环境信息监控中基于压缩感知的移动众包成本控制研究

Research on Cost Control of Mobile Crowdsourcing Based on Compressive Sensing in Environmental Information Monitoring

基于压缩感知的移动众包模式是解决环境信息监控中成本问题的有效途径.压缩感知能够通过部分采样点恢复出全部数据,而其恢复质量取决于采样点所包含信息与噪声的数量.本文针对这两方面对压缩感知进行了优化,从而进一步减少环境信息监控所需成本.首先,本文提出了一种基于经验的采样点选择算法EBCS(Experience Based Cell Selection),通过选取包含信息更多的采样点,减少了数据恢复所需要的采样点数量.其次,本文提出了一种改进的k-means算法IK(Improved K-means),对参与者提交的任务数据中可能存在的伪造数据进行检测,避免了众包平台为了抵消伪造数据对恢复算法性能造成的负面影响而不得不对更多的采样点进行采样.经实验证明,本文提出的方法在成本控制上有非常好的表现.

Mobile crowdsourcing based on compressed sensing is an effective way to solve the cost problem in environmental information monitoring.Compressed sensing can recover all the data through partial sampling points and its recovery quality depends on the amount of information and noise contained in the sampling points.In this paper,compressed sensing is optimized for these two aspects to further reduce the cost of environmental information monitoring.First of all,this paper proposes an experience based cell selection algorithm EBCS(Experience Based Cell Selection),which reduces the number of sampling points needed for data recovery by selecting sampling points with more information.Then,this paper proposes an improved k-means algorithm IK(Improved K-means),which is used to detect the possible forged data in the task data submitted by the participants so that the crowdsourcing platform doesn′t have to sample more sampling points in order to offset the negative impact of forged data on the performance of the recovery algorithm.Experimental results show that the proposed methods have a good performance in cost control.