群智感知中的参与者信誉评估方案

摘要

For a Mobile Crowd Sensing (MCS) network has a large group of participants,and the acquisition and submission of tasks are almost unrestricted,so that data redundancy is high and data quality cannot be guranteed.To solve the problem,a method called Participant Reputation Evaluation Scheme (PRES) was proposed to evaluate the data quality and the reputation of participants.A participant's reputation was evaluated from five aspects:response time,distance,historical reputation,data correlation and quality of submitted data.The five parameters were quantified,and a regression equation was established by using logistic regression model to get the participant reputation after submitting data.The reputation credibility of a participant was in the interval [0.0,1.0],and concentrated in [0.0,0.2] and [0.8,1.0],making it easier for the group of mental perception network to choose appropriate participants,and the accuracy of the evaluation results by the crowd sensing showed that PRES was more than 90％.%针对群智感知网络中参与者群体大,且获取和提交任务几乎不受限制,使得群智感知网络存在数据冗余度高和数据质量不能得到保证的问题,提出了针对参与者提交数据质量和可信度的信誉评估方案——参与者信誉评估方案(PLES).从参与者提交数据的响应时间、距离、历史信誉度、数据相关性和数据质量五个方面对参与者信誉进行评估,将这五个参数数值化,并利用逻辑回归模型建立回归方程,得出参与者本次提交数据后的信誉度.PRES得出的参与者信誉度在[0.0,1.0]范围内,且集中分布于[0.0,0.2]和[0.8,1.0]区间,使得群智感知网络容易选择合适的参与者,且评估结果表明PRES评估的准确率均在90％以上.