Improving Intrusion Detection in a Software Defined Network using Hybrid CNN and Bi-LSTM

A Software Defined Networking is a modern offering against traditional network systems to make network control more efficient by reducing the dependency on the hardware side. With a logical control plane, it is essential to distinguish between benign and malicious data traffic. We explore the possibility of a hybrid neural network architecture for intrusion detection and traffic monitoring in an SDN environment. The suggested architecture stacks a Convolutional Neural Network (CNN) with Bidirectional Long Short Term Recurrent Neural Networks (BiLSTM) to provide a specialized architecture for distinguishing the incoming network traffic patterns in SDN. The results emphasize the hybrid architecture’s effectiveness in detecting spatial and temporal features, contributing to improved network security in SDN environments.