Enhancing Secure Communication in IoT-Based Automated Vehicle Systems through Accurate Prediction of Abnormal Traffic Data

Ensuring secure communication among Automated Vehicles (AV) on the Internet of Things (IoT) applications requires
accurate prediction of abnormal traffic data. Information security within vehicles is crucial for transmitting traffic-related information reliably and guiding vehicles along the correct path. Existing algorithms for classification and prediction have aimed to provide clear predictions of malicious data. However, traditional techniques have faced challenges in achieving both
accuracy and computational efficiency. This study proposes a three-stage implementation to address these issues. The first phase involves pre-processing to reduce computational complexities. This initial step streamlines the data for further analysis. The processed data then undergoes feature selection in the second phase, employing a multi-GGA (Greedy Genetic Algorithm) approach to identify the most relevant features. By utilizing this algorithm, the system can detect significant information even in the presence of misleading data. Finally, the third phase involves classification using a combination of Random Forest (RF) and AdaBoost algorithms. This integrated approach enables the system to distinguish between normal and abnormal traffic data in
vehicle-to-vehicle datasets. Through experimental evaluations and comparative analysis, the efficiency of the proposed system
is demonstrated in terms of accuracy, precision, recall, and F1-score, outperforming several existing algorithms. Overall, this
proposed prediction system shows great potential in effectively classifying misleading and normal data with high accuracy.
Addressing the limitations of traditional techniques offers a reliable solution for secure communication and decision-making in
AV systems