PiTree: Practical Implementation of ABR Algorithms
Using Decision Trees


Zili Meng      Jing Chen      Yaning Guo      Chen Sun      Hongxin Hu      Mingwei Xu     

Tsinghua University            Clemson University


Abstract


Major commercial client-side video players employ adaptive bitrate (ABR) algorithms to improve user quality of experience (QoE). With the evolvement of ABR algorithms, increasingly complex methods such as neural networks have been adopted to pursue better performance. However, these complex methods are too heavyweight to be directly implemented in client devices, especially mobile phones with very limited resources. Existing solutions suffer from a tradeoff between algorithm performance and deployment overhead. To make the implementation of sophisticated ABR algorithms practical, we propose PiTree, a general, high-performance and scalable framework that can faithfully convert sophisticated ABR algorithms into lightweight decision trees to reduce deployment overhead. We also provide a theoretical analysis on the upper bound of performance loss during conversion. Evaluation results on three representative ABR algorithms demonstrate that PiTree could faithfully convert ABR algorithms into decision trees with less than 3% average performance loss. Moreover, comparing to original implementation solutions, PiTree could save operating expenses by up to millions of dollars every month for large content providers.


Paper


PiTree: Practical Implementation of ABR Algorithms Using Decision Trees
Zili Meng, Jing Chen, Yaning Guo, Chen Sun, Hongxin Hu, Mingwei Xu
Accepted by the ACM International Conference on Multimedia (2019)


Code


Will be released in early October.


Demo


Will be released in early October.


Supporters


This project is supported by National Key R&D Program of China (2017YFB0801701) and National Science Foundation of China (61625203, 61832013, 61872426).


Contact


Zili Meng (zilim@ieee.org)