IoT-Driven Vehicle Management: Fully Elman Neural Network with Red Piranha Optimization-Based Drowsiness and Alcohol Consumption Detection in Smart Cities

Recent research has focused on supporting drivers, revealing that the primary causes of road accidents are driver drowsiness and alcohol consumption. Thus, Drowsiness and alcohol consumption detection (DACD) are critical for IoT-based smart cities as they improve public safety by detecting and preventing incidents related to sleep and alcohol consumption. In this manuscript, an AI-enabled DACD using Fully Elman Neural Network (FENN) with Red Piranha Optimization (RPO) is proposed for Internet of Things (IoT) based smart cities. Initially, the IoT kit consists of several normal cars, ambulance cars, and roadside devices. The roadside devices which are transceivers fixed at predetermined locations, relay information to both normal and ambulance car devices. The system is designed to detect alcohol consumption, and driver drowsiness using data for each vehicle in the initial setup. The data collected by the IoT kit is preprocessed using the MaxAbsScaler Normalization approach. After that the deep learning model, specifically using FENN is applied in the preprocessed data to validate the detection results. Also, Red Piranha Optimization (RPO) is proposed for enhancing the weight parameters of FENN. By then the performance of the proposed FENN-RPO-DACD method is evaluated using the MATLAB platform, and the the performance evaluation is analysed using calculations like accuracy, False Positive Rate (FPR), Sensitivity, False Negative Rate (FNR), Precision, Recall, F-1 Score, Specificity, computational time. Thus, the proposed FENN-RPO-DACD method has achieved 18.98%, 21.56%, and 24.96% higher accuracy, 12.39%, 19.56%, and 29.67% lower Computation Time, 28.78%, 34.09%, and 38.67% lower FPR, 14.98%, 18.67%, and 21.09% higher sensitivity, 18.97%, 21.56%, and 24.38% higher precision than other conventional techniques like O-SNN-DADSS, AI-SIoT, and CNN respectively.