| This dissertation aims to improve multimedia streaming performance over a heterogeneous networking environment, which merges wired Internet, wireless local area networks (WLANs), and wireless mesh networks (WMNs) together. First, we propose a novel streaming architecture, called Mobile INfrastructure Enablers for STReaming Optimization and NEw Service (MINESTRONE), to facilitate high-quality multimedia streaming services over heterogeneous networks.; Then, we explore how to design better multimedia streaming schemes based on lower-layer feedback gathered by the proposed MINESTRONE architecture. We address this issue from two aspects: (a) providing reliable high-quality streaming services over the Internet; and (b) enhancing streaming quality over wireless networks, including one-hop WLANs and multi-hop WMNs.; For reliable streaming services over the Internet, we focus on how to combat streaming quality degradation caused by congestion and/or node/link failures by properly adapting streaming schemes. By simulating realistic network failure events using data collected on Tier-1 backbone networks, we demonstrate that our failure-aware rate-adaptive streaming schemes can recover faster from quality degradation than schemes that cannot distinguish different types of packet losses.; To enhance streaming quality over WLANs, we use a hybrid infrastructure/peer-to-peer approach to leverage nearby wireless peers and the remote media server to form a joint sender group, collaboratively streaming information to wireless clients. Since most peers in the multi-source streaming scenario are battery powered, to avoid premature power draining of low-energy peers, we present an energy-aware scheduling scheme to better distribute streaming load among serving peers.; Following the same spirit, we design a multi-source multi-path streaming scheme for supporting concurrent streaming services over WMNs. By taking media source characteristics, underlying WMN conditions, WMN connectivity, and application requirements into consideration, we develop an interference-aware routing path selection scheme to choose concurrent streaming paths for various clients.; Besides combating quality degradation of video streaming services, we also address the inefficiency problem of transmitting voice traffic over WMNs. Toward this end, we propose to implement traffic shaping strategies on intermediate nodes to change traffic characteristics in terms of packet sizes and arrival interval to improve bandwidth efficiency. By adopting physical (PHY) layer rate adaptation, we manage to mitigate the impact of channel errors. |
