|
" QoE-aware resource allocation in video communications "
Danish, Emad
Fernando, Anil
| Document Type
|
:
|
Latin Dissertation
|
| Record Number
|
:
|
831901
|
| Doc. No
|
:
|
TLets683868
|
| Main Entry
|
:
|
Danish, Emad
|
| Title & Author
|
:
|
QoE-aware resource allocation in video communications\ Danish, EmadFernando, Anil
|
| College
|
:
|
University of Surrey
|
| Date
|
:
|
2016
|
| student score
|
:
|
2016
|
| Degree
|
:
|
Thesis (Ph.D.)
|
| Abstract
|
:
|
Video streaming is expected to acquire a massive share of the global internet traffic in the near future. Meanwhile, it is expected that most of the global traffic will be carried over wireless networks. This trend translates into considerable challenges for Service Providers (SP) in terms of maintaining consumers’ Quality of Experience (QoE), energy consumption, utilisation of wireless resources, and profitability. However, the majority of Radio Resource Allocation (RRA) algorithms only consider enhancing Quality of Service (QoS) and network parameters. Since this approach may end up with unsatisfied customers in the future, it is essential to develop innovative RRA algorithms that adopt a user-centric approach based on users’ QoE. This research thesis contributes to three areas relevant to video communication systems. Firstly, a novel QoE-aware RRA scheme is designed. In the context of video transmission over a multiuser Orthogonal Frequency-Division Multiplexing (OFDM) network, the scheme considers a joint optimisation of users’ QoE, energy efficiency, and SP’s Quality of Business (QoB). Based on the evolutionary genetic algorithm (GA), the RRA scheme addresses the multi-objective optimization problem of maximising average video quality, minimising total energy, and maximising utility value. In contrast to comparable methods, the algorithm has improved energy efficiency by up to 22.3% at the expense of up to 5.6% reduction in quality. Secondly, for the practicality concern of the RRA scheme, a QoE prediction system is modelled to facilitate instantaneous estimation of QoE. Hence, based on Fuzzy Inference Systems (FIS), a No-Reference (NR) prediction model for video quality in the wireless domain is designed. The model has achieved a prediction accuracy of 0.199 RMSE, with a correlation factor of 0.94 between the measured and the predicted QoE, which outperforms comparable models. Thirdly, a QoE-driven efficient resource utilisation study is conducted for video delivery over WiMAX. The study has shown that significant bandwidth and power efficiency can be achieved when Modulation and Coding Scheme (MCS) selection is based on QoE. This research thesis emphasises the fundamental and crucial importance of adopting users’ QoE as a parameter in the RRA process for future multimedia communications. Moreover, it is expected that this approach would help limit the carbon footprint of wireless communications, while guaranteeing consumers’ QoE.
|
| Added Entry
|
:
|
Fernando, Anil
|
| Added Entry
|
:
|
University of Surrey
|
| |