Performance analysis for a CDMA network based on statistical algorithms

Abstract

Traditional performance analysis aiming at OMC performance data mainly focuses on two aspects: longitudinal comparison of the same network element in different time periods and horizontal comparison of the different network elements in the same time. It's lack of association analysis between performance indexes and in-depth data mining. In this paper, a network performance analysis approach based on correlation analysis and regression analysis is presented which focusing on solving the difficult problem that general diagnosis algorithms can hardly build for CDMA network optimization. Based on the characteristic of performance indexes, the statistical method was introduced to performance optimization. The correlation coefficients and degree of curve fitting were proposed to measure the related degree between performance indexes, which is the foundation of fault diagnosis for engineers. With the practical application to the network optimization, it's proved that the statistical analysis is a novel and efficient tool for performance optimization and fault diagnosis.