5G NETWORK ACCESS SECURITY MODEL THROUGH DEEP NEURAL NETWORKS CLUSTERING

Considering that a problem in the security of access to 5G networks is DOS attacks due to its orientation to IoT, an alternative security model is proposed that provides a solution to this problem and that requires little information from users, and is easy to use, train, configure, with little processing and high portability. This research proposes a security model for 5G Networks wireless access (5GDoSec) intended to detect possible intruders and malicious users through the use of Deep Neural Networks and the machine learning technique; this is based on the access data collected from a delimited entrance point that groups, identify and classify the authenticated users in the network to detect, based on the access numbers and the active time, the ones that could represent a threat. The 5GDoSec model follows an evolutive character proved to be reliable when classifying hazardous users showing better performance in its validation by DaviesBouldin than other techniques such as Kmeans and Linkage.