5G Network Slice Prediction using Hybrid Neural Network and Random Forest Model

5G networks are designed to be highly flexible and dynamic in order to meet the diverse needs and demands of users, such as faster internet, increased reliability, low latency and the ability to connect multiple devices. Network slicing, a prominent feature of 5G, enables flexibility by efficiently allocating resources through the creation of multiple virtual or logical connections on the same physical network structure. Recently, learning based models have found applications in various network slicing tasks such as predicting traffic, controlling slice admissions and efficiently allocating resources to make intelligent decisions. In this paper a novel hybrid model has been developed using the combination of a neural network and the random forest to predict the accurate network slice based upon network characteristics and user devices. The comparative performance analysis on a standard dataset shows that the proposed model performs better with an accuracy of 99.1% compared to other conventional deep learning algorithms in predicting the correct slice.