移动群智感知中发掘潜在高质量用户的激励机制  

Incentive Mechanism for Exploring Potential High Quality Users in Mobile Crowd Sensing

作　　者：江海峰[1] 商景杰 王树豪 张寿军 JIANG Haifeng;SHANG Jingjie;WANG Shuhao;ZHANG Shoujun(School of Computer Science and Technology,China University of Mining and Technology,Xuzhou 221000,China;Engineering Research Center of Digital mine Ministry of Education,China University of Mining and Technology,Xuzhou 221000,China)

机构地区：[1]中国矿业大学计算机科学与技术学院,江苏徐州221000 [2]矿山数字化教育部工程研究中心(中国矿业大学),江苏徐州221000

出　　处：《小型微型计算机系统》2025年第3期620-626,共7页Journal of Chinese Computer Systems

基　　金：国家自然科学基金项目(62071470)资助.

摘　　要：在移动群智感知的激励机制中,用户的感知质量和能力是重要的指标,对不同类型的任务是有差异的,用笼统的感知质量与能力标准选择用户往往会埋没潜在的高质量用户.针对这一问题,本文将用户的感知质量与能力根据任务的不同类型进行细分,在用户感知质量与能力未知的情况下,将用户选择问题建模成反向拍卖与多臂赌博机模型,不断学习与更新用户的感知质量与能力值,使用置信区间上界的方法估计用户的感知质量,并将其与用户的能力和报价作为选择用户的标准,提出了基于置信区间上界的质量与能力并驱的激励机制.当能力值均值达到平台规定的阈值时,用户将拥有招募其他用户的权限,并从其招募的用户完成的任务中获得额外的收益.本文证明了该激励机制满足计算有效性、真实性和个体理性.仿真实验结果表明,本文所提的激励机制在用户平均效用、任务平均质量和不同任务类型高能力值用户占比等方面具有良好的性能.In the incentive mechanism of mobile crowd sensing,users′perceived quality and ability are important indicators,which are different for different types of tasks.Selecting users with general criteria of perceived quality and ability will often bury potential high-quality users.In order to solve this problem,this paper subdivides the user′s perceived quality and ability according to different types of tasks.When the user′s perceived quality and ability are unknown,the user selection problem is modeled as reverse auction and multi-arm gambling machine model,and the user′s perceived quality and ability values are continuously learned and updated.The upper confidence bound is used to estimate the user′s perceived quality,and the user′s ability and quotation are used as the criteria for selecting users.An incentive mechanism based on the upper bound of confidence interval is proposed.When the average value of ability reaches the threshold specified by the platform,users will have the right to recruit other users and get additional benefits from the tasks completed by the recruited users.This paper proves that the incentive mechanism meets the calculation validity,authenticity and individual rationality.This paper proves that the incentive mechanism satisfies computational effectiveness,authenticity,and individual rationality.The simulation experimental results show that the incentive mechanism proposed in this paper has good performance in terms of average user utility,average task quality,and the proportion of users with high capability values for different task types.

关 键 词：移动群智感知 激励机制 多臂赌博机 置信区间上界 

分 类 号：TP393[自动化与计算机技术—计算机应用技术]