Whether for the purpose of crowd management or economic incentives, one of the most important (and certainly the most tangible) indicators of a crowd is its size. When access to an event is restricted (e.g., through tickets or turnstiles at access points), counting a crowd is trivial. For open-access events, however, this is much more challenging. Additionally, crowd size estimations often differ from reality due to subjectivity or contradicting motives of the different stakeholders [6]. Given that the success of an organized event (e.g., a protest march) is often measured by its attendance, organizers may for example be tempted to exaggerate attendance figures in order to put more weight on public opinion.

Perhaps one of the most telling examples is that of the Million Man March held in 1995 in Washington DC, where depending on the source, crowd size estimations varied between 400,000 and 1.5�C2 million.Public safety cannot afford such margins of error and requires objective and accurate crowd density figures. While various methodologies (an overview is given in Section 2) have been suggested to estimate the size of a crowd in an objective manner, they often entail high levels of uncertainty and are impractical when applied in scenarios with high levels of movement. The mapping of a crowd, for which information is needed on the specific location or sequence of locations of individuals within the crowd, is even more complex than just counting and requires more advanced methodologies.

In this paper, we will present an alternative methodology for counting and mapping a crowd based on the Bluetooth technology.

The usefulness of our approach will be illustrated in a case study where spectators of a road cycling race are mapped using Bluetooth sensors installed on a mobile platform moving along Batimastat the track, delivering detailed spatiotemporal information on the crowd assembled for the sporting event. In Section 2, we discuss the current methodologies used to count and/or map crowds, their most important deficiencies and how Bluetooth technology might offer a valuable alternative��especially when individual mobility needs Brefeldin_A to be accounted for. Subsequently, we present a Bluetooth tracking methodology and its specific deployment in this paper in Section 3.

In Section 4, we then give background information on the case study (Tour of Flanders 2011). The results of an experiment carried out prior to the cycling race, and the main case study experiment itself are outlined in Section 5. Finally, we interpret and discuss these results (Section 6) and give a short conclusion (Section 7).2.?Counting and Mapping a CrowdDifferent methodologies have been proposed to estimate crowd sizes.