علي عاليشوندي

Nowadays, with the increase in services offered through mobile networks and the growing usage of this technology by users, the concept of Quality of Experience (QoE) has gained significant importance. QoE is generally a subjective metric that can yield different results depending on the individual user. One of the key areas where QoE becomes particularly crucial, due to its popularity, is video streaming. Specifically, video streaming holds a special place in cellular networks and constitutes a large portion of data traffic. This leads to significant engagement of operators' resources and equipment with video streams. On the other hand, due to this popularity, user satisfaction with this service is of great importance. Poor service quality in this area can result in dissatisfaction among a large segment of users. Mobile operators are compelled to compete with other operators and must provide a satisfactory Quality of Experience to users; otherwise, they risk losing their customers. As a result, operators need appropriate tools to estimate the QoE received by users in different areas covered by their network. Without such tools, they will fail to compete with their rivals. Assuming operators can obtain an accurate estimate of users' Quality of Experience , how can they establish a proper mapping between QoE and their technical status and Quality of Service (QoS) metrics? In this case, there needs to be a correspondence between QoS metrics and QoE, so that mobile operators can identify, for example, which part of their network is experiencing issues when users report poor QoE. To this end, in addition to subjective QoE assessments, objective eva‎luations are also defined. These eva‎luations typically use mathematical functions to create an appropriate mapping between QoS and QoE. Moreover, recent research has employed machine learning models to perform this assessment. In this study, alongside reviewing methods for eva‎luating Quality of Experience in mobile networks' video streaming—both subjective and objective—we will also collect data from mobile networks, including radio parameters and the Mean Opinion Score (MOS), which is a subjective eva‎luation metric. Additionally, we will use machine learning to present models based on nonlinear regression, KNN, and decision trees. These models, utilizing 4G radio parameters such as RSRP, RSSI, RSRQ, SNR, and CINR, will help estimate the QoE experienced by users. To achieve this, we have compiled a dataset that includes radio parameters from the MCI network and users' QoE derived from video streams on the Aparat platform.

كيفيت تجربه , پارامترهاي راديويي , شبكه‌هاي تلفن همراه , جريان ويدئو , يادگيري ماشين

Quality of Experience , Radio parameter , Mobile networks , video streaming , Machine Learning

ali aalishvandi

Abolfazl Dianat