Semantic-based Adaptation of Quality of Experience in Web Multimedia Streams

Video streaming accounts for the majority of worldwide Internet traffic, and HTTP-based multimedia has the largest share among technologies and protocols. The wide availability of mobile devices and wireless broadband networks currently leads to wider heterogeneity of fruition contexts and frequent condition changes during a streaming session. MPEG-DASH is the reference standard for Dynamic Adaptive Streaming over HTTP: a provider defines several representations for a segmented multimedia source, with different bit rates, allowing a client to dynamically select the best one based on current conditions, and to download the corresponding sequence of segments for smooth playback. MPEG-DASH does not mandate specific bit rate adaptation schemes; conventional approaches are divided in buffer-based, bandwidth-based and hybrid. Nevertheless, Quality of Experience (QoE) can be influenced by many additional factors. This paper proposes a novel QoE adaptation approach based on dynamic ontology-based annotation of streaming context and mobile matchmaking with DASH representation profiles in a Web Ontology Language (OWL) fragment, exploiting a WebAssembly port of an embedded reasoning engine. The proposed framework enables adaptation based not only on network status, but also on client device capabilities, ambient conditions and multimedia content type. A case study validates the proposal, while early experiments support its sustainability.