Multiresolution traffic prediction: combine RLS algorithm with wavelet transform

Authors:
Yanqiang Luan
Affiliations:
School of Electrical and Information Engineering, University of Sydney, Sydney, NSW, Australia
Venue:
ICOIN'05 Proceedings of the 2005 international conference on Information Networking: convergence in broadband and mobile networking
Year:
2005

Citing 7
Cited 1

On the self-similar nature of Ethernet traffic (extended version)

IEEE/ACM Transactions on Networking (TON)
Wide area traffic: the failure of Poisson modeling

IEEE/ACM Transactions on Networking (TON)
Self-similarity in World Wide Web traffic: evidence and possible causes

IEEE/ACM Transactions on Networking (TON)
The changing nature of network traffic: scaling phenomena

ACM SIGCOMM Computer Communication Review
Time Series Analysis, Forecasting and Control

Time Series Analysis, Forecasting and Control
On the approximate decorrelation property of the discrete wavelet transform for fractionally differenced processes

IEEE Transactions on Information Theory
ScaleNet-multiscale neural-network architecture for time series prediction

IEEE Transactions on Neural Networks

Comparative study of different wavelets for hydrologic forecasting

Computers & Geosciences

Quantified Score

Hi-index	0.00

Visualization

Abstract

Numerous research in the literature has convincingly demonstrated the widespread existence of self-similarity in network traffic. Self-similar traffic has infinite variance and long range dependence (LRD) which makes conventional traffic prediction method inappropriate. In this paper, we proposed a traffic prediction method by combining RLS (recursive least square) adaptive filtering with wavelet transform. Wavelet has many advantages when used in traffic analysis. Fundamentally, this is due to the non-trivial fact that the analyzing wavelet family itself possesses a scale invariant feature. It is also proved that wavelet coefficients are largely decorrelated and only has short range dependence (SRD). In this paper, We investigate the computation characteristics of discrete wavelet transform (DWT) and shows that the $\grave{a } \ trous$ algorithm is more favorable in time series prediction. The proposed method is applied to real network traffic. Experiment results show that more accurate traffic prediction can be achieved by the proposed method.