ANALYSIS OF P2P FILE SHARING NETWORK’S CREDIT SYSTEM FOR FAIRNESS MANAGEMENT

Analysis of P2P File Sharing Network’s Credit

System for Fairness Management

Abstract:

                     Fairness is an important management issue for peer-to-peer file sharing systems. In this paper, we study the credit system of the P2P file sharing network through a simple queuing network model. Numerical analysis and experimental results show that this local credit strategy could effectively deal with free riders and provide fairness for the system during a single file exchange. Using this model, we also investigate different management strategies for dealing with the newcomer fairness issue. We propose a simple, private history-based scheme to balance the fairness between two types of newcomers.

Architecture:

The recent popularity and success of peer-to-peer (P2P) file sharing has established its importance while also contributing a majority of the traffic on the Internet. In contrast to the traditional client-server content distribution system, every member of a P2P file sharing network has an equivalent role. Not only can each peer download from other peers, but it is also responsible for uploading content as a server. This often results in fairness issues as many peers called free riders may

Only want to download without uploading or sharing their own content.

Existing System:

                                   The fairness policy of peer-to-peer file sharing system such as Bit Torrent is shown to not be robust, and the free rider could obtain a higher download rate than a tit-for-tat compliant client. In global reputation approaches are suggested for dealing with free riders and malicious peers. However, reputation is always vulnerable when the free rider repeatedly changes its ID for additional benefit, or more than one free rider work together as a coalition. Furthermore, because of the complexity of implementation, the global reputation approach hasn’t been popularly employed from existing P2P file sharing systems in the real world.

Proposed System:

          The primary contributions of this paper are:

                   1) A simple queuing network model is developed to investigate the impact of credit on system fairness; both the numerical analysis and the experiments in the real world illustrate that even if the incentive algorithm

is local-based, it can still deal with free riders and provide fairness during a single file exchange when compared to Bit Torrent’s “TFT” incentive  algorithm.

                   2) Our model is used to compare two different types of credit strategies for providing fairness to the newcomer. A simple, long-duration, and private history-based credit scheme is proposed which will better reward the generous newcomer while limiting the selfish free rider.

This means there are more peers blocked at the beginning and end segments of the

Download process than peers at the other download time period. In a general stochastic analytic framework for incentive-based file-swarming research is proposed, and the first-chunk problem is also shown in the analytical bound and simulation result. The first-chunk problem is related to how the system manages newcomers.

Algorithm Used:

                                Incentive Algorithm Are Used.

This credit-based rule does not only cover the general peers, but it theoretically covers the seeds in a long time period. Because the general peer always follow the

Credit-based incentive policy and thus get benefit through the credit, its average download rate could not be less than the free rider’s. TFT incentive algorithm is rate-based, greedy, and its general peers may achieve higher download rate.

                       

                                Average Download Rate= File Size/ Download Time

Modules:

1.     File Sharing System

                         As one of the most popular file sharing systems. It uses a hybrid architecture, which first obtains online sources information for its expected content, then employs multiple sources in downloading by dividing the whole file into equal-sized pieces called chunks. Through this scheme, more servers appear in the system at the same time to really enhance the system capacity. Mole employs an incentive fairness algorithm based on a local private history credit record to encourage uploading, which means the credit is used to reward a peers’ sharing

Behavior and provide benefit for future downloading.

2.     Comparison with Bit Torrent

                      The popular peer-to-peer files sharing application

“Bit Torrent”, whose “TFT” incentive algorithm was employed for fairness promotion, does not prohibit free riders from completing downloads. Free riders can still obtain enough bandwidth through two kinds of download channels:

                                      1) Free riders download from the seeds which only provide

Uploads and don’t need to conform to the TFT strategy.

                                    2) Free riders download from the other peers without obeying the incentive policy, as each peer periodically unchokes its part of the upload slot to randomly chosen peers through the “optimistic unchoking”  mechanism.

Long-time History Improvement

                    At issue is the case when a free rider attempts to gain extra benefit by pretending to be a newcomer via regularly changing its ID. In this case, both the penalty and reward strategies for coping with the newcomer will not continue to keep the system fair. Furthermore, this newcomer issue could bring to damage the fairness of a P2P network that is based on the global reputation system. Thus, additional, complex authentication or a global reputation system based on public history is proposed. However, this may aggravate the network’s burden due to the large information exchange, or it may require a management center as an addition to the traditional peer-to-peer structure.