**2. Related works**

Since the deployment of CoolStreaming, many measurement based studies are published. Some useful measurable parameters such as *buffer width*, *playable video* and *peer offset* are defined in (Ali et al., 2006; Hei et al., 2007a, 2007b; Vu et al., 2006), and startup performance is addressed in user perceptive (Zhou et al., 2007). In fact, nearly all the reports assume a small buffer system, which is far from the real one like PPLive that adopts much large buffer to resist network fluctuant. For a system like PPLive, one can no longer simply assume the same situation for both stable and startup peers. Studies on a mixed system of CDN server and peers can help our study. It is shown in (Lou at el., 2007; Small at el.,2006; Tu at el., 2005; Xu at el., 2003) that, there is a *phase-transition point C*(*t*) at time *t* in the mixed network, any chunks below *C*(*t*) is easy to fetch. The issue like *C(t)* in P2P steaming media system has never been studied. Besides, data fetching strategies are theoretically discussed in many reports. The algorithms of rarest first and greedy (Zhou at el., 2007) are two extreme strategies arise from BT and a mixed strategy of them is proposed in (Vlavianos at el., 2006; Zhou at el., 2007), while what is the suitable fetch strategy in P2P streaming media system needs to be answered. On VoD system aspect, very few studies (Cheng, 2007; Huang, 2007) based on so-called P2P VoD system were ever seen in 2008, however the target network is far from we discussed at all. The server-based VoD users' behavior is studied in (Yu et al., 2006; Zheng et al., 2005) based on core server's log file, but it is questionable whether P2P user has the same feature. Besides, intuitionally, data-sharing environment and user behavior will influence each other in P2P VoD system unlike in server based VoD system, however no relative research reports that.

## **3. Signalling crack and network measurement**

Reverse-engineering-based protocol crack is the first step. It helps understand the working mechanism in depth, but also makes our large-scale measuring possible by developing network crawler. To the best of our knowledge, the work presented here and in related papers by the same authors and colleagues is the first in the world who succeeded in cracking and measuring all the top popular P2P streaming media systems in large scale.

#### **3.1 Brief description of P2P VoD system**

Referring to Fig.1, a typical P2P media streaming system uses few servers to serve large number of audiences (named as *peer*) with both live and VoD programs (Ali et al., 2006; Hei, et al., 2007a; Zhang, et al., 2005). There are significant different design concerns about P2P VoD system and live system: *i).* VoD peer uses much more storage space to cache nearly the whole video in long term than live peer to cache very few latest contents temporarily. Besides, VoD peer may share all the cached contents even if he is in a different channel. (b) P2P live system is of source-driven such that seeder controls the content feeding rate, while P2P VoD system is of receiver-driven and each peer controls playback rate by himself. Unlike live peer, VoD user has more flexibility to choose different playback patterns, such as skipping, fast forwards and fast backwards.

Fig. 1. The system structure

96 Reverse Engineering – Recent Advances and Applications

point" of the analysis is to crack the system protocol, and then develop measurement platform to access to the system legally; next, based on the mass raw data, we investigate and study the user/peer behaviors, especially the startup behaviors which are believed to involve much more systematic problems rather than stable stage; at last, the system's performance, scalability and stability are discussed and the design models and philosophy are revealed based on the peer behavior models. The research steps outlined previously in this paragraph are detailed in Sections 3 to 5. In addition, Section 2 presents related work.

Since the deployment of CoolStreaming, many measurement based studies are published. Some useful measurable parameters such as *buffer width*, *playable video* and *peer offset* are defined in (Ali et al., 2006; Hei et al., 2007a, 2007b; Vu et al., 2006), and startup performance is addressed in user perceptive (Zhou et al., 2007). In fact, nearly all the reports assume a small buffer system, which is far from the real one like PPLive that adopts much large buffer to resist network fluctuant. For a system like PPLive, one can no longer simply assume the same situation for both stable and startup peers. Studies on a mixed system of CDN server and peers can help our study. It is shown in (Lou at el., 2007; Small at el.,2006; Tu at el., 2005; Xu at el., 2003) that, there is a *phase-transition point C*(*t*) at time *t* in the mixed network, any chunks below *C*(*t*) is easy to fetch. The issue like *C(t)* in P2P steaming media system has never been studied. Besides, data fetching strategies are theoretically discussed in many reports. The algorithms of rarest first and greedy (Zhou at el., 2007) are two extreme strategies arise from BT and a mixed strategy of them is proposed in (Vlavianos at el., 2006; Zhou at el., 2007), while what is the suitable fetch strategy in P2P streaming media system needs to be answered. On VoD system aspect, very few studies (Cheng, 2007; Huang, 2007) based on so-called P2P VoD system were ever seen in 2008, however the target network is far from we discussed at all. The server-based VoD users' behavior is studied in (Yu et al., 2006; Zheng et al., 2005) based on core server's log file, but it is questionable whether P2P user has the same feature. Besides, intuitionally, data-sharing environment and user behavior will influence each other in P2P VoD system unlike in server based VoD system, however no relative research reports that.

Reverse-engineering-based protocol crack is the first step. It helps understand the working mechanism in depth, but also makes our large-scale measuring possible by developing network crawler. To the best of our knowledge, the work presented here and in related papers by the same authors and colleagues is the first in the world who succeeded in cracking and measuring all the top popular P2P streaming media systems in large scale.

Referring to Fig.1, a typical P2P media streaming system uses few servers to serve large number of audiences (named as *peer*) with both live and VoD programs (Ali et al., 2006; Hei, et al., 2007a; Zhang, et al., 2005). There are significant different design concerns about P2P VoD system and live system: *i).* VoD peer uses much more storage space to cache nearly the whole video in long term than live peer to cache very few latest contents temporarily. Besides, VoD peer may share all the cached contents even if he is in a different channel. (b) P2P live system is of source-driven such that seeder controls the content feeding rate, while

**2. Related works** 

**3. Signalling crack and network measurement** 

**3.1 Brief description of P2P VoD system** 
