Published Paper Details:

This study examines the growing complexity of modern telecommunication networks and the increasing risks associated with cyber threats. With the expansion of technologies such as 5G, cloud computing, and IoT, network infrastructures have become more vulnerable to attacks like intrusions and service disruptions. Traditional intrusion detection approaches often fail to identify new and evolving threats due to their reliance on predefined rules Traditional intrusion detection systems (IDS) rely on static rule-based and signature-based mechanisms that struggle to cope with modern cyber threats. These systems are limited in their ability to detect unknown attacks, adapt to dynamic traffic patterns, and scale with increasing data volumes. Furthermore, high false alarm rates significantly reduce operational efficiency and increase the workload of network administrators. This research proposes a Artificial Intelligence driven anomaly detection framework tailored for telecom network environments. The proposed system integrates advanced preprocessing techniques, hybrid feature selection using Weighted Adaptive Feature Selection (WAFS), Synthetic Minority Oversampling Technique (SMOTE) for data balancing, and a hybrid ensemble model combining a Deep Neural Network (DeepAnomNet) and Random Forest classifier. The UNSW-NB15 dataset is employed for training and evaluation as it represents modern network traffic and diverse attack scenarios. The model is evaluated using multiple performance metrics including accuracy, precision, recall, F1-score, and ROC-AUC. Results demonstrate significant improvements in detection capability and reduction in false positives compared to conventional machine learning models. A real-time Telecom Security Information and Event Management (SIEM) dashboard is also developed to visualize anomaly probability and threat severity. The proposed system offers a scalable, adaptive, and intelligent solution for securing next-generation telecom infrastructures.