Mobility and opportunistic resource allocation in wireless multimedia networks
Abstract: In order to support increasing traffic loads, mobile operators need cost-effective solutions to improve the spectral efficiency of their cellular networks, or to off-load them by diverting some of the load to other networks. Advances in the radio resource management may to some extent reduce the need for costly new deployments. The resource management should not only focus on spectrum efficiency—it should try to meet the service requirements of applications that are expected to contribute large data volumes, such as video streaming. Many of those applications are multicast/broadcast in nature (e.g., mobile TV, data podcasting). Our focus in this thesis is on resource allocation mechanisms that exploit the mobility of users. The mobility induces channel quality fluctuations and creates intermittent connectivity, which both can be used to improve the resource efficiency of wireless multimedia systems. The thesis concentrates on two areas: link-layer resource allocation for video streaming in cellular networks and mobility-assisted content distribution in hybrid cellular/ad-hoc networks.In the area of wireless video streaming, we study bit-rate allocation, statistical multiplexing, and channel-aware scheduling. The bit-rate allocation should provide a distortion-optimal assignment of source, channel, and pilot bit-rates under link capacity constraints. We derive an analytical model that captures the video distortion as a function of the bit-rates and, based on it, we study various bit-rate allocation strategies and their robustness to varying radio conditions. The statistical multiplexing can be used to smooth out the burstiness of video streams and avoid over-provisioning of transport channels. We study the statistical multiplexing gains of H.264 video streams, both in terms of bit-rate requirements and video quality. When multiple flows are multiplexed on a shared transport channel, multi-user scheduling becomes crucial for the performance. Channel-aware scheduling exploits fluctuations in radio conditions to optimize the assignment of channel resources. We study the impact of channel-aware scheduling on the performance of delay-sensitive applications and possible extensions of channel-aware schemes to multicast scenarios.In the area of mobility-assisted content distribution, we study the resource efficiency of mobility-assisted podcasting and we propose an analytical model for pedestrian content distribution. The mobility-assisted podcasting exploits random encounters of mobile terminals equipped with short range radios to forward the podcast episodes, thereby relieving the strain on cellular networks. We provide results on the achievable spectrum and energy savings of such scheme. Finally, we introduce the “street model”, the first building block in a conceived library of analytical models that would be used to study the performance of pedestrian content distribution in some common case scenarios of urban mobility. Based on the “street model”, we study how various system parameters and node mobility affect the efficiency of content distribution in a grid of streets that represents a part of Stockholm’s downtown area.
This dissertation MIGHT be available in PDF-format. Check this page to see if it is available for download.