Font Size: a A A

Research And Application Of Mobile Network Characteristics Based On Cell Performance

Posted on:2019-06-14Degree:MasterType:Thesis
Country:ChinaCandidate:Y YuFull Text:PDF
GTID:2428330563958639Subject:Electronic and communication engineering
Abstract/Summary:PDF Full Text Request
With the continuous development of mobile communication technologies,the equipment and data generated by mobile communication networks are also increasing.In order to meet the ever-increasing experience requirements of mobile users,mobile operators need to optimize the mobile communication network based on existing equipment and data and conduct accurate marketing to users in order to improve the market competitiveness of operators.Specifically,the operators not only hope to deal with the anomalies encountered in the network through the post-processing mode of resolving problems,but also hope to implement real-time location analysis of network performance and to avoid possible abnormalities in a timely manner.This paper designs a set of cell performance analysis system based on network characteristics for a mobile branch of a city.This system collects the original performance data of the cell to achieve cell performance acquisition,calculation of cell historical performance benchmark values,real-time prediction of key cell performance,etc.The function,combined with the MFC client,provides a reliable and timely basis for the development of network rescue and marketing personnel marketing plans for network optimization personnel.Firstly,the system implements functions such as quickly importing and exporting the original data of the cell and interpolating the null values of the performance data,which lays the foundation for the analysis of the network characteristics.Secondly,the cell performance index that K-means is most concerned about by mobile operators.Carry out clustering to distinguish cells as key and non-key cells,use the arithmetic mean filter,median filter,and pulse anti-alias filter to fit the performance data of the cell,extract the historical performance data of the cell,and then calculate the cell's historical performance.Reference value: Calculate the performance warning threshold according to the pre-warning threshold range set by the user,determine whether the cell performance is in a high load state,help the optimized staff to keep abreast of the overall operation of the cell network,and report abnormal alarms to the persistently high-loaded cell.,To facilitate the optimizer to make corresponding measures to the abnormal community,provide relevant information for the community to make relevant measures and according to regional performance status,provide a basis for the mobile marketing staff to develop a precise marketing plan;Finally,the use of exponential smoothing,BP neural network method,extreme learning machine,On principal component analysis ELM major focus area of real-time performance prediction,optimization of personnel to facilitate timely manner to avoid potential network anomalies and improve the user experience a sense of the Internet.The cell performance analysis system based on network characteristics research has been implemented in the mobile branch of a city for six months,and has passed a large number of test systems basically stable and continuous operation.Calculate the baseline value of cell performance once a week,and the accuracy rate of warning is over 85%.Real-time prediction of key communities,and the accuracy of warnings exceeds 80%,reaching the system design requirements.
Keywords/Search Tags:Network Performance Analysis, Performance Prediction, Performance Benchmark, Extreme Learning Machine, Early warning
PDF Full Text Request
Related items