«End-to-End QoS Provision over Heterogeneous IP and non IP Broadband Wired and Wireless Network Environments A dissertation submitted in satisfaction ...»
QoS requirements can be either hard (i.e. deterministic) or soft (i.e. statistical). In the hard QoS case, guarantees are provided and strictly enforced based on a contract between the users and the service network. In the soft QoS case, guarantees are promised in a statistical sense, but may not be strictly enforced for a single instance. It must be added that even packets of the same media application may have diﬀerent QoS requirements in terms of delay and packet loss preference, which leads to a soft QoS rating for the application. Soft QoS services can be divided into classes characterized by diﬀerent QoS assurance levels. In the current best-eﬀort service environment, no QoS guarantees are supported.
To provide QoS for media delivery, it is important to consider the interaction between the application and the network, and also to achieve end-to-end QoS continiuty across heterogeneous network domains. Analyzing and designing such interactions and mapping are the central themes of this research.
1.1 Contribution of the dissertation
One main contribution of this research si to set-up a framework within which core networks, wireless and mobile access networks, and ﬁnally end-systems (i.e.
streaming server) can cooperate for better end-to-end QoS provision. This framework includes the following key components and addresses the following requirements:
1. An applications data segments of single or multiple video streams are packetized and then categorized according to the application level QoS sensitivity to packet loss and delay. A quantitative index is given to each packet top reﬂect its importance relative to receiving acceptable QoS from the network. This mechanism guides a streaming server to eﬃciently diﬀerentiate, on the basis of video data content, the QoS levels of the network service to be requested.
2. The mapping from application data’s QoS categories to the network service classes, which will often be called QoS mapping must be cost eﬀective. The QoS mapping should be designed with an awareness of both the meaning of the application’s QoS categorization and the QoS provided by the network side.
3. For a service contract to be constructed, optimal or eﬀective QoS mapping per ﬂow or per aggregated class requires a balance between the QoS requests assigned by a user and the limited number of QoS levels of a DiﬀServ network.
4. The proposed framework includes proper resource manangement schemes, which are to be employed by the network to realize stable and consistent diﬀerentiation fo QoS levels among diﬀerent classes under time-varying network load conditions.
5. The proposed framework includs an intelligent traﬃc-conditioning mechanism at boundary nodes, which is necessary to optimize performance while meeting Service Level Agreement (SLA) between the access network and the network service provided.
6. Our framework includes a rate adaptation module at the streaming server side, because scalable source-encoded stream is employed.
7. The eﬀective combination of application-level and network-level eﬀorts needs to be considered for QoS support.
The contributions of this research include the following:
• I propose QoS mapping framework between the prioritized continuos multimedia streams segments and the service leveles of the QoS-enabled network in terms of packet loss and delay performance.
• I propose a normalized and uniﬁed indexing scheme for the QoS request of an application, which it is call the relative priority index. This index is obtained by combining diﬀerent video factors in a video stream and categorized video data segments according to their importance with respect to acceptable QoS in delivery.
• I investigate optimal or eﬀective QoS mapping between a video stream and a QoS-enabled network. The network consists of diﬀerent wireless and wired network domains, that can support QoS guarantees. Under a given total pricing budget, severla packets from a video stream, categorized on the basis of the index, can be forwarded to the QoS mapping mechanism to achieve improved end-to-end quality.
• I propose an adaptive packet-forwarding algorithm to provide relative service diﬀerentiation in terms of packet loss and delay. This algorithm enables the measured network DS level to stay within a stable range and not ﬂuctuate too much under variable netowrk load conditions.
• I propose a seamless integration of rate adaptation, prioritized packetization, and simpliﬁed diﬀerentiation for MPEG-4 ﬁne granular scalability video stream over heterogeneous networks.
• I proposes a framework for the pricing of video streaming over heterogeneous networks that support QoS and Service diﬀerentiation, based on the cost of providing diﬀerent levels of quality of service to diﬀerent classes. Pricing of network services dynamically based on the level of the service, usage and congestion allows a more competitive price to be oﬀered, and allows network to be used more eﬃciently.
1.2 Organization of the document
The main objective of this research was to construct a system in which multimedia applications and the network service cooperated positively to realize eﬃcient end-to-end QoS provision. With this goal in mind, the research content can be conceptually delineated as: (1) the eﬀorts to be made by the application side; (2) the eﬀorts to be made by the network side to facilitate the cooperation; (3) QoS mapping from the application’s content classes to the network’s service classes;
(4) to guarantee the end-to-end QoS across heterogeneous network domains, including wired and wireless/mobile network domain, by employing eﬃcient QoS
traﬃc class coupling across network domains.
Chapter 1 cover essential background material that is required for an understanding of MPEG-4 Visual and H.264/MPEG-4 AVC, state-of-the-art prioticzed packetization scheme, and widely used network QoS architecture including wired and wireless technologies. In this chapter, the thesis introduces the basicconcepts of digital video coding, concerning scalable video coding and packetizations schemes, and also network architectures, which can support QoS provisiong for wired and wireless/mobile network domain.
Chapter 3 targets to demonstrate through a set of experimental studies that the common operation of IP DiﬀServ and DVB Bandwidth Management (BM) mechanisms can oﬀer quality gains for prioritized MPEG-4 FGS media delivery across an heterogeneous IP/DVB setting. The experimental studies refer to the delivery of eight YUV QCIF 4:2:0 diﬀerent video sequences across a heterogeneous IP/DVB testbed that includes two IP autonomous systems interconnected through a DVB MPEG-2 autonomous system acting as a trunk network.
