ABSTRACT The accurate measurement of human activity with high spatial and temporal granularity is crucial for understanding the structure and function of the built environment. With increasing mobile ownership, the Wi-Fi ‘probe requests’ generated by mobile devices can act as a cheap, scalable and real-time source of data for establishing such measures. The two major challenges we face in using these probe requests for estimating human activity are: filtering the noise generated by the uncertain field of measurement and clustering anonymised probe requests generated by the same devices together without compromising the privacy of the users. In this paper, we demonstrate that we can overcome these challenges by using class intervals and a novel graph-based technique for filtering and clustering the probe requests which in turn, enables us to reliably measure real-time pedestrian footfall at retail high streets.

中文翻译：

---

根据 Wi-Fi 探测请求估计实时高街客流量

摘要 以高时空粒度准确测量人类活动对于理解建筑环境的结构和功能至关重要。随着移动设备所有权的增加，移动设备生成的 Wi-Fi“探测请求”可以作为建立此类措施的廉价、可扩展和实时数据源。在使用这些探测请求来估计人类活动时，我们面临的两个主要挑战是：过滤不确定测量领域产生的噪声，以及在不损害用户隐私的情况下将同一设备生成的匿名探测请求聚集在一起。在本文中，我们证明我们可以通过使用类间隔和一种新颖的基于图的技术来过滤和聚类探测请求，从而克服这些挑战，