III. ARCHITECTURE JOURNAL OF NETWORKS, VOL. 7, NO. 10, OCTOBER Figure 1. Resource monitoring architecture for P2P IPTV systems.

Transcription

1 1624 JOURNAL OF NETWORKS, VOL. 7, NO. 10, OCTOBER 2012 IPTV-RM: A Resources Monitoring Architecture for P2P IPTV Systems Wenxian Wang Network and Trusted Computing Institute, College of Computer Science, Sichuan University, Chengdu, China Institute of Information Security, Sichuan University, Chengdu, China catean@163.com, catean@scu.edu.cn Xingshu Chen and Haizhou Wang Network and Trusted Computing Institute, College of Computer Science, Sichuan University, Chengdu, China chenxsh@scu.edu.cn, whzh.nc@qq.com Abstract Resources monitoring is an important problem of the overall efficient usage and control of P2P IPTV systems. The resources of IPTV can include all distributing servers, programs and peers. Several researches have tried to address this issue, but most of them illuminated P2P traffic characterization, identification and user behavior. The main contributions of this paper are twofold. Firstly, a resources monitoring architecture for P2P IPTV systems, IPTV-RM, was presented based on previous work. The monitoring architecture employs a hierarchical structure and provides systemic monitoring including resources discovery, relative information extraction and analysis, trace and location. It gives a systematic framework for IPTV resources monitoring. Secondly, a distributed program crawling system (DMP-) was first proposed to collect information of programs, and a peer crawling system was put forward to harvest peers of a program. The results show that they are efficient and can be used for resource collection of other P2P system. Index Terms P2P, IPTV, IPTV-RM, Resources Monitoring, I. INTRODUCTION Peer-to-Peer (P2P) applications take advantage of resources such as storage, CPU cycles, content or human presence available at the edge of the Internet in order to provide a service [1]. With the development and maturity of P2P technology, P2P applications become more and more popular in recent ten years, which include filesharing systems, audio-based VOIP systems, and videobased IPTV systems [2-5]. However, they account for a significant proportion of Internet traffic. According to a survey from CacheLogic [6] in June, 2004, 60% of the Internet s traffic is P2P. In Addition, P2P IPTV applications, such as PPLive [2], QQLive [4] in China, Foundation item: Project (2007CB311106) supported by Major State Basic Research Development Program of China; Project ((242)2009A82) supported by Special Plan Program of National Information Security. Corresponding author: Xingshu Chen, Professor of Sichuan University; Tel: ; chenxsh@scu.edu.cn. become popular gradually and contribute a great amount of P2P traffic to Internet [7]. And it was reported that PPLive has more than 200 million user installations and its active monthly user base (as of Dec 2010) is 104 million, i.e., PPLive has a 43% penetration of Chinese internet users [8]. Now P2P IPTV systems are difficult to be monitored because they use proprietary protocols and there are scarcely source codes or official documents published. Monitoring P2P IPTV systems is a critical problem, and several researches have tried to address this issue. But most researches illuminated P2P traffic characterization, identification, system performance and user behavior. This paper proposed IPTV-RM, a resource monitoring architecture for P2P IPTV systems. And the prototype system was developed to monitor popular P2P IPTV systems, such as PPLive, PPStream, UUSee and SopCast. The remainder of this paper is structured as follows: Section II presents relate work of P2P systems monitoring. Section III describes IPTV-RM, a resources monitoring architecture for P2P IPTV systems. Section IV provides implementation of program and peer crawling system. Section V presents monitoring results of program and peer crawling system. Finally, section VI concludes the paper and gives the future work. II. RELATED WORK There have been many studies about P2P systems monitoring. The monitoring approaches they use can be classified in two discrete tracing methods: passive tracing and active tracing. The passive approach is performed by deploying code at suitable points in the network infrastructure. The P2P traffic is identified from general Internet traffic with the known behavior (e.g. connection ports, feature or patterns) of P2P systems. The passive approach does not increase the traffic on the network. Sen et al. developed a signature-based payload methodology to identify P2P traffic [9]. Karagiannis and Broido proposed an identification method based on statistical characterization of P2P traffic flows [10]. Sebastian first introduced a machine learning method to P2P traffic recognition, and doi: /jnw

