Research On The Complaint Warning Problem In LTE And WLAN

Complaint is the most direct feedback of customer satisfaction to network quality,the appearance of complaint can confirm the deterioration of network health condition,and provide important guidance for network coverage and network optimization.By correlating the network KPI data with the customer complaint data,we can explore the change of network quality from the point of view of complaint,and on the other hand,we can find out the main reasons for the complaint from the changes of network metrics.Research on the customer complaint warning issue is helpful for network Operation Department to solve network problem,improve customer satisfaction and improve profit of network operation department.The current research focuses on the two kinds of network environments such as LTE and WLAN,analyzes the research background,research significance and research status,elaborates the related communication basic theory and machine learning algorithm in the research of the complaint warning.On the one hand,this article aims at LTE network,we associate the network KPI data with the customer complaint data and establish a Novel Warning Complaints Model in LTE,before the customer complains to discover the problems appearing in the LTE network and realizes the cell-level customer's complaint warning.The complaint warning model in LTE includes three parts:data association,sample screening and complaint prediction.The Data Association part correlates the network KPI data with the customer complaint data in LTE through time and CI.The sample screening part is based on the neighborhood algorithm to modify the sample data.The complaint prediction part is based on the BP neural network algorithm.We use Python to simulate the model of the complaint early warning in LTE network environment and compare the complaint warning model with traditional classification models in the experiment.The comparison results show that the complaint warning model in LTE network environment is better than the traditional model in predicting precision and the prediction effect is good,and can reach 86,9%.On the other hand,this article also aims at WLAN network,we associate the network KPI data with the customer complaint data and establish a Novel Warning Complaints Model in WLAN,before the customer complains to discover the problems appearing in the WLAN network and realizes the AP-level customer's complaint warning.The complaint warning model in WLAN includes three parts:data association,sample screening and complaint prediction.The Data Association part correlates the network KPI data with the customer complaint data in WLAN through longitude and latitude data and CI.The sample screening part is also based on the neighborhood algorithm to modify the sample data.The complaint prediction part is based on the BP neural network and the AdaBoost algorithm.We use Python to simulate the model of the complaint early warning in WLAN network environment and compare the complaint warning model designed in LTE network with the new complaint warning model designed in LTE network.The comparison results show that the predicting precision of the new complaint warning model in WLAN network environment is improved and the prediction effect is good,and can reach 91.6%.