点击次数：

论文类型：期刊论文

发表时间：2016-11-15

发表刊物：KNOWLEDGE-BASED SYSTEMS

收录刊物：SCIE、EI、Scopus

卷号：112

页面范围：105-113

ISSN号：0950-7051

关键字：Incentive mechanism; Dynamic reward; Replicator dynamics; Evolutionary game

摘要：Reward-based incentive mechanisms are used most commonly to promote cooperation in peer-to-peer (P2P) systems. Specifically, contributing agents are rewarded by the system. For a centralized P2P system, the central server is responsible for implementing the mechanism and rewarding those cooperative agents. Thus, providing incentives could be costly for the system server. In this paper, we propose a dynamic reward mechanism in which the reward changes accordingly as the system evolves, i.e., when there are more cooperative agents, the reward for each agent decreases, and when there are no free-riders, rewards are no longer needed. Through theoretical analysis using replication dynamics, we determined the evolution stable strategy (ESS) for different scenarios and qualitatively proved that our dynamic reward mechanism can reduce the overall cost of the system. The simulation proved that this dynamic reward mechanism can promote cooperation and reduce the cost of the system. (C) 2016 Elsevier B.V. All rights reserved.