The detection of critical situations through the adoption of multi-sensor Emergency Detection Units (EDUs) can significantly reduce the time between the initial stages of urban emergencies and the actual responses to relieve its negative effects, usually through the rescuing of endangered people, the attending to eventual victims, and the mitigating of its causes. However, although the benefits of such units are well known, their proper positioning in a city is challenging when considering a limited set of available units. In this sense, data-driven approaches can be leveraged to provide a better perception of the urban environments under consideration, allowing emergency management systems to be tailored to the specificities of a target city, thus improving the positioning of EDUs. This article proposes the processing of geospatial data of emergency-related urban infrastructure to support the computing of risk zones in a city, which is retrieved from the OpenStreetMap database together with the map of streets within a defined area. Since risk zones indirectly indicate the proportional number of detection units to be deployed, for each configuration setting of the EDUs, we propose an algorithm that computes the positions for such units only on streets, in a balanced way. Furthermore, considering that EDUs are expected to report detected emergencies through a wireless connection, we have also modelled the coverage area of existing networks in a city, which is also processed according to a suitable dataset. The proposed algorithm performs a fine-grained positioning of EDUs based on the number of active networks, flexibly favouring the EDUs' connectivity requirements such as reliability, throughput, latency, and transmission costs according to the actual demands of any urban emergency management system. Experimental results with real data demonstrated the applicability of the proposed mathematical model and the associated algorithm, reinforcing its practical application for the planning and construction of smart cities.

通过采用多传感器紧急检测单元（EDU）来检测危急情况，可以显着缩短城市紧急情况的初始阶段和实际响应之间的时间，以减轻其负面影响，通常是通过救援濒危人群、参与救援最终的受害者，以及减轻其原因。然而，尽管此类单位的好处众所周知，但在考虑有限的可用单位时，它们在城市中的适当定位具有挑战性。从这个意义上说，可以利用数据驱动的方法来更好地感知所考虑的城市环境，从而使应急管理系统能够根据目标城市的具体情况进行定制，从而改善教育单元的定位。本文提出处理与紧急情况相关的城市基础设施的地理空间数据，以支持城市风险区域的计算，这些数据是从 OpenStreetMap 数据库中与定义区域内的街道地图一起检索的。由于风险区域间接指示要部署的检测单元的比例数量，因此对于 EDU 的每种配置设置，我们提出了一种算法，以平衡的方式仅计算这些单元在街道上的位置。此外，考虑到EDU需要通过无线连接报告检测到的紧急情况，我们还对城市现有网络的覆盖区域进行了建模，并根据合适的数据集进行处理。该算法根据活跃网络的数量对EDU进行细粒度定位，根据城市应急管理系统的实际需求，灵活地满足EDU的可靠性、吞吐量、延迟和传输成本等连接要求。真实数据的实验结果证明了所提出的数学模型和相关算法的适用性，增强了其在智慧城市规划和建设中的实际应用。