Performance Evaluation of Video Streaming Applications with Target Wake Time in Wi-Fi 6

Govind Rajendran,Rishabh Roy,Preyas Hathi,Nadeem Akhtar,Samar Agnihotri

DOI: https://doi.org/10.1109/COMSNETS56262.2023.10041325

2023-10-04

Abstract:The Target Wake Time (TWT) feature, introduced in Wi-Fi 6, was primarily meant as an advanced power save mechanism. However, it has some interesting applications in scheduling and resource allocation. TWT-based resource allocation can be used to improve the user experience for certain applications, e.g., VoIP, IoT, video streaming, etc. In this work, we analyze the packet arrival pattern for streaming traffic and develop a synthetic video streaming traffic generator that mimics real-world streaming traffic. We propose a two-stage approach where we calculate the TWT duty cycle in the first step. In the subsequent step, we determine the Multiplication Factor(MF), which jointly dictates the required TWT schedule for the synthetic traffic model. Initial testing shows that key QoS metrics can be met for sustained performance of synthetic traffic upon enabling TWT, even in the presence of peak background congestion in the network.

Networking and Internet Architecture

What problem does this paper attempt to address?

The paper primarily aims to address the issue of optimizing the performance of video streaming applications in Wi-Fi 6 networks by using the Target Wake Time (TWT) feature and exploring how TWT can enhance user experience under high background congestion.

Specifically, the paper proposes a two-stage approach to calculate the TWT scheduling period (Duty Cycle) and the Multiplication Factor to meet the Quality of Service (QoS) requirements of video streaming applications. First, it determines the minimum TWT working period that satisfies the average throughput requirements, and then adjusts the Multiplication Factor to ensure buffer health, avoiding buffer underflow caused by instantaneous throughput fluctuations. Additionally, the paper develops a synthetic video stream generator to simulate real-world video streaming patterns and defines relevant QoS metrics for performance evaluation. Experimental results show that with TWT enabled, key QoS metrics can be met even under high network congestion conditions.