Channel Allocation for Smooth Video Delivery over Cognitive Radio Networks

Abstract

Video applications normally demand stringent quality-of-service (QoS) for the high quality and smooth video playback at the receiver. Since the network is usually shared by multiple applications with diverse QoS requirements, QoS provisioning is an important and difficult task for the efficient and smooth video delivery. In the context of cognitive radio (CR) networks, as the secondary or unlicensed users share a pool of bandwidth that is temporarily being unused by the primary or licensed users, there is an inevitable interference between the licensed primary users and the unlicensed CR devices. As a result, efficient and smooth video delivery becomes even more challenging as the channel spectrum is not only a precious resource, but also much more dynamic and intermittently available to secondary users.

In this thesis, we focus on the provision of guaranteed QoS to video streaming subscribers in CR network. In video streaming applications, a playout buffer is typically deployed at the receiver to deal with the impact of the network dynamics. With different buffer storage, users can have different tolerance to the network dynamics. We exploit this feature for channel allocation in CR network. To this end, we model the channel availability as an on-off process which is stochastically known. Based on the bandwidth capacity and the specific buffer storage of users, we intelligently allocate the channels to maximize the overall network throughput while providing users with the smooth video playback, which is formulated as an optimization framework. Given the channel conditions and the video packet storage in the playout buffer, we propose a centralized scheme for provisioning the superior video service to users. Simulation results demonstrate that by exploiting the playout buffer of users, the proposed channel allocation scheme is robust against intense network dynamics and provides users with the elongated smooth video playback.