| Popis: |
Owing to a rise in the number of vehicles on route, traffic has been increased dramatically in recent years. It has been noted that driving certain vehicles manually has become a challenging task. The challenges faced during an accident are the congestion of vehicles which blocks the ambulances during emergencies. Currently, traffic congestion is posing a major challenge for the public transportation system. To overcome these issues, a Priority federated learning with Multi-head CNN (PFL-MHCN2) has been proposed. Smart sensors and cameras equipped with IoT capabilities were used to collect data for the proposed model. Using this technique, signals from one junction are transmitted to another junction and updated in the cloud. A congestion spot is identified based on the input characteristics that are contained in the cloud and sensor data that are received. Initially, pre-processing reduces noisy values and predicts missing values in the acquired data. After pre-processing, the data are transferred to the detection layer. The Detection layer detects congestion free route using Federated Multi-head CNN. The proposed model is evaluated based on parameters like accuracy, precision, specificity, F1 score, and Miss rate. The proposed PFL-MHCN2 model produces a 0.65% miss rate, which is less than existing methods. The proposed technique improves the accuracy of 1.45%, 1.66%, and 4.32% better than TCC-SVM, TC2S-DNN and MSR2C-ABPNN respectively. |
