Several distributed applications, implemented overtoday’s Internet, are based on the assumption that participatingagents collaborate in order to achieve their own goal. However,when these applications are modelled as unstructured distributedsystems, the greater autonomy and decentralization encourageantisocial behaviours, which are likely to cause performancedegradation for the whole system. This paper presents a fullydistributed reputation management system that allows the evaluationof agent reputation in unstructured environments withoutany centralized coordination. The proposed approach is basedon game theory and is capable of capturing the highly dynamicnature of the involved communities. As a representative exampleof an unstructured environment, Peer-to-peer (P2P) networks areconsidered. Those dynamic communities are affected by severalantisocial behaviours, such as free riding. Since this phenomenontypically causes and exacerbates an unbalanced and unfair use ofsystem resources, it has been considered as the case study in ourwork. The proposed solution exploits peer reputations in orderto define an incentive system, whose main goal is the dissuasionfrom free riding.
|Numero di pagine||5|
|Stato di pubblicazione||Published - 2008|
All Science Journal Classification (ASJC) codes
- Computational Theory and Mathematics
- Signal Processing
- Control and Systems Engineering