Providing video QoE in a wireless environment

Abstract

The spreading of video streaming services in the last few years is presenting new challenges in wireless networking; namely, one of the biggest hurdles is defining and providing consistent Quality of Experience (QoE) to users in complex environments such as congested wireless channels. In this talk, we present our latest works on the topic, focusing on both distributed and centralized solutions. First, we will present a client-side Reinforcement Learning-based adaptation algorithm that overcomes the issues of previous learning-based solutions. In particular, since our algorithm uses a parallel learning technique that accelerates the learning and limits sub-optimal choices, the efficiency of our controller is not sacrificed for fast convergence. Simulation results show that our algorithm achieves a higher QoE than existing RL algorithms in the literature as well as heuristic solutions, as it is able to increase average QoE and reduce quality fluctuations. One of the most important challenges in trying to improve the performance of bitrate adaptation algorithms is the variability of the physical channel; in another recent work, we tried a machine learning approach to predict the long-term variation of the SNR of a channel, obtaining promising results that we plan to integrate into the reinforcement learning model. Finally, we present the centralized adaptation problem, including both quality adaptation and video admission control and aiming at providing QoE fairness among users. Since the problem is NP-hard, we define two heuristic algorithms and discuss their advantages and limitations.

Short Bio

Venue

Seminar room, Dunlop-Oriel