Cellular Network Quality Analysis System Based On Mobile Crowd Sensing

With the rapid evolution of cellular network communication technology and the popularization of smart phones,the mobile data service grows explosively.The operating philosophy of operators is also shifting from a network-centric approach to being usercentric.The ability to provide users with network quality that satisfies their service experience is the key for operators to compete in the mobile user market.The traditional network quality analysis method focuses on the extraction and analysis of network element indicators.It fails to directly reflect the user's real experience.Therefore,it is of great significance to build a network quality analysis system from user side.To solve the above problems,the thesis combines the characteristics of MCS(Mobile Crowd Sensing)with human-centered and wide coverage,and proposed a useroriented cellular network quality analysis system to analyze the weak coverage,connectivity and QoE of the network.The specific work of this thesis is as follows:1.According to the process of data collection,analysis and display,this thesis divides the system into three modules: mobile sensing terminal,data analysis platform and web server.The design and implementation of each module is completed according to the specific functional division.2.To solve the problem that the mobile terminal cannot obtain the real service data of the user in real time due to the limitation of user rights and privacy control,this thesis adopts the technology that simulates the service behavior of the user to extract the indicators for QoE assessment at the service layer in real time.3.Most of the current mature QoE evaluation models are oriented to a single service,such as audio and video,and cannot be applied to multi-service scenarios in a cellular network environment.To solve this problem,this paper establishes a three-tier evaluation model for multi-service scenarios to quantify QoE.4.Aiming at the problem that a single server cannot efficiently process and analyze massive sensing data in a cellular network,a data analysis platform based on a Hadoop cluster is built,and a parallel distributed computing model named MapReduce is used to complete processing and analysis of sensing data.The system test results show that the cellular network quality analysis system designed in this thesis has good compatibility,concurrency and stability.The accuracy of service QoE has reached more than 88%,and the overall QoE is closer to the user's real experience compared with the road test.The system results have a guiding significance for wireless network optimization.