پارسا حسني شريعت پناهي

Today, the Internet is recognized as one of the necessities of human life and plays an important role in communication, business and people's lifestyle. Disruption or low quality of internet services can have extensive negative effects on individual and social levels. In the intense competition to provide superior services, the result of quality of service (QoS) Internet has become a basic need. Operators are subscribers and pioneers in technologies that improve service quality, because this quality determines their survival and success. With the development of mobile networks, a new concept of quality called quality of experience (QoE) has been proposed, which in addition to service quality parameters, also considers user satisfaction. Accurate measurement of QoE helps operators to adjust their services accordingly. However, the eva‎luation of the quality of experience for operators has its own challenges. One of the main problems is the lack of information and modeling of QoE is only the operators' point of view and the feedback of the service quality at the time of actual use. Currently, researches in the field of user experience have moved towards the use of machine learning and deep learning techniques Using a large amount of user and network data, develop models to accurately predict user satisfaction. However, there are several challenges such as real data collection, variety of user scenarios and complexities of the network environment that require further research. This research is a step towards solving these challenges by providing a practical and efficient framework for eva‎luating and improving the quality of user experience. In this research, a comprehensive framework for measuring and eva‎luating the quality of experience in multimedia services is presented. This framework includes the steps of automatically collecting data, calculating the user satisfaction score according to the T-ITU .1203P standard, and then using machine learning algorithms to predict user satisfaction based on network parameters. By collecting more than 20,000 data records from users and different network conditions, the random forest model was able to predict user satisfaction with 95.8% accuracy by only receiving five main network parameters. These results show that operators can estimate the quality of users' experience in real time by monitoring network parameters and allocate network resources optimally and dynamically. This approach helps operators to maintain a high level of customer satisfaction at minimal cost.

كيفيت تجربه , كيفيت خدمات , يادگيري‌ ماشين , چندرسانه‌اي

Quality of experience , Quality of services , Machine learning , Multimedia

Parsa Hassani Shariat Panahi

Abolfazl diyanat