A Network-Centric View on DASH in Wireless Multihop Networks

Abstract

Video streaming in wireless multihop networks is a challenge due to different capabilities of end-user devices and changing network conditions. This challenge is addressed at the application layer with adaptive video streaming schemes like dynamic adaptive streaming over HTTP (DASH), which is widely applied by content providers. DASH copes with diverse end-user device capabilities by storing several representations of the same video such that DASH can offer a video in multiple qualities to users. Nevertheless, adjustments in DASH are solely taking place at the application layer. Especially in wireless multihop networks, adaptions on the lower layers are of particular importance. Therefore, we propose a novel application-aware cross-layer framework which adapts network support structures at the network layer, performs resource allocation at the medium access layer, switches between communication types at the physical layer and takes into account the properties and requirements of DASH at the application layer. Furthermore, we present a unified graph model, which takes into account the application layer, the network layer, the medium access layer and the physical layer jointly. We formulate a binary linear problem which chooses the optimal video representation for each user and finds the best combination of mechanisms on the lower three layers to optimally distribute the video content through the wireless multihop network. We show that our application- aware cross-layer framework which utilizes transitions leads to gains between 15-83 % compared to conventional approaches that do not switch between different mechanisms.