Dynamic Cooperation Enforcement through Trust-based Allocation Policies in P2P Systems

Abstract

In this paper we propose distributed trust-based resource allocation policies to provide fairness in p2p-like systems. According to these policies, the portion of the available resource received by each competing peer depends on (a) his trust/reputation, (b) his resource demands and (c) his request generation rate. The reputation of a peer is quantified and an approach for continuously updating the reputation is proposed such that it reflects the contributions of the peer in the transactions he engages. We study both homogeneous and heterogeneous systems of peers with different resource capabilities and needs and we see from theoretical analysis and simulation results that proposed policies motivate peers to contribute resources in the network by guarantying that peers will only receive resources in proportion to their contributions; thus misbehaving (non contributive) peers cannot exploit the system. Furthermore, proposed policies lead to the dynamic formation of coalitions (cooperation) between peers who mutually benefit by their transactions, according to their capabilities and needs. Peers' coalitions are adaptive to network changes and self-organized as new peers enter the system or strategic peers vary their contributions.