Research On Base Station Traffic Forecast And Coverage Optimization Algorithm Based On Big Data Mining

With the development of network technology and the widespread application of the Internet,wireless networks not only need to meet the growing demand for mobile data,but also provide users with high-quality services.Facing the rapid increase of user data and complex network,how to extract information and optimize network coverage is a challenging and critical research task.However,the traditional network optimization method is hard to extract information effectively in the face of huge data sets,which results in a certain lack of accuracy.Therefore,this thesis deeply studies the coverage optimization of LTE network with data mining method.This thesis mainly focuses on the following three aspects of base station traffic forecasting and base station network planning:(1)Proposing a base station traffic prediction algorithm based on neural network.Firstly,analyzing the temporal and spatial statistical characteristics of base station traffic sequence to extract the distribution characteristics,and dividing the base station traffic sequence into ordinary user base station traffic sequence and high volatility base station traffic sequence through volatility.Next,different prediction algorithms are proposed for different base station traffic sequences.The autoregressive prediction model is used to predict the steady traffic sequence of ordinary user base stations.For high volatility base stations through traffic pattern clustering and LSTM network,a base station traffic prediction model based on Trend-PCCs-LSTM is proposed.Finally,the prediction algorithm is applied to the actual base station traffic data set for simulation.According to the experimental results of the live network data,the algorithm has a higher accuracy improvement compared with the traditional prediction algorithm.(2)In order to figure out the problem of weak coverage and improve network quality,a research on LTE network weak coverage analysis and coverage optimization based on big data mining is carried out.Firstly,the correlation analysis and data cleaning of existing network data sets are carried out by data mining technology,and the messy data are converted into useful data sets.Secondly,the weighted KNN algorithm is used to establish the signal prediction model for the data set,and then the particle swarm optimization algorithm is used to establish the network coverage optimization model,so as to obtain the minimum number of base stations and the optimal deployment of base stations that meet the requirements.The experimental results of existing network data show that the model can optimize all the weak coverage grids in the sampling area and reduce the number of base stations.(3)The visualization of network planning optimization data based on big data is completed.The realization of data visualization is based on web front-end and Baidu AI interface.This part shows network planning data from three aspects,including coverage interface,coverage statistics and weekly base station load.The real-time coverage of the current planning interface is displayed through the interface and various functions.The visual interface provides a more intuitive and visual way to extract the temporal and spatial distribution characteristics in network planning.