User Perceived Qos Provisioning for Video Streaming in Wireless OFDMA Systems: Admission Control and Resource Allocation

In this paper, we investigate a joint admission control (AC) and resource allocation (RA) scheme for providing quality of service (QoS) guarantees for video streaming in OFDMA networks. The QoS provisioning is posed as a network utility maximization (NUM) problem. We prove that the QoS requirements are mainly dominated by the average data rate with the aid of diffusion approximation approach, allowing us to decompose the NUM problem into two tractable sub-problems, namely, the AC and the RA sub-problems. We develop an AC algorithm, which enables us to select a maximum user set such that all selected users can meet their QoS requirements related to user perceived video quality during the video playback period. In addition, in order to handle the computational complexity, we further split the RA sub-problem into the constrained Markov decision process associated problems performed at users together with a transmission problem performed at the base station. Correspondingly, we develop a power and sub-carrier allocation algorithm to minimize the average power consumption. The decision of AC is made in a long timescale and depends on the average channel state information (CSI), whereas the decision of RA is made in a short timescale and is subject to both instantaneous CSI and queue state information. Simulation results demonstrate the advantages of the proposed algorithms.