0 valutazioniIl 0% ha trovato utile questo documento (0 voti)
123 visualizzazioni4 pagine
This paper introduces an approach for measuring the Quality of Experience (QoE) and Quality of Service (QoS) for Internet Protocol television (IPTV) services employing MPEG-2 codec and MPEG-2 transport stream. A testbed which emulates the Internet and xDSL access networks is used for analyzing the QoE and QoS parameters. Our results show that the jitter and loss take most effect on the QoE while the behavior of delay is likely not predictable.
Descrizione originale:
Titolo originale
Analysis of transport network QoS and QoE of IPTV services.pdf
This paper introduces an approach for measuring the Quality of Experience (QoE) and Quality of Service (QoS) for Internet Protocol television (IPTV) services employing MPEG-2 codec and MPEG-2 transport stream. A testbed which emulates the Internet and xDSL access networks is used for analyzing the QoE and QoS parameters. Our results show that the jitter and loss take most effect on the QoE while the behavior of delay is likely not predictable.
This paper introduces an approach for measuring the Quality of Experience (QoE) and Quality of Service (QoS) for Internet Protocol television (IPTV) services employing MPEG-2 codec and MPEG-2 transport stream. A testbed which emulates the Internet and xDSL access networks is used for analyzing the QoE and QoS parameters. Our results show that the jitter and loss take most effect on the QoE while the behavior of delay is likely not predictable.
Analysis of Transport Network QoS and QoE of IPTV Services
Tran Trong Son, Bong-Kyun Lee, Young-Tak Kim
Dept. of Information and Communication Engineering, Graduate School, Yeungnam University 214-1, Dae-Dong, Kyeong-San, Kyeongbuk, 712-749, KOREA trongson@ynu.ac.kr, zakare@ynu.ac.kr, ytkim@yu.ac.kr
Abstract
This paper introduces an approach for measuring the Quality of Experience (QoE) and Quality of Service (QoS) for Internet Protocol Television (IPTV) services employing MPEG-2 codec and MPEG-2 transport stream. A testbed which emulates the Internet and xDSL access networks is used for analyzing the QoE and QoS. We figure out the relationships between QoE and QoS parameters. Our results show that the jitter and loss take most effect on the QoE while the behavior of delay is likely not predictable.
I. Introduction The advances in broadband internet and scalable video coding techniques such as MPEG-1, 2, and MPEG4/H264, etc., have made the Internet Protocol Television (IPTV) possible [1]. IPTV describes a system capable of sending, receiving, and displaying a video stream as a series of IP packets. IPTV can be viewed in several ways. A computer may be used to reassemble protocol packets and convert them into a television signal that can be recognized by standard television set. Another option is the set top box, commonly used by cable companies. The set-top box would be used in place of a computer and would do the same job a computer does. The coming together of voice, data, and multimedia (IPTV) services brings to customers convenience but from the service provider side, there are much tasks needed to be done for the real triple-play service. There are some limitations to the IPTV. IPTV is based on the Internet Protocol, and then it is very sensitive to packet loss, delay and jitter. A windows Media-encoded stream, for instance, takes up 1.0 to 1.5Mbps for Standard Definition TV (SDTV). Ten channels could be sent at one without any effect on the voice or data traffic. The situation is explicitly changed when High Definition TV (HDTV) joins the market. The bandwidth necessary for HDTV is about 20 to 25 Mbps. In this situation, we need QoE/QoS-supported mechanisms. There are many well-known organizations involved in studying general architecture capable of supporting QoE/QoS for IPTV services. The most active one is ITU-T (International Telecommunication Union) [2]. The recommendation Y.1541 [9] and G.1010 [10] of ITU-T could be a good reference for supporting QoS and network performance, and for supporting QoS at application layer, respectively. The IPTV Focus Group of ITU-T is currently working on the network architecture and QoS model for IPTV system. The National Telecommunications and Information Administration (NTIA), United States of America, also has actively activities on introducing a QoE (Quality of Experience) metric, VQM (Video Quality Metric), for evaluating the quality of video stream (IPTV stream, etc.). Other objective metrics of service quality such as the duration of periods degraded service (e.g. Degraded Seconds, Errored Seconds, Unavailable Seconds) provide less information but may be easier to measure. There are many studies on supporting QoS at network and application layer, and on testing the quality of picture (such as VQM, PSNR (Peak Signal to Noise Ratio), etc.). But we do not see much study on the relation between QoE and QoS. In this paper, we discuss about our study on this relation. The rest of this paper is organized as follow. In section II, we explain about the IPTV architecture, and the QoE/QoS requirements for IPTV services. We present our approaches to fully comprehend the relation between QoS and QoE in section III. We show our experimental results in section IV. In section V, we conclude our paper and indicate directions for future work.
