A Multi Soft-sensor Approach for the Development of Emergencies Detection Units on the Edge

The adoption of multi-sensor hardware units to detect different types of emergencies in complex scenarios such as Industry 4.0 and Smart Cities is a recent development trend, which have demanded research efforts to optimize the way these units are assembled, configured, and deployed. However, although their expected advantages, usual multi-sensors emergency-oriented approaches may be constrained by the tight association between physical sensor devices and the emergency detection process. Therefore, this paper proposes an innovative approach to concurrently detect multiple emergencies exploiting the concept of soft-sensor modules, which are implemented on Emergency Detection Units (EDUs) as concurrent tasks along with a high-priority emergency alerting task. The soft-sensor concept defines physical sensors as ordinary input data streams, which will be processed according to the defined algorithms by each particular soft-sensor, allowing detection decisions to be performed on the edge. In order to support the practical exploitation of this concept, a formal specification based on Petri Nets is defined, modeling the concurrent operation of the tasks within an EDU. A proof of concept is also designed based on the affordable Raspberry Pi Pico microcontroller, demonstrating how the proposed approach would behave when two soft-sensors are implemented in a typical detection unit.