A Web Service QoS Forecasting Approach Based on Multivariate Time Series

Abstract

In order to accurately forecast Quality of Service (QoS) of different Web Services, this paper proposes a novel QoS forecasting approach called MulA-LMRBF (Multi-step fore-casting with Advertisement and Levenberg-Marquardt improved Radial Basis Function) based on multivariate time series. Considering the correlation among different QoS attributes, we use phase-space reconstruction to map historical multivariate QoS data into a dynamic system, use Average Dimension (AD) to estimate the embedding dimension and delay time of reconstructed phase space. We also add the short-term QoS advertisement data of service provider to form a more comprehensive data set. Then, RBF (Radial Basis Function) neural network improved by the Levenberg-Marquardt (LM) algorithm is used to update the weight of the neural network dynamically, which improves the forecasting accuracy and realizes the dynamic multiple-step forecasting. The experimental results demonstrate that MulA-LMRBF is better than previous approaches in term of precision and is more suitable for multi-step forecasting.