Quality of experience prediction model for progressive downloading over mobile broadcast networks

The paper introduces a new quality of experience (QoE) prediction model for progressive downloading over the mobile broadcast networks. The proposed model covers new QoE metrics, a novel QoE aware buffering method, and a novel QoE measurement method that predicts the perceived service quality in real time. Progressive download is a streaming technology using “play while download” approach. As a case study, progressive downloading is used over the multimedia broadcast multicast service (3GPP’s MBMS) as the underlying network. The broadcast networks are unidirectional delivery platforms, and hence exposed to many unwanted conditions such as the packet losses, delays, and bandwidth problems. Quality of service (QoS) is a way of classification that manages how these conditions are controlled and mapped to the service quality. However, QoS parameters could not reflect how the end-user experience is. At this point, the QoE describes the achieved QoS and the end-user satisfaction with the service. Conventional QoE metrics for multimedia streaming assumes that packet losses cause artifacts, such as blurring and color distortions, on media presentation. However, with the progressive downloading the packet losses could be tolerated, e.g., using forward error correction. With the protection against loss errors, the overall network errors are projected onto the behaviors of the buffering model. The buffering characteristics should be described by well defined states and expected behaviors in that expected behaviors, from the user expectation point of view, are better than the random ones. In this article, first a buffering method for progressive downloading is proposed. Then, a real time QoE measurement method is proposed to map the buffering characteristics to the achieved performance of the service. Finally, some subjective study using mean opinion score are provided to prove the accuracy of the proposed model.