Application Of Intelligent Algorithms In QoE Analysis Of Telecom Services

In recent years, Chinese telecom industry is in the critical transitional period. After the full service operations in the telecom field, the core competence of this industry inevitably transfers to the QoE (Quality of Experiences) and user experience. Hence, the quality of service provided and the measurement of QoE attract more and more attention. As telecom operators'business philosophy becomes clearer, the analysis of customer experience has become core capabilities for operators to build the operational supported architecture with high quality.In the telecom domain, QoE analysis predicts the customers'recognition of telecom services and network quality via imperfect quantifiable methods. It can be used to reflect not only the subjective feelings of users, but also the gap between customers' expectations and actual qualities. For the present, QoE analytical methods are based on empirical models; intelligence algorithms have not been widely applied. Model training and analyzing usually takes long cycles. Moreover, the requirements of adjusting models after practical applications can not be met quickly and effectively.With respect to the complex and subjective QoE analysis in telecom services, intelligence algorithms have good adaptability. In this trend, the applications of intelligent algorithms in customer's QoE analysis of telecom services have bright market prospect.Including the features of telecom industry, we study the way to utilize intelligence algorithms in QoE analysis.First, on perceiving the problems in the analysis of user experience, such as slow modeling, analysis tools decentralization, lack of integrated management and others, we propose the idea of establishing modeling management tools of QoE analysis. With Decision-Tree, Fuzzy Neural Network and the previous target system as fundamental parameters, we implement all of the steps of customers QoE models'training, analysis and calculating. In the thesis, we display the detailed descriptions on integrated management of data mining, decision analysis and strategies-making, and establish the closed management cycle - Training to Application, to Analyzing, and to Improving. Thus, the study can help operators in telecom industry respond rapidly to the changes of the markets and the demands of the customers.1. Based on logic and numeric algorithms, we provide a platform to build the training and analysis supported QoE analysis model. Also, we observe the effects of the model and support quick model retraining to get continued optimization. Furthermore, we integrate the user experience analysis on the closed cycle and keep unified management of model training, analysis, calculation, verification, decision-making, and the reports process.2. We provide unified interface for different kinds of business data from various data sources in the telecom industry, to obtain real-time data entry for calculating QoE value immediately. Make the practical model of QoE analysis normalized, and promptly forecast, monitor and analyze the user experience.3. With aim to get rapid application, we integrate the running environment and define import interface for external models to gain excellent extensibility for algorithms.4. For the depth operation of telecom services, we integrate multi-topics analysis to make monitoring and supervision, with excellent services expandability.5. In the final section of this thesis, we describe the practical effects and the testing results of our research.Nowadays, the achievement of this work has already got excellent demo applications in several telecom sections in different provinces, and won the business opportunity of further developments.