Chapter 4 discusses the end-to-end QoS provisioning for scalable video streaming traﬃc delivery over heterogeneous IP/UMTS networks. A prototype architecture is proposed, and is further validated, that explores the joint use of packet prioritization and scalable video coding (SVC) together with the appropriate mapping of UMTS traﬃc classes to the DiﬀServ traﬃc classes. A complete set of simulation scenarios, involving eight diﬀerent video sequences and using two diﬀerent scalable encoders, demonstrates the quality gains of both scalable video coding and prioritized packetization.
Chapter 5 addresses the end-to-end QoS problem of MPEG-4 FGS video streaming traﬃc delivery over a heterogeneous IP/DVB/UMTS network. It proposes and validates an architecture that explores the joint use of packet prioritization and scalable video coding together with the appropriate mapping of UMTS traﬃc classes to the DiﬀServ traﬃc classes. A set of experimental scenarios, involving eight diﬀerent video sequences, demonstrates the quality gains of both scalable video coding and prioritized packetization.
Chapter 6 discusses scalable video streaming traﬃc delivery over heterogeneous DiﬀServ/WLAN networks. A prototype architecture is proposed and further validated that explores the joint use of packet prioritization and scalable video coding (SVC) together with the appropriate mapping of 802.11e access categories to the DiﬀServ traﬃc classes. A complete set of simulation scenarios, involving four diﬀerent video sequences using the scalable extension of H.264/MPEG-4 AVC, demonstrates the quality gains of both scalable video codingand prioritized packetization.
Chapter 7 proposes a framework for the pricing of video streaming over heterogeneous networks that support QoS and Service diﬀerentiation, based on the cost of providing diﬀerent levels of quality of service to diﬀerent classes. Pricing of network services dynamically based on the level of the service, usage and congestion allows a more competitive price to be oﬀered, and allows network to be used more eﬃciently. Our framework incorporates the quality of the delivered video in the given networking context into a dynamic service negotiation environment, in which service prices increase in response to congestion, the applications adapt to price increases by adapting their sending rate and/or choice of service.
Finally, concluding remarks and extensions of this research are given in Chapter 8.
2.1 Introduction Multimedia/video coding to further enable its transportation over various network infrastractures, due to the outstanding demand for video streaming applications, is an active reasearch area. Typically, video streaming applications require information to be available to a variety of receivers interconnected through network links with widely varying characteristics. A number of recent video-coding standards have proposed methods to facilitate video communications for diﬀerent QoS enabled networks. Furthermore, multimedia description frameworks, like MPEG21  deﬁne standarized semantic descriptions of multimedia content and network context of use in terms of delay, loss and bandwidth variation. Both video coding techniques and semantic descriptions oﬀer the ability to develop multimedia streaming techniques that are QoS aware and can be adapted to static or dymanic context of use.
For streaming video, the user and network heterogeneity requires both highly scalable video coding and ﬂexible delivery techniques to overcome the problems imposed by Best Eﬀort service. The bandwidth variation, due to this heterogeneity, can be partly compensated for with scalable coding of conventional coding formats, like MPEG-2. Many network technologies address the problem of
QoS guarantees from a network provider point of view, dealing with network
performance and bandwidth utilization, ignoring the quality needs of the application and the end user. It is necessary to develop an overall architectural framework in order to achieve the necessary collaboration of the existing notion of quality of service at diﬀerent system levels and among diﬀerent types of network technologies. The design of a system that satisﬁes both should maximize the utilization of network resources and guarantee diﬀerent levels of QoS. For this, a basic two step process is required. At the ﬁrst step, the application QoS requirements of the multimedia services to be run over the network have to be identiﬁed. At the second step, these requirements have to be mapped to network ones, that should adapt its behaviour accordingly to allow for eﬃcient end-to-end QoS management.
The structure of this section is as follows. In Section 2.2, recent scalable coding methods are surveyed. The section describes the recenlty proposed video streaming techniques that include methods to facilitate video communications for diﬀerent QoS aware networks. It discusses how the QoS requirements are reﬂected in the application layer using scalable video coding, prioritized packetization schemes and network related semantic descriptions. Section 2.5, gives an overview of the recent work on QoS support in the network layer, considering mobile and ﬁxed QoS networks. The application and network perspectives faced out the QoS problem as a single layer problem. Emphasizing on the cross-layer context, Section 2.6 presents the available techniques for mapping application QoS related semantics with the appropriate network low-level description schemes. In Section 2.7, I discuss recently proposed QoS architectural frameworks and stateof-the-art research followed by a short qualitative comparison.
2.2 Scalable Video Coding
Scalable Video Coding  should meet a number of requirements in order to be suitable for multimedia streaming applications. For eﬃcient utilization of available bandwidth, the compression performance must be high. Also, the computational complexity of the codec must be kept low to allow costless but real time implementations. For example, in videoconferencing applications both encoding and decoding processes must be performed in real time and the latency of encoding/decoding must be low. In contrast to real time streaming applications, there are streaming applications where asymmetrical codecs with no-realtime encoding capabilities are acceptable and where requirements on decoding latency are in reasonable levels. In addition to the previously mentioned requirements, lareyed coding can trade-oﬀ among the diﬀerent aspects of video quality, such as frame rate and spatial resolution. For example, a receiver must have the ability to choose high frame rate over high resolution and vice versa in order to meet the available bandwidth. Many scalable video compression algorithms based on discrete cosine transform  (DCT) have been proposed leading to the MPEG-2 scalable proﬁles , MPEG-4 scalable proﬁles  and Scalable Extension of H.264/MPEG-4 AVC [?]. The MPEG-2 standard deﬁnes three scalable proﬁles that can be used independently or in combination: spatial, temporal and Signal-to-Noise (SNR) scalability.