|
|
 |
| |
|
|
|
A Small Scale Forecasting Algorithm for Network Traffic based on Relevant Local Least Squares Support Vector Machine Regression Model |
|
|
|
PP: 653-659 |
|
|
|
Author(s) |
|
|
|
Tao Peng,
Zhoujin Tang,
|
|
|
|
Abstract |
|
|
| Real-time monitoring and forecasting technology for network traffic continues to play an important role in network
management. Effective network traffic prediction detects and avoids potential overload problems before they occur, which significantly
improves network availability and stability. Recent research has centered around Time Series Analysis based traffic prediction methods
that primarily extend Neural Network Forecasting (NNF) and Least Squares Support Vector Machine (LSSVM) algorithms, which are
not without their drawbacks. Given the vulnerabilities of existing nonlinear prediction methods in forecasting modeling, this paper
presents a novel, Relevant Local (RL) forecast method and its accompanying Pattern Search (PS) parameter-optimization approach
to introduce a new small-scale network traffic forecasting algorithm called RL-LSSVM (Relevant Local-Least squares support vector
machine). Furthermore, we demonstrate our new algorithm on network traffic data collected from wired campus networks and show
that the RL-LSSVM can effectively predict the small scale traffic measurement data while exhibit significantly improved prediction
accuracy than existing algorithms. |
|
|
|
|
|
|
|
