User Preferences Based Preloading and ABR Algorithm for Short Video Streaming

Due to mobile networks and multimedia technology, short video streaming has become increasingly popular. However, traditional ABR algorithms are not suitable for short videos and waste bandwidth. To address this, a user preferences-based preloading and ABR algorithm called USP-ABR has been proposed. It analyzes QoE preferences and divides users into three categories, improving accuracy and saving bandwidth. Experimental findings show significant improvements compared to benchmark algorithms. According to the experimental findings, the three QoE models have improved accuracy by 21.43%, 26.75% and 16.39%, respectively, when compared to the standard user QoE model. Besides, by comparison with two benchmark algorithms, the USP-ABR algorithm can save up to 66.1%, 65.6% and 60.7% of bandwidth overhead and improve QoE by 15.1%, 13.5% and 9.6% under the three categories of QoE models.