II. Overview of IPTV services, architecture, and QoE/QoS considerations 2.1. Overview of IPTV services and architecture Recently, the ITU-T FG IPTV reached a consensus on the following definition of IPTV [4]: IPTV is defined as multimedia services such as television/video/audio/text/graphics/data delivered over IP based networks managed to provide the required level of QoS/QoE, security, interactivity and reliability. Figure 1 depicts the meaning and scope of IPTV definition incorporating all discussions and results. Fig. 1 Meaning and Scope of IPTV
Generally, the basic reference model of IPTV service is shown in Figure 2. This architecture is studied in ITU-T with the analysis of relevant roles that involve in the provision of an IPTV service. As describer in the Figure 2, there are four roles [4]: Content Provider: The entity that owns or is licensed to sell content or content assets. Service Provider: The entity that provides the IPTV services to Customer. The Service Provider relieve on the QoS in the networks provided by Network Provider to guarantee QoE of IPTV services. Network Provider: The entity that provides network services for both Customer and Service Provider. The responsibility of this entity is very high. And it may need to have contracts with not only Service Provider but also Customer about the network parameters (e.g. bandwidth, availability of service,
etc.). Customer: The entity that consumes and pays for the IPTV services. The basic model looks quite simple, but this clear definition let us distinguish IPTV services further by considering each role. For example, we can classify the IPTV services in terms of Content (e.g. live content, managed services, unmanaged services) [1].
Fig. 2 IPTV Reference Model
Looking at the basic IPTV Reference, there are many factors that impact to the Quality of the IPTV services. Typically, the area that covered by Network Provider contains of wide area network (e.g. Internet or BcN, etc.) and access network (e.g. xDSL, FTTH, wireless access networks). These networks influence the quality much. The delay in a small sector of the Internet typically excesses one hundred milliseconds on the normal load. During the high traffic time, the delay, jitter and loss is very high. This is one of the biggest challenges in guaranteeing QoS of the network and so QoE. We also need to consider the Customer network. Nowadays, the Customer network is considered as an infrastructure with many kinds of end- point device and Layer 2 technologies, such as Ethernet, IEEE1394, Wireless, and so on. Study on QoS support in this network is also needed for support the overall QoS [3]. To fully support QoE/QoS we also need to consider the Content and Service Provider roles such as video/audio encoding/compression technologies (e.g. MPEG-2, MPEG- 4/H.264, etc,), and also transport methods. In this paper, we analyze the impact of network parameters on the Quality of Video service (e.g. QoE of IPTV service).
2.2. QoE/QoS Considerations The DSL Forums technical report TR-126 [5] defines Quality of Service (QoS) is a measure of performance at the packet level from the network perspective. Quality of Service (QoS) also refers to a set of technologies (QoS mechanisms) that enable the network operator to manage the effects of congestion on application performance as well as providing differentiated service to selected network traffic flows or to selected users. Quality of Experience (QoE) is the overall performance of a system from the point of view of the users. QoE is a measure of end-to-end performance at the services level from the user perspective and an indication of how well the system meets the users needs. QoE and QoS are two different type of metrics used for evaluating the quality of service. But there is the major difference between them. This difference is the point-of- view. The QoS is considered at network side while the QoE is considered by customers. Customers do not know about the QoS, the only thing they concern is the quality of video (e.g. picture frames in video stream) and the quality of audio. Therefore, the QoE is a measure of end-to-end performance, while the QoS can be measured at different points inside the network. The Video QoE can be measured in three ways [5]: Subjectively using a controlled viewing experiment and participants who grade the quality using rating scales such as Mean Opinion Score (MOS). Objectively at the service layer by using electronic test equipment to measure various aspects of the overall quality of the video signal. Indirect using measurements of network impairments (loss, delay, jitter, duration of the defect) to estimate the impact on video quality, where there is an established relationship between QoE and QoS. The first way is dependent on participants, and the results may be changed widely. There are many studies on the send method, such as PSNR, VQM, etc. VQM accurately tracks subjective quality judgments of video scenes and in addition to providing technology- independent perception-based estimates of subjective quality, the VQM has low computational complexity and can be used for continuous real-time in-service quality monitoring applications. The VQM consists of a linear combination of four parameters that have been optimized for the standard viewing distance of six times picture height. Three parameters are extracted from spatial gradients of the luminance component (Y) of ITU-R BT.601 [7] input and output video streams while one parameter is extracted from the vector formed by the chrominance components (CB, CR). The sampled input and output video streams are assumed to have been calibrated before the processes described herein are performed. This calibration includes compensation for system gain and level offset, as well as spatial and temporal registration of the images [8]. PSNR is a classic metric to assess quality of compressed video. PSNR is defined according to the following formulae: 2 10 2 1 255 10log (1) 1 ( ( , , ) ( , , )) ( 2 1 1)( 2 1 1)( 2 1 1) n n N PSNR MSE MSE d p mn o p mn P P M M N N =
=
= + + + 2 2 2 1 1 (2) p P m M N p P m M = = = = =
Where d(p,m,n) and o(p,m,n) represent respectively degraded and original pixel value at frame p, row m and column n. In spite of this metric does not generally conform to human perception of image distortion, it is a common medium to estimate video codecs quality for the last decades. Currently, there are several implementations of Video quality measurement based on VQM, PSNR, etc, metrics such as Tektronix [11] and Elecard StreamEye Tools [12]. The recommendations Y.1541 [9] and G.1010 [10] of ITU-T provide guidance for supporting QoS at IP layer and application layer. These recommendations, which are shown in Table 1 and Figure 3, respectively, could be good references for Network and Service Providers for guaranteeing QoE to customers.
