of a real time streaming session is set to 32 secs. The call arrival rate is exponentially distributed with =0.1. The metrics measured include the delay and jitter per packet per active connection and throughput. The following section discusses the results in detail. Figure 5.5 Avg. delay per packet per connection The evaluation observations on x-axis start from 150 sec, because the NS-2 initially needs a few simulation seconds to stabilize the routing table etc. As mentioned above, on an average rtps session duration for simulation is 32 sec. For example, at 150 seconds 2 sessions will be parallely active. After another 32 seconds i.e; 182 seconds of simulation 4 sessions are active in parallel. In the same manner, 6 sessions are active at 214 seconds at 312 seconds 12 sessions are active. 30

43 Figure 5.5 gives the delay per packet per active connection. From Figure 5.5 observe that initially the delay is in the range of 10 msec. However over a period of time delay increases to 1.4 seconds which is not within the tolerable delay limits. This increase in delay is due to increase in the average queue occupancy at the BS. The average jitter per active connection is shown in Figure 5.6. From Figure 5.6, observe that initially the jitter is in the range of 0.1 msec. This is due to the small queue occupancy at the BS. However, over a period of the time the jitter increases to about 10msec which is well within the tolerable limits for real time streaming traffic. In simulations it is observed that at most six active connections are supported over a single BS. It is also observed that the average Packet Loss rate is of the order of 60-70% which is a substantial loss. We have not however presented the packet loss studies in this thesis in detail, as the complete study in this respect is in progress. It is also interesting to observe that when on the x-axis, the simulation time is 278 seconds, where 10 sessions are active in parallel, the jitter starts reducing. This can be attributed to the random selection of video sources (as mentioned in the Table 5.4 for the five video sources) in such an order that the average traffic load for 10 parallel sessions is in fact lesser than previously selected 6 parallel sessions at the simulation time 214 seconds. Figure 5.6 Avg. jitter per packet per connection Figure 5.7 gives the average throughput per active connection. From Figure 5.7 we observe that the peak data throughput per active session at any given instant during simulation does not cross as average of 800 kbps, which is substantially low as compared to source data rates of multimedia flows in IPTV scenarios. Further, investigations were carried out in order to observe the data carrying capacities of the 31

44 Figure 5.7 Avg. throughput per connection Figure 5.8 Relative avg. throughput per connection network. The simulation scenario included data sources with peak data rate requirement varying from 100kbps to 800 kbps to model thee lower bit rate multimedia transmissions. Figure 5.8 gives the comparative plot of average throughput per active connection. From Figure 5.8 we observe that the peak throughput of kbps is experienced by connections having a peak bandwidth requirement of around kbps. This indicates that the MAC and PHY overhead brings down the network throughput, as for low data rates the packet loss rates are within tolerable limits (5% as observed in simulations). Another interesting observation from Figure 5.8 is with regard to the simultaneous establishment of active sessions in the network which translates into the load on the network. As the number of simultaneous sessions are increased from two to six the throughput achieved per active connection initially peaks, after which there is a gradual degradation in the observed throughput. This observation is supported by the delay experienced per connection when the number of simultaneous sessions goes 32

45 above 10; Figure 5.9 gives the comparative plot of average delay per active connection. In Figure 5.9, it can be clearly observed that for low data rate, Figure 5.9 Avg. delay per connection traffic flows on WiMAX network, the delay experienced per connection is below 50 msec which is within the tolerable limits when the total load on the system is within a threshold. Above this sustainable threshold it is observed that the delay per active connection rockets beyond the permissible limits 5.5 In summary Based on the simulation results the following observations can be made: 1) The average delay increases with the increase in number of active sessions. In order to maintain the delay within tolerable limits, rate control mechanisms must be implemented at the BS or the queue size at the BS must be increased. 2) The Packet loss rate is high, and beyond tolerable limits for real time streaming data. The high packet loss rate is due to the high bandwidth requirement of the application coupled with low queue size of the BS. The packet loss rate can be reduced by conditioning the buffer at the BS and also by using the flow splitting technique. 3) The response of the system to variable loads indicates that the PHY-MAC overhead is substantial. The overhead at the MAC level could be due to the frame size used for transmission or the high data rate of the multimedia sources. In effect intelligent schemes to aggregate these MAC message 33

