Loading...

Quality of Experience (QoE) Modeling and Resource Allocations based on QoE in Wireless Communication Networks

Zabetian, Negar | 2023

44 Viewed
  1. Type of Document: Ph.D. Dissertation
  2. Language: Farsi
  3. Document No: 56612 (05)
  4. University: Sharif University of Technology
  5. Department: Electrical Engineering
  6. Advisor(s): Hossein Khalaj, Babak
  7. Abstract:
  8. The quality of experience (QoE) is a qualitative metric of users’ satisfaction with the received service. The competition for providing services is fierce, and the one who satisfies the most users wins. Therefore, the method of evaluating user satisfaction is important. In this dissertation, a model for VoIP QoE assessment is proposed. In the proposed method, appropriate features for measuring the quality of voice calls are first extracted solely from the degraded audio signals. Then, an ensemble learning model is created to estimate MOS, acceptance, and probabilistic metrics accurately and close to real opinions. Then, the performance of the model in terms of correlation coefficient and mean absolute error is compared with the real opinions. It is also shown how the more accurate predicted QoE values can be used by service providers to properly modify network parameters to get closer to the required QoE values. Users’ opinions, which are evaluated through QoE, depend on how resources are allocated for a given service. Therefore, simultaneous consideration of optimal resource allocation and QoE evaluation is one of the challenges for service providers, and resource allocation according to QoE is one of the problems that has been considered. After evaluating the QoE, we propose a method to allocate resources based on MOS. In the proposed method, we first model the MOS of the wireless communication network with the regression algorithm. Then we investigate the maximization of the sum MOS of users, the maximization of the number of satisfied users, and the joint maximization of user utility and operator revenue and analyze the results. In all of the problems, we show the superiority of the proposed user satisfaction maximization problem over rate maximization problems and the superiority of the proposed hybrid MOS evaluation model over existing objective models. In the end, by examining the tradeoff between the number of users served and their level of satisfaction, we show how the operator can use the proposed model to provide services
  9. Keywords:
  10. Experience Quality ; Machine Learning ; Power Allocation ; Optimization ; User Admission ; Mean Opinion Score (MOS) ; Resources Allocation

 Digital Object List

 Bookmark

No TOC