Table 1 ITU-T Y.1541 IP network QoS class definitions and network performance QoS classes Network performance parameter 0 1 2 3 4 5 IPTD 100ms 400ms 100ms 400ms 1s U IPDV 50ms 50ms U U U U IPLR 10 -3 10 -3 10 -3 10 -3 10 -3 U IPER 10 -4 U
The ITU-T recommendation Y.1541 focuses on the IP layer performance with five different QoS classes. It also provides guidance for IP QoS classes, but no mention
about IPTV. The ITU-T recommendation G.1010 provides delay guidance for several kinds of service. There is a comment for streaming audio and video (timely, delay ~10s, see Figure 3), it is likely not suitable for IPTV because in IPTV service scenario, the content may come from some live sources. This content may come to customers in a different way such as traditional TV. In this case, if the delay guaranteed for IPTV services is about 10s there is a big gap in displaying systems. Moreover, IPTV is provided through IP network where the fluctuation in delay and jitter is very high. If we guarantee the delay about 10s it can result in the high jitter.
Conversational voice and video Command/control (eg Telnet, interactive games) Transactions (eg E-commerce, WWW browsing, email access) Messaging, Downloads (eg FTP, still image)
Other consideration about QoE and QoS is the relation between QoE and QoS. Typically there will be multiple service level performance (QoS) metrics that impact overall QoE. The relation between QoE and service performance (QoS) metrics is typically derived empirically. Having identified the QoE/QoS relationship, it can be used in two ways [5]: Given a QoS measurement, one could predict the expected QoE for a user Given a target QoE for a user, one could deduce the network required service layer performance. Having this relation is very useful in system design and traffic engineering process. When Service Providers start providing IPTV services they need to ensure the level of QoE guaranteed for customers and by translating the QoE into QoS and vice versa they can construct a good network design, maintenance, and planning in future. In the rest of this paper, we will present our study on investigating how the network parameter (e.g. loss, jitter, and delay) affects to the QoE of IPTV service.
III. Testbed and Approaches In this study, we configure a testbed for the performance analysis of the end-to-end IPTV services. The IPTV client utilizes the IPTV services from Service Provider through IP networks. Figure 4 shows our testbed implementation. NMS Streaming Server Client Performance Monitoring Reports Internet Access network xDSL Access network xDSL RTP/UDP
Fig. 4 Testbed for QoE/QoS relation discovery
The Streaming Server acts as IPTV server which delivers IPTV services to the Client. The Client utilizes the IPTV services and also acts as Performance Monitoring point. We choose to measure the network performance at the RTP layer because RTP is the highest layer under the application layer. In order to measure the performance of IPTV services in an environment which is most similar to the real one, we configure the backbone network as the Internet with the Wide Area Network Emulator (WANEM) [13]. The access networks are configured as xDSL ones. Using WANEM, we can simulate the Internet characteristics like network delay, jitter, packet loss, etc. We use Elecard StreamEye Tools [12] as the main tool to analyze the VQM of video streams. In our scenario, we study the relation between QoE and QoS by investigating SDTV service: MPEG-2 codec MPEG-2 transport stream Seven 188-byte packets per RTP/UDP/IP packet. No loss concealment Network Metrics are for RTP flows containing video stream only. Figure 5 represents our IPTV model implementation for measuring the QoE and QoS.
