Employing LRCN Model for Application Classification in SDN

The rapid growth of network technologies has complicated networking operations and network resource optimization. Software-Defined Network (SDN) overcomes the limitations of traditional networks by separating the forwarding layer and the control layer of the network. Application classification in SDN manages network by categorizing traffic into various application classes. Efficient application classification in SDN facilitates network operators to allocate resources efficiently for different services. However, current machine learning classifiers do not give high classification accuracy on large-scale data. This article proposes a new hybrid deep learning-based network application classifier—Long-Term Recurrent Convolutional Network (LRCN), which is a combination of Convolutional Neural Network (CNN) and the Long Short-Term Memory (LSTM) models. The proposed LRCN model leverages SDN’s robust computing potential and logically centralized control. The performance of the proposed model has been compared with prominent machine learning and deep learning techniques. The Unicauca (version 2.0) dataset has been pre-processed and used for experimentation purposes. The results infer that the proposed model is highly accurate, having an accuracy of 99.43%.