Karma: Adaptive Video Streaming via Causal Sequence Modeling

TL;DR

Karma is an adaptive video streaming algorithm that uses causal sequence modeling to better understand interrelated factors affecting quality, leading to improved generalization and higher user experience across diverse network conditions.

Contribution

Karma introduces causal sequence modeling with a decision transformer for adaptive bitrate decisions, enhancing generalization and performance over existing methods.

Findings

Achieves 10.8% to 18.7% QoE improvement over state-of-the-art algorithms.

Demonstrates strong generalization to unseen networks in simulations and real-world tests.

Utilizes causal modeling to incorporate interrelated observations, returns, and actions.

Abstract

Optimal adaptive bitrate (ABR) decision depends on a comprehensive characterization of state transitions that involve interrelated modalities over time including environmental observations, returns, and actions. However, state-of-the-art learning-based ABR algorithms solely rely on past observations to decide the next action. This paradigm tends to cause a chain of deviations from optimal action when encountering unfamiliar observations, which consequently undermines the model generalization. This paper presents Karma, an ABR algorithm that utilizes causal sequence modeling to improve generalization by comprehending the interrelated causality among past observations, returns, and actions and timely refining action when deviation occurs. Unlike direct observation-to-action mapping, Karma recurrently maintains a multi-dimensional time series of observations, returns, and actions as input and employs causal sequence modeling via a decision transformer to determine the next action. In the input sequence, Karma uses the maximum cumulative future quality of experience (QoE) (a.k.a, QoE-to-go) as an extended return signal, which is periodically estimated based on current network conditions and playback status. We evaluate Karma through trace-driven simulations and real-world field tests, demonstrating superior performance compared to existing state-of-the-art ABR algorithms, with an average QoE improvement ranging from 10.8% to 18.7% across diverse network conditions. Furthermore, Karma exhibits strong generalization capabilities, showing leading performance under unseen networks in both simulations and real-world tests.