46 exchanges amongst frames as well as static or dynamic determination of optimal frame sizes could provide a solution to the observed shortcoming. In this work a performance evaluation of Triple-Play applications over an e network was considered. The simulation results indicate that at most 6 simultaneous Triple-Play applications can be supported over the network, beyond which the delay exceeds tolerable limits. The simulations also indicate that the high delay is due to high buffer occupancy at the base station. Hence we see that though the broadband wireless access provided by WiMAX is state of the art in itself, but it does suffer if the network load is over a threshold, especially while serving the media streaming connections. In the following section, a proposed Call Admission control algorithm and a corresponding node architecture that can support it is explained. 34

47 6 Chapter 6 Split Flow Based Approach for Providing QoS in Triple Play Applications This chapter describes a proposed split flow based architecture for providing Quality of Service (QoS) for Triple-Play applications in rural environments. In section 6.1 an overview of the proposed call admission control algorithm is provided. In section 6.2 a description of the node architecture that supports this call admission control algorithm at the Base Station is provided. A summary of this chapter is presented in section Split flow algorithm In this section we elaborate the split flow algorithm for call admission and resource management. As discussed in Section 5.1, the MPEG stream comprises a Group of Pictures (GOP) composed of I, B and P frames as shown in Figure 6.1. I B B P B B P B B P B B P B B I Figure 6.1MPEG GOP-16 The following flows are derived from the GOP as given in Table 6.1. Flow F 1 Composition I F 2 B 1 B 2 P 1 F 3 B 3 B 4 P 2 F 4 B 5 B 6 P 3 F 5 B 7 B 8 P 4 F 6 B 9 B 10 P 5 F 7 I Table 6.1 Split flow structure 35

48 Class Description Gold Hard guarantees for all flows Silver = 0.1 F th = 2 Bronze = 0.2 F th = 3 Table 6.2 Class of customers As flow F 1 comprises the I frame, it is referred to as the Essential flow and the other flows F 2 F 7 are referred to as the Non-Essential flows. Hard QoS guarantees are provided for the essential flow, and soft QoS guarantees are provided for the nonessential flows. We define Qos Degradation Factor (QDF) as the fraction of the bandwidth that can be scaled from a Non-Essential flow. If is the bandwidth required for the i th flow of the k th session, and if k is the associated QDF, then the flow can be scaled according to the equation 6.1 as per an algorithm 2. (1) We define the Flow Threshold Factor (FSF) F th as the number of flows that are permitted to be scaled within a given session. Based on QDF, the users are classified as per Table 6.2. On arrival of a new session with bandwidth B j, the session is admitted as per equation 2 (2) where B ins denotes the instantaneous bandwidth and B tot denotes the total available bandwidth. If Equation 2 is not satisfied then the flows from some of the admitted sessions are scaled to accommodate the new session. The flow graph of the flow scaling algorithm is given in Figure Due to the proprietary nature of the work details of the algorithm are omitted from this work 36

49 Start Is the available bandwidth sufficient to establish incoming session? No Run the Dynamic Flow Scaling algorithm at BS Yes Establish the new session Can the session be dropped? No Yes Drop the new session Stop Figure 6.2 Outline of dynamic Flow Scaling algorithm 37

50 6.2 Proposed node architecture Figure 6.3 Proposed node architecture The proposed architecture of a WiMAX node is shown in the Figure 6.3 The upstream packets can be either the data packets or the management packets. The data packets is then classified by the traffic classifier into one of the five provisioned traffic classes by the WiMAX standard, namely, UGS, rtps, ertps, nrtps and BE. The rtps packets would then go through the stream classifier, where it is classified according to the nature of the application such as movies, news, sports, advertisements and e-education videos. These streams are then split into the seven traffic flows consisting of two I frame flows and five other consisting of a triplet of 38