Research And Application Of VoLTE Voice Quality Evaluation System

Nowadays,the services provided by mobile communication technology are more and more diversified.However,voice business has always been the most basic business and the advantage of mobile communication.At present,we are in the stage of large-scale commercial use of LTE system,and voice services have been transformed from circuit bearers to bearers in the packet domain.Research shows that VoLTE will be the main solution for voice services in LTE systems.Therefore,how to improve the VoLTE user experience and ensure good voice service quality is one of the hot research topics.Based on the project of "Ministry of Education-Mobile" Research and Innovation Fund,this thesis focuses on the design of voice quality evaluation system in VoLTE network.And it puts forward a method to guarantee the accuracy of the data of the evaluation system.Then this thesis proposes an algorithm for speech quality evaluation,which is based on entropy method and random neural network,and realizes the demands of accurate,real-time and objective evaluation.The main contents of this thesis are as follows:Firstly,based on studying of LTE technology deeply,VoLTE voice communication technology,protocol analysis technology,and neural network,this thesis proposes voice quality evaluation system,which includes four levels,data collection,data analysis,data management,and evaluation algorithms.Secondly,research and analyze the realations between KQI and KPI index system in VoLTE,establish the mapping relation model of VoLTE voice service KPI to KQI index on the basis of protocol parsing,so that the corresponding KQI can be evaluated accurately by the KPI index obtained by the protocol analysis.Thirdly,based on the analysis of VoLTE related business processes and the message structure of related protocols,design the analytic process of RTP,RTCP and SIP protocols,and the relevant parameters are calculated by data correlation analysis,which lays a foundation for accurate evaluation.Fourthly,aiming at researching the E-model evaluation algorithm,which has some shortcomings such as complexity and imperfect,which can not meet the evaluation demand of VoLTE voice service proposed by the project.Aiming at the point of VoLTE technology and the requirement of project,this thesis proposes an objective evaluationalgorithm based on entropy weight method and random neural network,realizing the accurate,real-time and objective evaluation.Finally,the system was tested and operated in a certain community in southwestern China.The quality of VoLTE voice services in the region is evaluated and the performance is verified.