2 JOURNAL OF NETWORKS, VOL. 7, NO. 10, OCTOBER used the unsupervised Bayesian learning algorithm for traffic classification [11]. After that, several P2P traffic identification algorithms based on machine learning were proposed [12], such as semi-supervised clustering algorithm, decision tree algorithm, neural network algorithm and support vector machine. The passive approach is potentially transparent, scalable and allows comparison of traffic from multiple domains side-by-side. However, it is dependent upon access to core network infrastructure, which is not always feasible. So it is often used for flow control in firewall or gateway devices. The active approach relies on special crawler to inject test packets into P2P network or send packets to servers and peers, following them and measuring characters of P2P network. Saroiu et al. performed the first active tracing of the Gnutella and Napster Network [13]. They sought to precisely characterize the population of enduser hosts that participate in these two systems. Ripneau et al. developed a crawler that joined the Gnutella network as a servant and used the membership protocol (the PING-PONG mechanism) to collect topology information [14]. From then on, many studies based active method were performed on P2P file sharing system such as BitTorrent, Kazaa and emule. P2P IPTV technology has been restricted by low broadband penetration in the past; however, the rapid and large-scale popularization of broadband technology makes it possible to become the next disruptive IP communication technology, which will greatly revolutionize the people s lives and entertainment [15]. There have been several IPTV applications raised and gained great success commercially, including CoolStreaming [16], PPTV [2], PPStream [3], SopCast [17] and so on. And these IPTV systems draw people s attention, including research fellow. Hei et al. carried out the first active tracing of a commercial P2P IPTV system, namely, PPLive [18]. They further developed a dedicated PPLive crawler to study the global characteristics of PPLive system [19]. Wu et al. presented Magellan to characterize topologies of peer-to-peer streaming networks of UUSee [20]. Vu et al. mapped the PPLive Network to study the impacts of media streaming on P2P overlays [21, 28, 29]. Most of the previous work was performed from a single view of observation or from few nodes within direct access and lacked an automatic system for executing measurements and monitoring. These studies tended to mainly look at either network-centric metrics (e.g., traffic characterization, TCP or UDP connections, video traffic), or user-centric metrics (e.g., user arrival and departure, geographic distribution, channel population). Our studies focused primarily on systemic architecture for resources monitoring, including resources discovery, relative information extraction and analysis, trace and location. And a distributed program crawling system (DMP-) was first proposed to collect various information of programs. III. ARCHITECTURE As shown in Fig. 1, IPTV-RM, the architecture for P2P IPTV system resources monitoring, includes four main components: Task Scheduling Center which allocating various tasks to servers, Information Collector which collects information of relative websites, distributing servers and programs from network, Data Processor which filters and analyzes information obtained by collector, and Monitor Dashboard which performs system management, query, report and monitoring. Figure 1. Resource monitoring architecture for P2P IPTV systems. A. Task Scheduling Center Task Scheduling Center allocates various tasks to servers. As the task scheduling and load balancing is a NP complete problem in a distributed computing environment, fuzzy control theory is introduced to the distributed task scheduling. We proposed a new fuzzy scheduling algorithm based on the fuzzy control theory and task request quality of service [22]. While allocating tasks by the algorithm, the center can choose the best server node to execute every task, maximize the processing capacity of each node, and keep the load of each node in relative balance, especially in the case of large number of tasks. B. Information Collector Information Collector collects information of IPTV websites, distributing servers and programs from Internet and IPTV overlay network. It contains three modules: Websites Discovery, Severs Discovery and. Websites Discovery module is based on a focused crawler [23] which is designed to search relative IPTV systems websites and to collect various information, such as homepage, IP address and ICP license. Through statistic analysis of captured raw packets of communication between peers and servers, Servers Discovery module can grab information of distributing servers like program-list servers, media distributing servers and peer list servers module includes two kinds of special crawlers, namely, program crawler and peer crawler. A program crawler is a disguised IPTV client. It requests program information (e.g. Name, bitrate, description, play-link) from program-list servers. While a peer crawler is used to collect participating peers in the same program.

