完全竞争均衡的群智感知定价机制研究  

Completely-competitive-equilibrium-based crowdsensing pricing mechanism

作　　者：李美璇 孙玉娥[2,3] 黄河 辛煜[4] 卜霄菲 LI Meixuan;SUN Yue;HUANG He;XIN Yu;BU Xiaofei(Department of Computer Science and Technology,Soochow University,Suzhou215006,China;Department of Urban Rail Transportation,Soochow University,Suzhou 215137,China;Suzhou Institute for Advanced study,University of Science and Technology of China,Suzhou 215123,China;Beijing Institute of Remote Sensing Information,Beijing China;College of Software,Shenyang Normal University,Shenyang 110034,China)

机构地区：[1]苏州大学计算机科学与技术学院,江苏苏州215006 [2]苏州大学城市轨道交通学院,江苏苏州215137 [3]中国科学技术大学苏州研究院,江苏苏州215123 [4]北京遥感信息研究所,北京100011 [5]沈阳师范大学软件学院,辽宁沈阳110034

出　　处：《中国科学技术大学学报》2019年第7期595-602,共8页JUSTC

基　　金：国家自然科学基金面上项目(61572342,61672369);江苏省自然科学基金(BK20151240,BK20161258)资助.

摘　　要：群智感知通过将任务分配给大量普通用户,能够完成大规模、复杂的社会感知任务,受到了国内外学者的广泛关注.其中,如何激励用户参与感知任务是群智感知中的重要问题.现有激励机制相关研究重点关注了如何设置任务的定价,从而激励用户提交高质量感知数据,但忽视了潜在的盲目报价问题;这极易导致参与任务执行的用户数量失衡,进而无法保证平台得到最优收益.为了解决这一问题,提出了完全竞争均衡的群智感知定价机制.该机制首先将平台与用户之间的多人博弈抽象为平台与市场间的双人博弈.再引入市场类型概率,通过海萨尼转换,将双人不完全信息博弈转化为双人完全不完美信息博弈.最后通过平台多轮重复博弈,使平台报价收敛到完全竞争均衡水平.理论分析和实验结果表明所提激励机制能够收敛到完全竞争均衡状态.Crowdsensing accomplishes extended general and complex social sensing tasks through allocating tasks to a large number of ordinary users(or workers), and has attracted extensive attention in recent years. How to motivate users to participate in sensing tasks is one of the most important issues in crowdsensing. However, the existing incentive mechanisms mainly focus on how to set prices to enable users to submit high-quality sensing data,ignoring the problem of blind quotes, which can easily lead to the imbalance of the number of users participating in the task execution, so that the platform cannot obtain the optimal revenue. To tackle this challenge, a completely-competitive-equilibriumcrowdsensing pricing mechanism is proposed. Firstly, the multi-player game between platform and users is abstracted as a two-person game between the platform and the market. Then the market type probability is introduced and the two-person incomplete information game is transformed into the two-person complete imperfect information game through Harsanyi transformation. Finally, through multiple rounds of repeated games on the platform, the platform′s price converged to completely competitive equilibrium. Theoretical analysis and experimental results show that the proposed incentive mechanism can achieve completely competitive equilibrium.

关 键 词：群智感知 完全竞争均衡 激励机制 任务分配 

分 类 号：TP393[自动化与计算机技术—计算机应用技术]