| Along with the rapid development of computer technology, video compression technology and communication technology, the streaming media services have been fest developed and widely deployed. The research on efficient and highly reliable streaming media services is quite popular now.There are still a lot of difficulties for streaming media services. First, it’s still an open issue to measure the quality of streaming media services and it’s difficult to optimize the services from the point of service quality. The network QoS (Quality of Service) parameters are usually used by traditional network services to describe thenir service quality. However, they are not totally suitable for streaming media service. Second, sensitivity for real-time and smoothness makes streaming media services unsuitable to transmitted via IP networks for the inherent characters of IP netwoks such as delay, transmission error, packet loss, bandwidth fluctuation and the best-effort service it offored. In this paper, we focus on the description for QoE and streaming control strategies under different transmission protocols. The main content is as follows:1. In order to optimize the streaming media service from the point of QoE, we propose a scalable QoE description method by combine QoE and network QoE parameters together. The description provides quantitative analysis of QoE by consideration of inherent characters of state-of-art media codec compression algorithms and error concealment technologies. Then we propose a streaming media service framework based on the description method by adding a feedback loop to traditional streaming media servive framework with the idea of feedback control. The feedback loop consists of four major components as sensors, analytical module, decision module and controllers. Compared to other methods, the proposed description method doesn’t need to train sample data and increases description accuracy by5%at least.2. To tackle the quality fluctuation and low bandwidth utilization of adaptive streaming service based on HTTP nowadays, we propose a QoE-based adaptive streaming control method that joints both network situation and end-user buffer length. The method focuses on two major issues:the quick and accurate estimation of present network bandwidth and the choice of optimize segments to provide optimize QoE for customer. Compared to Microsoft and Adobe adaptive streaming service over HTTP, the proposed method increase bandwidth utilization by11%at least and improve service smoothness.3. Streaming protocol RTP/RTCP lacks of streaming control strategy and usually the TFRC method is used to provide streaming control. We pointout the shortcomings of the classical TFRC throughput model for fow throughput under the circumstance of network bandwidth fluctuation and slow adaption to the rapid change of network. We propose an QoE-aware top-friendly rate control algorithm (QTFRC) based on TCP throughput model. Furthermore we propose a lost packet classifying algorithm to improve accuracy of QTFRC in the wireless network. These two optimizations make QTFRC can achieve better adaption performance in network fluctuation situations while maintain tcp-friendly and smoothness and improve QoE by increase video quality up to2dB.Part of research results mentioned above has been successfully applied into projects as follows:863project Collaborative Environment for Distributed Broadband Service Production in Shanghai and Interative TV Integrated Service System in Jiading, Shanghai... |
PDF Full Text Request