3 1626 JOURNAL OF NETWORKS, VOL. 7, NO. 10, OCTOBER 2012 C. Data Processor Data Processor includes four modules: Data Filtering, Data Repository, Data Analysis and Storage. All the data obtained by Information Collector are filtered and stored in Data Repository. Data Filter module is mainly used for programs classification and sensitive programs marking through text-based analysis technologies. Data Analysis module is used for statistics and features analysis, which can be used to predict propagation speed and trend of popular programs. The purpose of Storage module is to store records of specified programs. With Data Processor, the monitoring system is able to automatically associate data with a given reference point (time, location, ICP, etc.) and to categorize the programs that are popular. The continuous trace of a program can also be built up to determine how a program is being distributed across various IPTV network by peer crawler. D. Monitor Dashboard Monitor Dashboard includes four modules: System Management, Resource Location, Query and Report, and Monitoring. System Management module is in charge of administration of monitoring system, and includes fault management, configuration management, performance management and security management. Through configuration management, the administrator can easily start or terminate various crawlers at regular time. Resource Location module is based on IP-to-location database like MaxMind [24]. This module can display location of distributing server or peers in a worldwide map. Query and Report module provides easy interface for query and generates statistic reports (e.g. statistical figures, spreadsheets). Monitoring modules is used for trace, record and pollution attack of popular or sensitive programs. Peer crawler mentioned previously can perform trace of a program. A program can be recorded, converted and saved as a video file by FFmpeg [25]. Pollution attack of a program is conducted by a dedicated crawler developed by us [15]. The results show that a single polluter is capable of compromising the whole system and its destructiveness is severe. IV. IMPLEMENTATION Fig. 2 briefly outlines the basic processes of data. First, a focused crawler [23] is designed to collect information of the IPTV systems websites. In addition, program crawler and peer crawler, based on principle of IPTV protocols through passive trace and reverse engineering technology, are developed to obtain information of distributing servers, peers and programs. Then the information is filtered and stored in data repository and is further analyzed and classified. According the command of monitor dashboard, trace of some program is carried out by a few peer crawlers, which can collect viewing peers information (e.g., IP address, TCP number) in a period of time. Figure 2. Processes of data. A. Program Crawling Program-list distribution is very important in IPTV systems. When a client starts up, it requests program-list file from program-list servers and updates the information of programs immediately. The list of program includes program name, program play-link which is the most important identification of signal communication among peers, program descriptions and so on. The client-server architecture is usually adopted to distribute program-list file in IPTV systems, as shown in Fig. 3. The program-list file uses XML to organize various properties of programs. With the number of programs increasing rapidly, the size of program-list file becomes bigger and bigger. For example, PPLive had about 300 thousand programs in 2010, and the size of program-list file was more than 20MB. That is a heavy burden to servers, and makes bad experience to users. Some IPTV systems use compression to decrease the file size, and others use multiple programlist files. Figure 3. Program-list distribution architecture of IPTV. An efficient Distributed Multi-Protocol System (DMP-) was proposed to obtain various information of programs in popular IPTV systems. We used {program name, IPTV protocol} to uniquely identify a program. Fig. 4 presents an overview of the program crawler system architecture. On the basis of crawler clients status, the crawler controller assigns tasks to multiple independent crawler clients. Each crawler client periodically reports its crawling status, CPU and memory consumption to crawler controller.

