Hybrid deep learning models and link probability based routing in software defined-DCN

Abstract

In an SDN-based Data Center Network (SD-DCN) environment, network traffic is highly vulnerable to handle the dynamic requirements of the users. It creates difficulty for network management tasks and future network routing path prediction. Deep learning is an emerging technology available to analyze the network environment to enhance network performance. The advanced deep learning models can easily handle dynamic changes as compared to conventional mechanisms in the network environment. The Convolution Neural Network (CNN) deep learning model provides limited congestion features extraction. The Recurrent Neural Network (RNN) deep learning model provides long feature dependency information from the given traffic dataset. To overcome the limitations of CNN and RNN models, an efficient deep learning technique is required to find both congestion features and feature dependency from previous input data. Moreover, in this paper, the mathematical model is proposed which improves network routing and flow management mechanisms by using available link probability and computing average network delay in the network. The proposed mathematical model provides the average network delay and traffic dataset to CNN-RNN hybrid deep learning models to analyze the network traffic. The improvement of the Quality of Service (QoS) in the network is the major aim to incorporate a mathematical model in the proposed system. In the proposed work, the hybrid deep learning models are a combination of two deep learning models, i.e., CNN with Long Short Term Memory (LSTM) network as CNN-LSTM and Bidirectional Long Short Term Memory (BiLSTM) network as CNN-BiLSTM. The experimental results show that the proposed approach outperforms the existing work in terms of several metrics such as average network throughput, average network delay, packet loss rate, and routing overhead. Furthermore, the result comparison is analyzed for the training and testing error analysis of different deep learning models.