A Robust Throughput Estimation in Edge-Assisted Adaptive Bitrate Streaming Networks
Shuaibu Yau, Suphakit Awiphan, Jakramate Bootkrajang, Jiro Katto · IEEE Access · 2025
Dynamic Adaptive Streaming over HTTP (DASH) ensures continuous video transmission via Adaptive Bitrate (ABR) algorithms reliant on accurate throughput estimation. Existing methods often falter in unstable network conditions, impacting Quality of Experience (QoE). We address two intertwined issues in mobile HTTP video delivery. First, we present Weighted Harmonic–Exponential Averaging (WHEA), a lightweight hybrid estimator that reduces the next segment throughput prediction error compared to state-of-the-art baselines. Second, we embed WHEA within a new edge-executed adaptation framework, called Extended-EQAH, which leverages the cross-client view available at the edge to coordinate bitrate choices. Experiments on realistic LTE traces show that Extended-EQAH raises the mean quality by 10.74 % and Jain’s Fairness Index (JFI) by 10.98 % compared to its baseline schemes. In higher-variability settings using 5G traces, Extended-EQAH with WHEA yields substantially larger average relative gains about 27.73% in mean quality and 11.11% in JFI over representative baselines. These results demonstrate that combining accurate edge-side forecasting with joint buffer and throughput control is the key to robust video streaming in dynamic wireless networks.