StareCrowd: A New Paradign for Mobile Crowdsensing

随着具有丰富传感器外设和计算能力的移动设备的激增，移动群体感知在实现各种环境、商业和社会应用方面显示出了希望。移动群体感知是社区使用移动设备贡献感知信息和人类智能的地方。然而，现有的移动众感解决方案受到众所周知的冷启动问题的影响，在这个问题上，合格贡献者的短缺使得很难提取任何有意义的信息并据此做出决定。此外，移动人群感知在处理来自资源受限的移动设备的噪声和错误数据、最大限度地减少用户干预以及实时数据采集、处理和分析方面提出了巨大的挑战。该项目的目标是开发一种新的人群感知范例，ShareCrowd，以编程方式促进感知和用户贡献的数据的共享和聚合。ShareCrowd是一个双面平台，允许两个不同的用户群体，即跨多个应用程序的数据请求者和数据贡献者，相互提供净收益。为了从大众感知的数据中提取有用的信息并检测数据篡改，将开发新的数据分析技术。为了解决资源限制，移动众感成本将以支付系统的形式在ShareCrowd的激励机制设计中明确考虑。基于主动学习的数据查询技术将减少对用户干预的需要，这种技术仅在必要时获得更多数据和注释。移动人群感知将传统的物联网扩展到以消费者为中心的移动设备，这些物联网将嵌入传感器的物理物品互联到已经部署在现场的移动设备。该项目解决了释放移动人群感应力量的关键障碍，从而可以提高我们的行业合作伙伴和其他组织的现有服务和业务的效率，并通过为加拿大的Apps经济价值链做出贡献来创造新的收入形式。

With the proliferation of mobile devices with rich sensor peripherals and computation capability, mobile crowdsensing, where communities contribute sensing information and human intelligence using mobiledevices, has shown promises in enabling a variety of environmental, commercial and social applications. Existing solutions to mobile crowdsensing, however, suffer from the well-known cold start problem, where the shortage of eligible contributors makes it difficult to extract any meaningful information and make decisions accordingly. In addition, mobile crowdsensing poses great challenges in handling noisy and faulty data from resource-constrained mobile devices, minimizing user intervention, and real-time data acquisition, processing, and analytics. The goal of this project is to develop a novel crowd sensing paradigm, ShareCrowd, that facilitates sharing and aggregation of sensing and user-contributed data programmatically. ShareCrowd is a two-sided platform and allows two distinct user groups, namely, data requesters and data contributors across multiple apps to provide each other with net benefits. To extract useful information from crowdsensed data and to detect data falsification, novel data analytical techniques will be developed. To address resource limitation, mobile crowdsensing costs will be explicitly considered in the incentive mechanism design of ShareCrowd in the form of a payment system. The need for user intervention will be mitigated by active learning-based data query techniques, which obtain more data and annotations only when necessary. Mobile crowdsensing extends the traditional IoT that interconnects physical items embedded with sensors to consumer-centric mobile devices that are already "deployed in the field". The project tackles the key roadblocks to unleashing the power of mobile crowdsensing, and thus can improve the efficiency of existing services and businesses of our industrial partners and other organizations, and generate new forms of revenue by contributing to the Apps economy value chain in Canada.