4 JOURNAL OF NETWORKS, VOL. 7, NO. 10, OCTOBER CPUs and 1GB Memory at Sichuan University of China with 10/100Mbps Ethernet network access. The detailed distributed peer crawler system architecture is plotted in Fig. 6. Controller Database Host Host Host... Process Process Process Crawling Thread Crawling Thread Crawling Thread P2P IPTV Network Figure 4. Distributed multi-protocol crawler system. According to crawling task type, the crawler client invokes Engine, requests program-list file from program-list servers and reports crawling status to crawler controller. When program-list file is downloaded, the crawler client extracts metadata of programs from the file, classifies these programs and stores all information in database. B. Peer Crawling Fig. 5 illustrates the procedures that a peer crawler joins the IPTV network to harvest peers [50]. Figure 5. Workflow of peer crawler. (1) the peer crawler requests the latest program file from program-list servers and updates it immediately; (2) after the peer crawler selects one program to watch, it registers itself in the peer-list servers (tracker servers) and sends out multiple query messages, generally five, to the servers to retrieve a small set of peers that are watching the same program; (3) upon receiving the initial list of peers, the peer crawler uses this seed peer list to harvest additional lists by periodically probing active peers which maintain a list of peers. From the knowledge of the measurement studies for P2P file sharing system presented in [26], the accuracy of captured overlay snapshots depends on the crawling speed because of the high dynamic nature of P2P overlay network. Therefore, we developed a fast and efficient distributed peer crawler system (Peer-), which can capture significantly more accurate snapshots of the IPTV overlay network than previous crawlers [19, 27]. In our experiment, the distributed crawler system was deployed on six PCs with two 2.60GHz Pentium (R) Peer Peer Figure 6. Distributed peer crawler system. controller: The crawler controller accepts play-link of a program as input argument of crawling startup, and then constructs query messages for retrieving peer-list from peer-list servers or partner peers. At the beginning of crawling, the crawler controller implement step 2 of Fig. 5 to retrieve the initial set of peers and then coordinates among multiple independent crawler hosts. In more detail, the crawler controller assigns part of the initial peers to each crawler host and waits for the crawling results from them by using asynchronous communications. Each crawler host then probes these seed peers to harvest additional peer lists and reports crawling results to crawler controller. The distributed crawler system uses adaptive scheduling algorithm to ensure each crawler host remain busy, which is of crucial importance for making full use of the heterogeneous processing capabilities of various crawler hosts. host: After receiving seed peers from crawler controller, each crawler host implements step 3 of Fig. 5 to harvest peer-list in parallel. Upon accomplishing the mission for tracing seed peers, the crawler host then reports crawling results to crawler controller and wait for new missions. hosts are independent crawlers to explore different parts of single program overlay network. Additionally, to achieve a high degree of concurrency, each crawler process creates multiple crawling threads that can independently harvest peer lists from ITV overlay network. C. Peer Crawling Performance Owing to the mismatch between the arrival and departure lags [19], the peer number of crawling results may overestimate the real active peers. Moreover, it is a challenge for Peer- to probe all the peers due to network congestion and NAT problem, which may not respond to peer-list requests at all. According to the resource distribution mechanism of IPTV system, we can harvest almost all the peers information from peer-list servers and partner peers depending on appropriate measurement intervals. To present Peer- performance in speed and efficiency of capturing IPTV

