A Model Approach to the Estimation of Peer-to-Peer Traffic Matrices
Peer-to-Peer (P2P) applications have witnessed an increasing popularity in recent years, which brings new challenges to network management and traffic engineering (TE). As basic input information, P2P traffic matrices are of significant importance for TE. Because of the excessively high cost of direct measurement, many studies aim to model and estimate general traffic matrices, but few focus on P2P traffic matrices. In this paper, we propose a model to estimate P2P traffic matrices in operational networks. Important factors are considered, including the number of peers, the localization ratio of P2P traffic, and the network distance. Here, the distance can be measured with AS hop counts or geographic distance. To validate our model, we evaluate its performance using traffic traces collected from both the real P2P video-on-demand (VoD) and file-sharing applications. Evaluation results show that the proposed model outperforms the other two typical models for the estimation of the general traffic matrices in several metrics, including spatial and temporal estimation errors, stability in the cases of oscillating and dynamic flows, and estimation bias. To the best of our knowledge, this is the first research on P2P traffic matrices estimation. P2P traffic matrices, derived from the model, can be applied to P2P traffic optimization and other TE fields.
PROJECT OUTPUT VIDEO: (Click the below link to see the project output video):
Researchers have proposed a variety of methods and models in recent years to make a more convenient and precise estimation. In both the methods and the models are well summarized. These works mainly focus on the estimation of matrices for general traffic regardless of the type of traffic carried over the network.
DISADVANTAGES OF EXISTING SYSTEM:
The large volume of P2P traffic significantly increases the load on the Internet, making networks more vulnerable to congestion and failure, and hence brings new challenges to the efficiency and fairness of networks.
Existing models designed for general traffic (e.g., the gravity model) fail to capture the features of P2P traffic, leading to undesirable estimation errors for P2P traffic.
In this paper, we propose a model to estimate P2P traffic matrices based on a close analysis of the traffic characteristics in P2P systems. To capture the critical properties of the P2P traffic, we take the following physically meaningful factors into consideration. Firstly, the number of peers is considered because, intuitively, networks with more peers might have larger volumes of P2P traffic. Another factor is the traffic localization ratio, which covers the internally exchanged portion of P2P traffic. Last but not least, the distance between different networks is also considered, which can precisely reflect the peer selection strategy of the concerned system.
Using real P2P traffic datasets derived from a P2P video on-demand (VoD) system and a P2P file-sharing application, we explore how parameters in the P2P model affect the estimation accuracy. To the best of our knowledge, this is the first work that deals with the estimation of P2P traffic matrices. Therefore, we also evaluate the estimation accuracy of our model through a comparison with two typical models proposed for general traffic matrices, namely the gravity model and the independent connection (IC) model. Evaluation results show that the newly proposed P2P model outperforms the other two models in several metrics, including spatial and temporal estimation errors, stability in the cases of oscillating and dynamic flows and estimation bias.
ADVANTAGES OF PROPOSED SYSTEM:
We argue that a model designed especially for estimating P2P traffic is needed and greatly useful. Existing models designed for general traffic (e.g., the gravity model ) fail to capture the features of P2P traffic, leading to undesirable estimation errors for P2P traffic.
- Neighbor selection
- Data request and Data Transmission
- Traffic matrices
In the neighbor selection phase, a peer newly in the system registers in a centralized server named tracker and retrieves a list of partial peers in the same swarm, which is a group of peers interested in the same file. In the mainstream implementation of trackers, peers in the list are selected randomly without any bias. But recently, many researchers focus on improving locality in this phase, and prefer to select the neighbors closer to the requester, such as P4P. The network distance is either measured by peers themselves or provided by ISP-operated services.
Data request and Data transmission
In Data request phase, the downloading peer will send data requests to its neighbors on the list. According to the default setting in BitTorrent, a peer can only concurrently upload data to at most 4 downloading peers, and will reject all received requests when in full uploading service. Leechers will prefer to respond to the data requests from the peers who have uploaded to them before, while free-riders will reject the majority of the received data requests. Connections are set up between a host and each of its neighbors who have accepted data requests, and then the data transmission phase begins.
We define basic P2P traffic matrices as traffic matrices reflecting traffic volumes among individual peers. The basic P2P traffic matrices are difficult to estimate, because individual peers dynamically join and leave the P2P system. To simplify the analysis, we assume that peers remain stable within a certain time interval t, and build up a probability model for basic P2P traffic matrices.
- System : Pentium IV 2.4 GHz.
- Hard Disk : 40 GB.
- Floppy Drive : 44 Mb.
- Monitor : 15 VGA Colour.
- Mouse :
- Ram : 512 Mb.
- Operating system : Windows XP/7/LINUX.
- Implementation : NS2
- NS2 Version : 2.28
- Front End : OTCL (Object Oriented Tool Command Language)
- Tool : Cygwin (To simulate in Windows OS)
Ke Xu, Senior Member, IEEE, Meng Shen, Yong Cui, Member, IEEE, Mingjiang Ye, and Yifeng Zhong, “A Model Approach to the Estimation of Peer-to-Peer Traffic Matrices”, IEEE TRANSACTIONS ON PARALLEL AND DISTRIBUTED SYSTEMS, VOL. 25, NO. 5, MAY 2014.