LoRaWAN-Based Artificial Intelligence Analytics for Forecasting Cyclone-Induced Floods on Urban Roads

Abstract

On urban roads, frequently severe floods were caused by monsoon or tropical cyclones, and oceans or several lakes and rivers overflowed. These cyclone-induced floods had massive loss and damage to road infrastructure, vehicles, and human life. In addition, many homes and businesses were destroyed, transportation disruption, power outages, and the spread of waterborne diseases such as dengue and malaria. This research proposes LoRaWAN-based artificial intelligence (AI) analytics for forecasting cyclone-induced floods to address the abovementioned destructions. The LoRaWAN network uses sensor devices in flood-prone areas to extract hydrological and weather conditions. The proposed analytics compares the sensor-measured data with past pre, during, and post-flood conditions and forecasts for future flood disasters. NVIDIA Jetson platform is used for performing image analytics and other factor inferences. Hyper-heuristic-based Convolution Encoder-Decoder using Gated Recurrent Units with attention mechanism (ConvED-GRUAT) is used to intelligently analyze satellite imagery and hydrological and meteorological factors to detect changes indicative of flooding. The proposed prediction model is trained on historical flood data to identify patterns and predict the likely impacts of a future flood event. The prediction model may look for rapid changes in water level or changes in the conductivity of the water. Once a flood is forecasted, the model can also be used to predict the severity of the flood, which can help to take preventative action and vehicle rerouting. Overall, the proposed solution accurately and timely estimates the climate and hydrological changes indicative of flooding and forecasts future flood tragedies, which can help to mitigate the effects of floods. Radiofrequency (RF) channel quality of service parameters such as RSSI, SNR, CR, and SF are considered to guarantee flood forecasting reliability and accuracy.