J4

• 计算机科学 • Previous Articles     Next Articles

Improvement to Forecasting Algorithms for Performance of Resource in NWS

ZHANG Hai-yang, JU Jiu-bin, HU Liang   

  1. College of Computer Science and Technology, Jilin University, Changchun 130012, China
  • Received:2004-06-16 Revised:1900-01-01 Online:2005-03-26 Published:2005-03-26
  • Contact: JU Jiu-bin

Abstract: In this paper a new forecasting method, the dynamic exponential smoothing algorithm is implemented, which is derived form the exponential smoothing algorithm. The parameters in the new method can be regulated dynamically, which will make the method more effective in the NWS system. This new for ecasting method is different from the existing exponential smoothing algorithm, which can also regulate parameters, and it may be useful in the time series of NWS and other time series with the same statisticcharacteristics. Compared with the existing method in NWS, this method is better in terms of time complexity, s pace complexity and forecasting accuracy.

Key words: network weather service, CPU availability, bandwidth TCP, time series analysis, exponential smoothing algorithm, dynamic exponenti al smoothing algorithm

CLC Number: 

  • TP393