A fusion model of SWT, QGA and BP neural network for wireless network traffic prediction

Abstract

In this paper a fusion model by combining the Stationary Wavelet Transform (SWT), Quantum Genetic Algorithm (QGA) and Back-propagation (BP) Neural Network is proposed to forecast wireless network traffic. In order to achieve guaranteed Quality of Service (QoS) in wireless networks, various managing measures can be taken only by knowing the network traffic in advance. This developed fusion model which is called the SWT-QGA-BP model can be efficiently used to assess the future network and provide adequate evidence for wireless network management. By using SWT, the original non-stationary wireless traffic data are transformed into multiple stationary components. With the QGA evolution, the BP neural network is optimized in both architecture and initial parameters. The simulation further indicates the effectiveness of the proposed SWT-QGA-BP fusion model, and the results show that our model can enhance the prediction performance significantly in accuracy.

Linear and Non-Linear Spatio-Temporal Input Selection In Wireless Traffic Networks Prediction using Recurrent Neural Networks

“…Optimization is achieved by preventing congestion, detecting network traffic instabilities, monitoring response, and allocating system resources. Its purpose is to decrease costs and enhance bandwidth services due to persistent high system demands [1].…”

“…The Multivariate Statistical Network Monitoring (MSNM) method is additionally used for detection intrusions and anomalies [5], the Deep Belief Networks (DBN) method for predicting network traffic data [6], and the Fuzzy Machine Neural Networks (FMNN) method for regulation of network traffic. Alternatively, use a combination model of Stationary Wavelet Transform (SWT), Quantum Genetic Algorithm (QGA), and Backpropagation (BP) to achieve higher service levels with a lot of efficient prediction model [1].…”