Analysis of Buffer Starvation with Application to Objective QoE Optimization of Streaming Services

TL;DR

This paper models buffer starvations in streaming services using queue theory, providing exact and approximate distributions, and applies this analysis to optimize user experience by balancing startup delays and rebuffering events.

Contribution

It introduces two novel analytical approaches for buffer starvation distribution, including explicit solutions via Ballot theorem and recursive methods for bursty arrivals.

Findings

Explicit starvation distribution results for M/M/1 queues.

Extension of analysis to bursty ON/OFF arrival processes.

Application of starvation analysis to optimize QoE in streaming.

Abstract

Our purpose in this paper is to characterize buffer starvations for streaming services. The buffer is modeled as an M/M/1 queue, plus the consideration of bursty arrivals. When the buffer is empty, the service restarts after a certain amount of packets are \emph{prefetched}. With this goal, we propose two approaches to obtain the \emph{exact distribution} of the number of buffer starvations, one of which is based on \emph{Ballot theorem}, and the other uses recursive equations. The Ballot theorem approach gives an explicit result. We extend this approach to the scenario with a constant playback rate using T\`{a}kacs Ballot theorem. The recursive approach, though not offering an explicit result, can obtain the distribution of starvations with non-independent and identically distributed (i.i.d.) arrival process in which an ON/OFF bursty arrival process is considered in this work. We further compute the starvation probability as a function of the amount of prefetched packets for a large number of files via a fluid analysis. Among many potential applications of starvation analysis, we show how to apply it to optimize the objective quality of experience (QoE) of media streaming, by exploiting the tradeoff between startup/rebuffering delay and starvations.