Adopting Long-Short Term Memory to predict reliability in different use cases in autonomous vehicles in 6G

In vehicular networks, reliability is a critical problem. This study suggests applying Deep Learning (DL) to automatically predict the reliability of vehicular networks including different types of components. First, a data set is generated using simulation parameters from our previous work. Next, a framework is proposed that use the deep learning method, Long-Short Term Memory (LSTM) to predict the networks’ reliability. The performance is compared with other machine learning techniques. The results showed, the proposed model outperforms other machine learning methods for reliability prediction. The performance is also evaluated excluding driver’s behavior and reaction. The results indicate that the reliability of the system decreases by $10 %$. Further, the performance was also evaluated excluding Prime Reliability Factor (PRF), code rate and coding. According to the results, reliability decreased by $3 %$ more.