From EigenTrust to SHAPE-Trust
Od EigenTrust k SHAPE-Trust
Authors
Supervisors
Reviewers
Editors
Other contributors
Journal Title
Journal ISSN
Volume Title
Publisher
České vysoké učení technické v Praze
Czech Technical University in Prague
Czech Technical University in Prague
Date
Abstract
Peer-to-peer sítě (P2P) jsou stále populárnějí, a proto je zásadní v nich vytvořit a udržet bezpečné prostředí. Bezpečnost P2P sítě je zajištěna prostřednictvím systémů pro správu důvěry (reputace) (TMSs). Tyto systémy používají nějakou formu lokálních hodnot důvěry k výpočtu globální důvěry přidělené každému uzlu v síti. Tato práce se zaměřuje na porovnání tří TMS aplikovaných v simulovaných sítích se specifickými vlastnostmi. EigenTrust je iterativní algoritmus, který přiřazuje skóre důvěry každému uzlu prostřednictvím agregace názorů. MaxTrust je rozšíření EigenTrust do nelineární algebry zvané Max-Plus algebra. SHAPE-Trust používá koaliční teorii her a Shapleyho hodnotu k spravedlivému rozdělení globálních hodnot důvěry. Všechny algoritmy byly implementovány v programovacím jazyce Julia. V případě EigenTrust byly implementovány pouze centralizované verze, protože zbylé algoritmy zatím nemají decentralizovanou verzi.
Peer-to-peer (P2P) networks have been rising in popularity, therefore the need to create and maintain a secure environment is pivotal. Security of a P2P network is preserved via Trust (Reputation) Management Systems (TMSs). Those systems use some form of local trust values, to compute global trust assigned to each peer in the network. This thesis focus on a comparison of three TMSs applied in simulated networks with specific properties. EigenTrust, is an iterative algorithm, which assigns a trust score to each peer through aggregating the opinions. MaxTrust is an expansion of EigenTrust to a non-linear algebra called Max-Plus algebra. SHAPE-Trust uses coalitional game theory and Shapley value, to fairly distribute the global trust scores. All algorithms were implemented in the Julia programming language. In case of EigenTrust only the centralized versions were implemented, because the remaining algorithms do not have a decentralized version yet.
Peer-to-peer (P2P) networks have been rising in popularity, therefore the need to create and maintain a secure environment is pivotal. Security of a P2P network is preserved via Trust (Reputation) Management Systems (TMSs). Those systems use some form of local trust values, to compute global trust assigned to each peer in the network. This thesis focus on a comparison of three TMSs applied in simulated networks with specific properties. EigenTrust, is an iterative algorithm, which assigns a trust score to each peer through aggregating the opinions. MaxTrust is an expansion of EigenTrust to a non-linear algebra called Max-Plus algebra. SHAPE-Trust uses coalitional game theory and Shapley value, to fairly distribute the global trust scores. All algorithms were implemented in the Julia programming language. In case of EigenTrust only the centralized versions were implemented, because the remaining algorithms do not have a decentralized version yet.