Correlation Analysis Of KQI And KPI In Wireless Network Based On Time And Space

With the popularity of wireless network,intelligent network has a broad development prospect,and its most important part is data collection and analysis and prediction.At present,the focus of operators has shifted from the construction of wireless network to operation and maintenance,eager to provide users with the best service experience.This requires the analysis of wireless network data,the use of historical data to find out the factors that affect the user experience index,and can predict the future network performance and deterioration reasons.In this paper,through the spatial-temporal analysis of wireless network operation data,we explore the correlation analysis method of KPI(key performance indicator)and KQI(key quality indicator)of wireless network,in order to get the correlation between KPI and KQI,and provide guidance and basis for effectively improving KQI by optimizing some KPIs.The main work of this paper is as follows(1)This paper expounds the basic concepts and relationships of QoE(quality of experience),KQI,KPI and QoS(quality of service),designs MRS data extraction scheme,and provides support for data analysis and mining research related to wireless communication network.(2)Based on the theory of time series and time series analysis,the STL time series decomposition method is used to study the time characteristics of wireless network.Based on the analysis of the time characteristics of wireless network,a variety of prediction models are tried,and a combination model of Holt winters model and LSTM model is proposed,which achieves the optimal effect on the experimental data set.(3)This paper analyzes the spatial characteristics of the wireless network,considers the influence of other cells on the current cell,uses Pearson correlation coefficient,with the help of spatial correlation thermal graph,obtains the correlation degree between different cells under the same base station,finds out the cell with the largest correlation with the current cell,and then applies it to the multi cell correlation analysis model.(4)The machine learning model is used for the correlation analysis of KQI and KPI in wireless network.The experimental results show that the logistic regression model and random forest model are the best,which shows the effectiveness of the above models in the correlation analysis of KQI and KPI.(5)Apriori algorithm is applied to the association analysis of KQI and KPI in wireless network.A single cell association analysis model based on Apriori algorithm and a multi cell association analysis model based on improved Apriori algorithm are proposed.The experimental results are better than machine learning model.