Pdas

Joint Preloading and Bitrate Adaptation for Short Video Streaming

“…Two main approaches have been used to develop preloading strategies for short video streaming: Learning-based approach [10]- [15], and conventional approach [7]- [9], [17]. In the learning-based approach, user data such as viewing duration, and swipe times are collected and used to train machine learning models that decide the chunk download strategy.…”

“…The above methods, however, do not consider bitrate adaptation. In [9], the authors propose PDAS, a probability-based adaptation method for short video streaming, in which the maximum buffer size is chosen based on chunk-level viewing probability and estimated network bandwidth. Furthermore, in [17], the authors present Dashlet, a system that predicts the rebuffering time for each potential video chunk to determine the optimal chunk buffering sequence for preloading.…”

“…Our proposed algorithm, described in Algorithm 1, takes inspiration from the PDAS study [9]. Both our proposed algorithm and PDAS share a fundamental strategy for managing short video streaming.…”

“…Subsequently, we calculate and present the average results across all cases in Figure 5, Figure 6, and Figure 7 in the following subsection. Related studies, including [9], [13], and [15], have also used the same evaluation approach to assess the performance of their proposed methods. This approach has allowed us to draw valuable insights and demonstrate the effectiveness of our proposed method.…”

“…Early works focus on designing appropriate chunk preloading strategies based on various factors such as predicted user viewing time [7], and network throughput [8]. Preloading and bitrate adaptation are jointly considered in [9]. However, this method results in high variations in video quality, causing negative effects on user experience.…”