5 1628 JOURNAL OF NETWORKS, VOL. 7, NO. 10, OCTOBER 2012 overlay snapshots, several notations were used to evaluate the completeness of crawling results and efficiency of Peer-. The notations are listed in Table I. TABLE I. THE NOTIONS Symbol Explanation for Symbols Δ t the interval of sending peer-list request T the overall time of crawling program N the actual peer number of program overlay network N the peer number of program overlay network which harvested by crawlers μ t the number of peers which harvested by crawlers at time t ρ the completeness of crawling results ψ the efficiency of crawler system From the above crawling workflow proposed by us, we can know that: μt+δ t μt lim = 0. (1) t T Δt The completeness of crawling results ρ is defined as follow: ρ = N N. (2) It is impossible for Peer- to obtain the precise number of peers in program overlay network, because just only the IPTV providers can know this information by their own platform. However, we can obtain this approximate number by setting appropriate measurement intervals in our experiment. With the method of curve fitting, we discovered a certain quadratic functional relationship between Δ t and T as follow: T 2 = α( Δt) + β Δt γ. (3) + Consequently, we defined the efficiency of Peer- as follow: ρ N N ψ = = =. (4) 2 T N T N ( α( Δt) + β Δt + γ ) That is, the efficiency of Peer- will increase with the reduction of Δ t. Therefore, we implemented a comparative experiment by setting Δ t as 0ms, 1ms, 2ms, 3ms, 5ms and 10ms respectively to choose a suitable parameter. We presented how the number of the participating users evolves for a very popular program with the different intervals of sending peer-list request, as shown in Fig. 7. From the experiment results shown in Fig. 7, we can observe that: (1) If Δ t =0ms, the process of crawling stops within 10 seconds, however, the crawler can only harvest about 2000 peers which has a large deviation compared with the real value. (2) As the value of Δ t increases, the overall crawling time is also growing. (3) When Δ t =1ms or Δ t =2ms, the crawling time of two situations are both about 38 seconds, but the later reaches faster convergence rate than the former. (4) As for Δ t =10ms, the process of crawling lasts about 94 seconds, which is too long time to capture accurate overlay snapshots. Consequently, we concluded that Δ t =2ms is the most valuable interval of sending peer-list request to archive more accurate and complete crawling results with relatively shorter period. Figure 7. Program-list distribution architecture of IPTV V. RESULTS A. Program Crawling The program crawling system was developed in December 2008 and was made of one Controller and ten Clients. The system was deployed on three PC Servers with Intel E5506 CPU and 4GB Memory at Beijing of China with 10Mbps Ethernet network access. From February 2009 to December 2010, the program crawling system collected resources from 28 IPTV systems in China, and obtained 7, 679, 331 program information, in which the proportion of videos on demand is 99.6%. In particular, PPfilm has no live program. Fig. 8 shows program s percent of IPTV systems. PPTV has 25.6 percent programs, while PPfilm has 20.8 percent programs. Figure 8. Program s percent of IPTV systems. We ranked each of the IPTV systems according to their number of program and plotted a typical Cumulative Distribution Function (CDF) of program s percent in Fig. 9. The 21.4% (6/28) popular IPTV systems count for about 80% programs. The 32.1% (9/28) popular IPTV systems have more than 90% programs. Some IPTV systems, like SopCast and TVUPlayer, have only a small proportion of programs, for they have no videos on demand.

6 JOURNAL OF NETWORKS, VOL. 7, NO. 10, OCTOBER Figure 9. CDF of program s percent. B. Peer Crawling A popular channel named My Own Swordsmen was chosen for the experiment because this program has a clear diurnal trend of the evolution of peers viewing program. Fig. 10 shows the peer participation evolution. In Fig. 10, the curves of the number of peers are very similar during the seven measurement days. The daily peers remain with a constant diurnal viewing pattern. The major peaks appear at 1PM and 10PM in one day. This might suggest that P2P IPTV users have different viewing habits from TV audiences. Figure 10. Evolution of the number of simultaneous users (March 10 to March 16, 2008). Fig. 11 and Fig 12 present how the number of participating users evolves for a popular and an unpopular program. Note