Performance evaluation of peer-assisted MPEG-DASH streaming in multi-user mobile scenarios

The dramatic surge in multimedia consumption has exposed the structural vulnerabilities of traditional client–server distribution models, particularly during massive concurrent demand peaks. While hybrid architectures offer a potential solution, their real-world effectiveness is fundamentally governed by the degree of content demand overlap—a factor that has remained largely under-explored in experimental literature. This paper addresses this gap by proposing SwarmLayer, a novel hybrid peer-to-peer (P2P) assisted streaming architecture designed as a transparent middleware between the DASH player and the content delivery infrastructure.

The proposed architecture leverages a combination of core web technologies, including WebRTC for low-latency peer connectivity, Service Workers for request interception, and Protocol Buffers for efficient data serialization. By operating at the browser’s network layer, SwarmLayer enables direct segment exchange among peers without necessitating modifications to the media player or application logic. Central to this work is the formalization of the Content Overlap Ratio (COR), a probabilistic framework introduced to quantify the potential for peer collaboration as the likelihood that two arbitrary players request segments from the same media presentation. This metric allows for a systematic characterization of the relationship between user demand patterns and systemic distribution efficiency.

The performance of the proposed architecture is evaluated using a controlled multi-user testbed that emulates vehicular streaming scenarios through time-varying bandwidth profiles derived from real 5G mobility traces. Experimental results demonstrate that even moderate content overlap (COR = 0.14) yields a 57.4% reduction in CDN traffic, with offloading gains reaching 91.7% in high-demand scenarios. Beyond bandwidth optimization, the analysis of Quality of Experience (QoE) indicators reveals that the peer-assisted layer significantly enhances playback stability by increasing buffer robustness and mitigating bitrate oscillations under non-stationary network conditions.