Rapid sensing-based emergency detection: A sequential approach

The upsurge of smart devices has enabled the realization of safe, efficient smart cities that improve the quality of life of their citizens. A prevalent class of smart city services that are attracting increasing attention are Smart Emergency Response and Management (SERM) systems, where sensing paradigms such as crowd sensing and IoT-centric sensing are employed to facilitate the detection of, and response to a crisis situation. In this paper, we study the detection of an abnormal change in a monitored variable through crowd sensed and heterogeneous data, where the change is suggestive of an emergency situation. We formulate our problem as a sequential change-point detection problem, where the underlying distribution of the variable changes at an unknown time. We aim to detect the change-point with minimal delay, subject to a false alarm constraint. We utilize Shiryaev's test to construct two variants of the solution depending on the structure of the received data contributions and mobility of participating sensing elements. We conduct simulations experiments to show the performance of these variants in terms of the delay-false alarm trade-off in different scenarios.