A cluster-based approach to cooperative mobile crowd monitoring

了解人群出现的位置对于许多应用程序至关重要，例如，要有效管理应急人员、志愿者或公共交通等资源。移动电话感知可能是获取实时人群信息的一种廉价而灵活的方式，因为它允许社区使用他们的移动电话收集传感器数据，否则需要部署专用的感知基础设施。然而，个人频繁将传感器数据上传到远程服务器的非合作方法会给参与的手机的网络基础设施和电池带来压力。对等网络和无线传感器网络的研究采用了设备到设备通信、分布式节点分簇、作业轮换、占空循环和分布式感知等协作概念，解决了流量分流和能量节约问题。然而，尚不清楚：1)这些概念应用于移动人群监控时有多合适，2)这些概念在移动电话上如何相互结合，以及3)它们共同应用时会产生什么影响。合作的手机感知，即个人将他们的数据联网并捆绑在一起，才刚刚开始出现，目前的方法只能部分解决这些研究问题。特别是，现有的解决方案是为小规模合作设计的，还不能适应人群的动态，或者没有同时研究多个概念的影响。本项目将通过迭代分析不同的集群和节能技术在移动电话、大规模模拟自组织网络和真实人群场景中的成本和收益来研究研究问题。它将开发一种用于协作移动人群监控的能效和自适应集群方案CONIC。通过最终的原型和量化数据，该项目希望有助于更好地理解如何通过能源受限的临时移动电话网络来监控人群。此外，它希望为研究人员和从业者提供证据，以支持他们关于移动传感应用的可行方法的决定。从长远来看，该项目将为基于人群的传感及其融入智慧城市概念的进一步研究奠定基础。

Knowing where crowds of people emerge is essential for many applications, e.g., to efficiently manage resources like emergency response personnel, volunteers, or public transport. Mobile phone sensing may be an inexpensive and flexible way to obtain real-time crowd information as it lets the community collect sensor data with their mobile phones where otherwise dedicated sensing infrastructure needs to be deployed. However, non-cooperative approaches where individuals frequently upload sensor data to a distant server strain the network infrastructure and the battery of the participating phones. This leads to frustration, less participation, and reduced information accuracy.Research on peer-to-peer and wireless sensor networks has addressed traffic off-loading and energy conservation with cooperative concepts like device-to-device communication, decentralised node clustering, job rotation, duty cycling, and distributed sensing. It is, however, unclear: 1) how suitable these concepts are when applied to mobile crowd monitoring, 2) how these concepts integrate with each other on mobile phones, and 3) what their impact is when collectively applied. Cooperative mobile phone sensing, where individuals network and bundle their data, has only started to emerge and current approaches addresses these research questions only partially. In particular, existing solutions are designed for small-scale cooperation, are not yet adaptive to the dynamics of the crowd, or do not study the effect of multiple concepts simultaneously. This project will investigate the research questions by iteratively analysing the costs and benefits of different clustering and energy conservation techniques on mobile phones, large-scale simulated ad hoc networks, and in real crowd scenes. It will develop Conic, an energy-efficient and adaptive clustering scheme for cooperative mobile crowd monitoring. With the resulting prototype and quantitative data, the project hopes to contribute to a better understanding of how a crowd of people can be monitored with a transient network of energy-constrained mobile phones. Further, it hopes to provide researchers and practitioners with evidence to support their decision on a viable approach for mobile sensing applications. In the long run, the project will establish the basis for further research in crowd-based sensing and its integration into smart city concepts.