that the number of participating users of popular program differs greatly from that of unpopular program. The maximum and mean numbers of users of the popular program are 3224 and 1675 respectively; however, those of the unpopular program are only 268 and 100. Figure 11. Diurnal trend of number of participating users of popular program ( My Own Swordsmen, March 10, 2008). Figure 12. Diurnal trend of number of participating users of unpopular program ( Sex and the City, March 10, 2008). VI. CONCLUSION AND FUTURE WORK With the popularity of P2P IPTV systems, it is necessary to monitor information in their overlay for security issue. Based on previous work [48-50, 54], we outline IPTV-RM, a resources monitoring architecture for P2P IPTV System, primarily addressing hierarchical structure and collector. It gives a systematic frame work for IPTV resources monitoring. In addition, Distributed Multi-Protocol System and Peer- system are proposed. To date an IPTV monitoring prototype system has been developed. This monitoring system carries out its monitoring activity using active monitoring. The monitoring system have collected more than 7 million programs. We can trace a program by Peer- system. In the next work, we focus on other monitoring methods and analysis of programs. ACKNOWLEDGMENT This work was supported by Major State Basic Research Development Program of China ( 973 Program) (Number 2007CB311106) and Special Plan Program of National Information Security (Number (242)2009A82). The research presented in this paper would not have been possible without our institutes, at Sichuan University, Chengdu. We wish to thank the reviewers for their valuable suggestions. REFERENCES [1] D. Hughes, J. Walkerdine and K. Lee, Monitoring Challenges and Approaches for P2P File-Sharing Systems, in First International Conference on Internet Surveillance and Protection (ICISP 06), Cap Esterel, Cote d Azur, France. USA: IEEE Press, [2] The PPTV (former PPLive) Homepage. [ ]. http: // [3] The PPStream Homepage. [ ]. http: // [4] The Tencent video (former QQLive) Homepage [EB/OL]. [ ]. http: //v.qq.com/. [5] The UUSee Homepage. [ ]. http: // [6] True Picture of P2P Filesharing. [ ]. [Online]. Available: http: // [7] R. Wang, Y. Liu, Y. Yang et al., Solving the App-Level Classification Problem of P2P Traffic Via Optimized

7 1630 JOURNAL OF NETWORKS, VOL. 7, NO. 10, OCTOBER 2012 Support Vector Machines, in Proceedings of the Sixth International Conference on Intelligent Systems Design and Applications (ISDA'06), 2006, Vol. 2, pp [8] The Synacast Homepage. [ ]. http: // [9] S. Sen, O. Spatscheck and D. M. Wang, Accurate, Scalable In-Network Identification of P2P Traffic Using Application Signatures, in Proceedings of the 13th international conference on World Wide Web (WWW2004), New York, USA, [Online]. Available: http: //www2004.org/proceedings/docs/1p512.pdf. [10] T. Karagiannis and A. Broido, Transport Layer Identification of P2P Traffic, in Proceedings of the 4th ACM SIGCOMM conference on Internet measurement, Sicily, Italy. USA: ACM, 2004, pp [11] S. Zander and T. Nguyen, Automated traffic classification and application identification using machine learning, in Proceedings of the IEEE Conference on Local Computer Networks 30th Anniversary. Sydney, NSW. USA: IEEE Press, 2005, pp [12] H. Liu, W. Feng, Y.g Huang, et al., A Peer-To-Peer Traffic Identification Method Using Machine Learning, in International Conference on Networking, Architecture, and Storage (NAS 2007), IEEE Press, July 2007, pp [13] S. Saroiu, P. Gummadi, and S. D. Gribble, A Measurement Study of Peer-to-Peer File Sharing Systems, in Tech. Report UW-CSE , Dept. of Computer Science and Eng., UniV. of Washington, Seattle, July [14] M. Ripeani, A. Iamnitchi, and I. Foster, Mapping the Gnutella Network, IEEE Internet Computing, vol. 6, no. 1, pp , [15] H. Wang, X. Chen, W. Wang, A Measurement Study of Polluting a Large-Scale P2P IPTV System, China Communications, Vol. 8, No. 2, pp , [16] X. Zhang, J. Liu, B. Li et al., DONet/CoolStreaming: A Data-driven Overlay Network for Peer-to-Peer Live Media Streaming, in Proceedings of 24th Annual Joint Conference of the IEEE Computer and Communications Societies, Miami, FL. USA: IEEE Press, 2005, pp [17] The SopCast Homepage. [ ]. http: // [18] X. Hei, C. Liang, J. Liang, Y. Liu, and K. W. Ross, Insight into PPLive: Measurement Study of a Large Scale P2P IPTV System, in IPTV workshop in conjunction with WWW2006, May [Online]. Available: http: //cis.poly.edu/ross/papers/ppliveworkshop.pdf. [19] X. Hei, C. Liang, J. Liang, Y. Liu, and K. W. Ross, A Measurement Study of a Large-scale P2P IPTV System, IEEE Transactions on Multimedia, Vol. 9, No. 8, pp , [20] C. Wu, B Li, and S. Zhao, Exploring Large-scale Peer-to- Peer Live Streaming Topologies, ACM Transactions on Multimedia Computer Communication Applications, Vol. 4, No. 3, pp. 1-23, [21] L. Vu, I. Gupta, J. Liang, and K. Nahrstedt, Measurement of a Large-scale Overlay for Multimedia Streaming, in Proceedings of the 16th International Symposium on High Performance Distributed Computing (HPDC 07), Monterey Bay, CA, USA. USA: IEEE Press, June 2007, pp [Online]. Available: http: //cairo.cs.uiuc.edu/publications/papers/hpdc-poster.pdf. [22] Y. Duan, X. Chen, and W. Wang, The Research of distributed task scheduling algorithm based on Fuzzy Control, in Proceedings of International Conference on Information Security and Artificial Intelligence (ISAI 2010), Chengdu, China. USA: IEEE Press, 2010, pp [23] W. Wang, X. Chen, and H. Wang, A focused crawler based on naive Bayes classifier, in Proceedings of the Third International symposium on Intelligent Information Technology and Security Informatics (IITSI 2010), Jinggangshan, China. USA: IEEE Press, 2010, pp [24] The MaxMind Free Geolocation Database Homepage. [ ] http: // [25] The FFmpeg Homepage. [ ]. http: //ffmpeg.org/. [26] D. Stutzbach and R. Rejaie, Capturing Accurate Snapshots of the Gnutella Network, in 24th Annual Joint Conference of the IEEE Computer and Communications Societies (INFOCOM 2005), Miami, FL, USA. USA: IEEE Press, 2005, pp [27] L. Vu, I. Gupta, K. Nahrstedt, et al., Understanding overlay characteristics of a large-scale peer-to-peer IPTV system, ACM Transactions on Multimedia Computing, Communications and Applications, Vol. 6, No. 4, pp , [28] R. Zhu, J. Wang, Power-Efficient Spatial Reusable Channel Assignment Scheme in WLAN Mesh Networks, Mobile Networks and Applications, vol. 17, no. 1. pp , [29] R. Zhu, Intelligent Rate Control for Supporting Real-time Traffic in WLAN Mesh Networks, Journal of Network and Computer Applications, vol. 34, no. 5, pp , Wenxian Wang was born in Jinjiang of Fujian province of China in He received his M.E. degrees from College of Chemical Engineering, Sichuan University, China, in He is a lecturer of Network and Trusted Computing Institute, and is currently a Ph.D. candidate at the Institute of Information Security, Sichuan University. He is a member of the International Association of Computer Science and Information Technology (IACSIT). He took part in project Key Technology of P2P Applications Monitoring System, awarded Scientific and Technological Progress Second-class Award of Sichuan Province of China in September His research interests include peer-to-peer networks, information security and trusted computing. Xingshu Chen was born in She received her M.S. degree from College of Computer Science of Sichuan University, Chengdu, China, in 2000 and Ph.D. degree from Institute of Information Security at Sichuan University, Chengdu, China, in From 2005 to 2010, she was with College of Computer Science as an assistant professor, working with teaching and research. Currently, she is a professor and Ph.D. supervisor of College of Computer Science. She is currently the director of Network and Trusted Computing Institute (NTCI) and vice director of Information Management Center. She awarded Scientific and Technological Progress Second-class Award of Sichuan Province of China in September Her general research interests include peer-to-peer networks, information security, computer networks and cloud computing. Haizhou Wang was born in He received his B.E. degrees from College of Computer Science, Sichuan University, China, in He is currently working toward the Ph.D. degree at the same University. He awarded Outstanding Graduate Student of Sichuan University twice in 2009 and 2010, and 3rd Prize in 2010 NVIDIA CUDA Collegiate Programming Contest of China. His research interests include peer-to-peer IPTV systems, information security, and network measurement.