The present digital era facilitates advanced services with the widespread use of the internet to society. Though the usage of the internet increased, the vulnerable threats through this interconnection were also increasing everyday. Therefore, maintaining effective and transparent network security becomes a significant challenge. However, the need for an intrusion detection system is very essential to detect and understand these threats. Thus, Explainable Artificial Intelligence (XAI) offers a clear transparency, and comprehensibility to the decision-making process of a model, thereby enhancing trust and facilitating human interpretability. Hence, to secure network security, this research focuses on the usage of Local Interpretable Model-agnostic Explanations (LIME), XAI-LIME technique with the deep learning (DL)-based models for classification and detection. Deep learning models such as Deep Neural Network (DNN), Recurrent Neural Net- work (RNN) and Capsule Network (CapsNet) are used that have the ability to handle complex patterns, detecting and classifying network traffic into normal and abnormal activities. For this experimentation, we used the CICIDS-2017 dataset and conducted both performance metrics and interpretability. The proposed methodology not only achieves the accuracy but also makes it easier with LIME explanations and transparent to human understanding, thereby making the network security protected from real-world scenarios.