| 摘要: | 近年來，隨著點對點傳輸技術的興起，許多用於點對點傳輸的軟體相繼出現，而如何有效的抑制free-riders一直是重要的議題，根據統計，大多數的點對點傳輸軟體採用Gnutella和BitTorrent協定， BitTorrent使用Choking algorithm來抑制free-riders，Choking algorithm只是單純的給予free-riders阻塞，Gnutella使用EigenTrust algorithm來打擊惡意節點，而現實證明EigenTrust algorithm不夠完善，所以兩者的效果都不好，根據2005年統計，在Gnutella網路中有85%的使用者是free-riders，甚至只有1%的使用者主動分享新的檔案。
由於點對點傳輸的軟體上的資源猶如公有資源，所有人都可以使用，以至於大多數的人都不願意分享自身的資源，造成社會困境，導致效率低落，因此本論文將導入社群網路的資料，透過長期的合作促進節點分享，也同時透過社群網路中節點間的關係排斥free-riders。
在本論文提出的系統架構中，我們將考慮各使用者的上傳頻寬、計算能力與電力等因素給與使用者籌碼，這些籌碼將儲存在伺服器中，由伺服器統一管理。系統分享檔案的模式可以分為即時影像傳流分享與檔案分享。在即時影像串流分享系統中，使用者可以使用其籌碼參與競標高畫質即時影像串流的門票。在檔案分享系統中，使用者每時間區段都必須為每單位的下載頻寬支付一定數量的籌碼，換句話說，使用者在本系統中必須使用這些籌碼來競標或購買系統的服務，如此一來，籌碼系統便能夠促進使用者自發性的分享自身的資源。
In recent years, with the emergence of P2P technology, there are many of the applications for P2P technology has emerged and how to effectively inhibit the free-riders is a hot issue. According to statistics, most of the P2P applications used Gnutella protocol and BitTorrent protocol. BitTorrent uses the Choking algorithm to isolate free-riders, but Choking algorithm is simply giving free-riders choke. Gnutella uses the EigenTrust algorithm to combat free-riders. However according to the actuality, EigenTrust algorithm is not a consummated algorithm. Therefore, both of Choking algorithm and EigenTrust algorithm have a bed effect. A research was conducted in 2005, there are nearly 85 percent of Gnutella network users are free-riders from all users. Furthermore, there is only 1 percent of users voluntary to share new files.
Due to the resource in the P2P applications are like public goods that everyone can use it for free. It makes most of the users were not sharing their own resource, to cause tragedy of the commons and inefficiency. Therefore, in this paper, we propose an incentive mechanism on the basis of game theory; it promotes the peers to spontaneous sharing their own resource. We also use social network to promote users to share their resource, to maintain a long-term cooperation between users, and also use user''s relationship of social network to exclude free-riders.
To deploy the mechanism to the architecture, the system will give users counters by considering bandwidth, computing power and electricity of each user. These counters will be stored in server and managed by server. In addition, the sharing model of the system can be divided into live media streaming sharing and file sharing. In live media streaming sharing, users can participate in the auction and using their counters to bid the tickets of high-quality live media streaming. In file sharing, users have to pay counters for every unit download bandwidth. In other words, users need to use the counters to bid or buy service from server. Therefore, the mechanism can promotes the peers to spontaneous sharing their own resource if they need some resources from server. |