Fig. 5 Our IPTV model for QoE/QoS measurement
IV. Experimental Results VQM is used to investigate the effects of jitter and delay on the QoE of IPTV services. Normally, the set-top box (STB) buffers every incoming RTP packets, reorders, and after that forwards them to the Decoder. The RTP reordering time in the STB could affect much to the quality of real-time video streams because in some cases, the jitter and delay in the network are very high, which may result in the misorder of RTP packets, the STB still keeps old RTP packets of one video frame in the buffer and waits for the complete income of others. In this case, other video frames are delayed to the video output even though they are ready. VQM vs Delay 0 1 2 3 4 5 6 7 8 50 100 200 300 400 500 600 Del ay (ms) V Q M averJ itter=30ms averJ itter=50ms averJ itter=60ms averJ itter=80ms averJ itter=100ms
Fig. 6 Relationship between Delay and VQM
To analyze the effects of delay and jitter on real-time characteristics of IPTV services, we implement a specific buffer in the STB. The STB receives, buffers RTP packets, and waits for 50 milliseconds delay between two packets. It means that once one RTP packet comes into the buffer if the next RTP packet does not come after this moment
50ms, it will be considered to be lost. By using this model, we could analyze the effects of jitter and delay on the real- time IPTV services. The effects of delay and jitter on QoE are shown in Figures 6 and 7, respectively. By setting the jitter in the network to relatively stable values, we can measure the impact of delay on VQM with specific jitter values. As shown in Figure 6, while the delay increases the VQM increases (the quality is decreased). In the case where the average jitter is set to 30ms, the delay does not affect to the VQM. It is because of the RTP reordering time is set to 50ms. VQM vs Jitter -1 0 1 2 3 4 5 6 7 8 0 20 40 60 80 100 120 Ji tter (ms) V Q M averDelay=50ms averDelay=100ms averDelay=200ms averDelay=300ms averDelay=400ms averDelay=500ms averDelay=600ms
Fig. 7 Relationship between J itter and VQM
Similarly, for measuring impact of J itter on VQM, we set the packet delay in the network to relatively stable values. Figure 7 represents the trend of impact of jitter on VQM. While jitter is less than 80ms, the VQM increases exponentially as the jitter increases. But if the jitter is higher (about 80 to 100ms), the VQM increases slowly. From Figure 6 and 7, good VQM values (less than about 1.5) can be archived if the jitter is less than 50ms and delay is less than 300ms. VQM vs Loss -1 0 1 2 3 4 5 6 7 0 0.0001 0.0002 0.0003 0.0004 0.0005 0.0006 Loss rati o (%), short vi deo V Q M averVQM maxVQM minVQM
Fig. 8 Effect of loss on VQM
Figure 8 represents effect of loss on VQM. Obviously, loss takes much effect on the VQM. The quality of video stream is guaranteed if the loss ratio is less than about 0.00015 percent (one packet loss in about 6666 packets).
V. Conclusion In this paper, we analyzed the QoE and QoS requirements for IPTV services as well as the relationship between these parameters. We configured a testbed for investigating the relationship between QoE and QoS by using VQM metric. The results showed that the effects jitter and loss is very critical, while the behavior of delay is likely not predictable. The results for VQM and PSNR metrics are quite similar. In this paper, we introduced our study on the QoE/QoS of typical SDTV services using MPEG-2 transport stream. Our future work is to study on other kinds of IPTV services such as HDTV and other type of Video Transport techniques. Currently, we are studying on the QoE/QoS provisioning for IPTV services in the home/office environment.
VI. References [1] J ames She, Fen Hou, Pin-Han Ho, and Liang-Liang Xie, IPTV over WiMAX: Key Success Factors, Challenges, and Solutions, IEEE Communications Magazine August 2007, page(s) 87 - 93. [2] http://www.itu.int/net/home/index.aspx. [3] Min Ho Park, Yeonjoon Chung, Wan Ki Park and Eui Hyun Paik, A Novel QoS Guaranteed Mechanism for Multicast Traffic Controls in the Home Network, ISCE '06. 2006 IEEE Tenth International Symposium on Consumer Electronics, 2006., page(s) 1 6. [4] Chae-Sub Lee, IPTV over Next Generation Networks in ITU-T, Broadband Convergence Networks, 2007. BcN '07. 2nd IEEE/IFIP International Workshop, May 2007. [5] Architecture and Transport Working Group, DSL Forum, Technical Report TR-126, Triple-play Services Quality of Experience (QoE) Requirements, 13 December 2006. [6] http://www.ntia.doc.gov/. [7] ITU-R BT.601-5 (1995), Studio encoding parameters of digital television for standard 4:3 and wide-screen 16:9 aspect ratios. [8] ITU-T J .144 (2001), Objective perceptual video quality measurement techniques for digital cable television in the presence of a full reference. [9] ITU-T Y.1541 (2002), End-user multimedia QoS categories. [10] ITU-T G.1010 (2001), Objective perceptual video quality measurement techniques for digital cable television in the presence of a full reference. [11] http://www.iptv-industry.com/sp/tkx/tkxhome.htm. [12] http://www.elecard.com/. [13] http://wanem.sourceforge.net/.