This dissertation proposes a novel Mobile Video Quality Prediction (MVQP) approach for 3G cellular networks. The MVQP is an objective approach that relies on a set of Quality of Experience (QoE) measurements. The MVQP was developed in two phases. The first phase entails development of a raw video database that contains forty representative 4K raw videos. This video database is accessible on the MVQP website and is free of cost for research purposes. This dissertation contains an evaluation of live video streaming measurement as a part of Phase One of the MVQP project.;A novel method for subjective Quality of Service (QoS) evaluation of streaming video was developed. The method depends on a non-reference QoE measuring tool. This is a hybrid between subjective and objective measurements. To validate this approach as part of the MVQP project, a live video streaming platform was designed. The platform includes different video sequences (high- and low-motion). The results presented in this dissertation are based on a subset of eight videos selected from the MVQP video quality database. A commercial Sprint PCS 3G mobile network was used for the test evaluation. Fifty subjects evaluated video quality using smart phones based on ITU recommendations. The Mean Opinion Score (MOS) was compared with packet loss at the end of Phase One. For MVQP (Phase Two), a QoE tool was built using the results that were collected in Phase One. This QoE tool can predict video quality by using a Radial Basis Function Neural Network (RBF-NN).
展开▼
