International Journal of Computer Networking, Wireless and Mobile Communications (IJCNWMC) ISSN 2250 1568 Vol.2, Issue 2 June 2012 72-78 TJPRC Pvt. Ltd., MEASUREMENT STUDY OF A P2P IPTV SYSTEM: SOPCAST SANDHYA C P 1 & S.KARTHIK 2 1 Final year M.E-CSE, Department of Computer Science Engineering SNS College of Technology, Sathy Main Road, Coimbatore, Tamilnadu,India. 2 Head, Department of Computer Science Engineering SNS College of Technology, Sathy Main Road, Coimbatore, Tamilnadu,India. ABSTRACT In the world of networking Peer-to-Peer live streaming approaches have become popular in recent years. The new arising P2P applications on the internet are VoIP or IPTV.Here in this paper we are going for the comprehensive study about P2P IPTV applications and their effect on the network. For that we are choosing SopCast as IPTV for case study. Through the study, we collect a huge amount of data then study the applications/users behavior and their effect of traffic in the network. KEYWORDS: IPTV, P2P Measurement, SopCast. INTRODUCTION A new generation of streaming systems over the internet is at an emerging stage with the aim of offering large bandwidth (1-10Mbits/s) to the users.several streaming systems are Joost, PPLive, Zattoo, PPStream, SopCast etc.
73 Measurement Study of A P2P IPTV System: SOPCAST Traditionally video streaming on the internet follows the client server model with a large number of servers. The main problem of this model is the capacity and the bandwidth of server is not infinite and the single point of failure. Based on the services provided P2P streaming can be classified into two categorize: Live streaming and Video-on-demand. Live streaming could be used to broadcast TV channels, radio channels and mainly sporting events. It provides the live programs. It is the scenario where all clients receive streaming content for the ongoing event. The lags between the play points of the clients and that of the streaming source are very small. On the other video-on-demand mainly focus on providing better quality of services. On-demand streaming would allow users watch whatever they want. It allows users to select and watch video content on demand. In addition, live streaming is important for latency and scalability. On the other hand on-demand video streaming is critical for better quality. The major issue in P2P live streaming is to provide users with real time and the playback experiences [1]. In this paper we present the study one of the P2P IPTV applications SopCast.Measured the network traffic generated by the IPTV users watching the UEFA European Football championship. From different networks we collect a large amount of data by using several applications. With this collected data, we can study the applications/users behavior and their effect of traffic in the network. The paper is organized as follows. In section 2 presents the experiment we made. Section 3 gives the result of our work. Finally, section 4 concludes the paper. EXPERIMENT SETUP Our works mainly focus on P2P IPTV. We can classify P2P video streaming as live streaming or Video-on-demand. In video-on-demand the packets are
Sandhya C P & S.Karthik 74 prerecorded where as in live streaming is not. So most of the users are interest to watch lively instead of prerecorded. Our main goal is to study P2P IPTV applications and we think users having live interest and choose the UEFA European Football Championship for our experiment. We collected data from various P2P applications to estimate their traffic effect on the network and study their global behavior. DATA COLLECTION Our experiment setup contains Intel Pentium computers with 1.8 GHz, running Windows XP operating system. Some nodes which were in our campus are connected to the Internet through 100Mbps Ethernet where some others were connected through residential DSL access with 20 Mps or 512 Kbps. When the game started each of the P2P applications are running in different nodes. Information about different P2P applications were obtained from Internet. We collected packets from each node using Ethereal [2].Finally we collected data from all games with different measuring points or applications. In this paper we going to present the result obtained with the SopCast [3] application running on one of our campus node. RESULT ANALYSIS In this section, we will analyze the result. From the general observation we obtained that SopCast relies mostly on UDP,ie for video and signaling SopCast mainly relies on UDP and a few TCP. Next our aim is to know about the download and upload rate. Figure 1 shows the upload and download rate of SopCast. From the figure we can see that the download rate almost uniformly distributed, but the download rate is fluctuating. This shows that there is a lack of tit-for-tat fairness in SopCast. We isolated top ten peers which sent high amount of data across the entire network. In the same way from the top ten, top peers are selected. Figure 2
75 Measurement Study of A P2P IPTV System: SOPCAST shows the download rate of the top ten and top peers. Here for the complete download, the top ten peers contribute to huge amount of data. In the same way Figure 3 shows the upload rate of the top ten and top peers. Here top ten peers contribute fewer amounts of data. By these facts we can explain that the number of uploading larger than download peers. Throughput(bps) 0.3 0.25 0.2 0.15 0.1 0.05 0 Download rate Upload rate 6 12 18 24 Time (Minutes) Figure 1: Total traffic of SopCast (Both download and upload rates) Figure 2 : Download Rate
Sandhya C P & S.Karthik 76 Throughput(kbps) 700 600 500 400 300 200 100 0 Upload Top ten peers Top peer 50 Time (Minutes) 150 Figure 3 : Upload Rate From the studies we obtained top peers of the download session contribute from 50 minute to 150 minute where in the case of uploading, top peers can contribute from 14 minute to 150 minutes. So the top upload peer, has more interest to continue relationship with the neighbor ones. To calculate the signaling overhead, we separated video and signaling traffic [4].We counted the amount of packets which is bigger or equal than 1200 Bytes. If the session had atleast10 packet, we took it as video session and small ones are t top ignored. And we removed all non-video sessions. From the calculation we obtained the signaling overhead for SopCast is 19.3%.
77 Measurement Study of A P2P IPTV System: SOPCAST 180 160 140 120 100 80 60 40 20 0 50 150 Avtive peers Upload Download Time (Minutes) Figure 4 : Churn of peers observerd with SopCast From the Figure 4 we obtained that the number of upload peers are more than the download peers. It incorporates with the video distribution activities. The number of peers is not constant at all time and we are analyzing the effect of arrival and departure of peers on the upload and download rate and the quality of video. CONCLUSIONS The aim of this work is to measure the network traffic generated by IPTV applications. we measured the network traffic generated by the IPTV users watching the UEFA European Football championship. We collected the data during the game and analyze the network and global behavior of the applications. Here we selected SopCast as the application. From the collected data we analyzed the upload and download policies, churn of peers ant the signaling overhead. At present time. we are observing network traffic and global behavior. In future these results help in more clarification in our observation and compare and find the strength and weakness of each IPTV applications.
Sandhya C P & S.Karthik 78 REFERENCES [1] Bit torrent. http://www.bittorrent.com/. [2] Chien-Chia Chen.Peer-to-Peer Live Video Streaming Services:Past, Current, and the Future.In IEEE Signal Processing Magazine, pp. 74 93. [3] X.Hei,C. Liang,J.Liang,Y. Liu, and K.W.Ross, Insights into iptv system, pplive:a measurement study of a large-scale p2p iptv system, in Proc. Of IPTV Workshop,2006. [4] SopCast.Available on http://www.sopcast.com