A Conditional-Probability-Distribution Model for Bandwidth Estimation with Application in Live Video Streaming Weijia Zheng1

2025-04-28 0 0 833.57KB 5 页 10玖币

侵权投诉

A Conditional-Probability-Distribution Model for Bandwidth

Estimation with Application in Live Video Streaming

Weijia Zheng1

1Department of Information Engineering, CUHK

1Department of Mathematics, CUHK

1wjzheng@link.cuhk.edu.hk

Abstract

Experience of live video streaming can be improved if the

video uploader have more accurate knowledge about the

future available bandwidth. Because with such knowl-

edge, one is able to know what sizes should he encode the

frames to be in an ever-changing network. Researchers

have developed some algorithms to predict throughputs

in the literature, from where some simple hence practical

ones. However, limitation remains as most current band-

width prediction methods are predicting a value, or a point

estimate, of future bandwidth. Because in many practi-

cal scenarios, it is desirable to control the performance to

some targets, e.g., video delivery rate over a given tar-

get percentage, which cannot be easily achieved via most

current methods.

In this work, we propose the use of probability distri-

bution to model future bandwidth. Speciﬁcally, we model

future bandwidth using past data transfer measurements,

and then derive a probability model for use in the appli-

cation. This changes the selection of parameters in appli-

cation into a probabilistic manner such that given target

performance can be achieved in the long run. Inside our

model, we use the conditional-probability method to corre-

late past and future bandwidth and hence further improve

the estimating performance.

Keywords

Network throughput estimation, conditional probability,

relative frequency, live video streaming, empirical method

1 Introduction

With the development of smartphones and high-speed mo-

bile data networks such as 3G and 4G/LTE, live video

streaming has long become part of our lives in the enter-

tainment or casual ﬁelds. In recent years, it has played a

vital role in the workplace as well. As we all experienced

in person, the year of 2020 witnessed Zoom’s signiﬁcant in-

crease [1] in the usage of remote work, distance education,

and online social relations.

Given the importance of video streaming and the high

peak bandwidth nowadays in mobile data networks, band-

width ﬂuctuation remains a challenging obligation for its

unpredictable characteristic by its wireless nature, which

may aﬀect the quality of clients’ experience. Several pre-

diction methods were established or used in the literature,

such as arithmetic mean (AM), multiple linear regression

(MLR), ARIMA, LSTM, etc. However, most of them pre-

dict a point estimate of future bandwidth, or gives a con-

ﬁdence interval with further assumptions, which is hard

to guarantee its accuracy to a speciﬁed level in practice.

Motivated by the ideas of using an empirical condi-

tional probability for prediction in ﬁnancial [2] and trans-

portation ﬁelds [3], this work contributes to establishing

a conditional-probability model in bandwidth prediction.

And a simulator of the uploading part of a live video

streaming, referencing the one in [4], is implemented for

simulation and demonstrating the feasibility of such a

method. To mimic a real network environment, the simu-

lator uses a packet-level with timestamp trace data mea-

sured from real-world network sources of 3HK 4G.

The rest of the paper is organized as follows. Section

2 introduces the settings of the problem. Some analyses

and related works are presented in Section 3. In Section

4 we will derive the proposed encoding scheme in detail.

Numerical results of it on diﬀerent network environments

are shown in Section 5, before conclusions are outlined in

Section 6.

Besides, unless otherwise stated, all the bitrate variables

are in Mbps, and all the time variables are in seconds in

this paper.

2 Problem Background

In our study, to concentrate on the uplink part, we con-

sidered the uplink part exclusively with some further sim-

pliﬁcations. We do not consider the downlink part of the

streaming process in the simulator and the general sce-

nario can be described as the following.

An uploader generates frames one by one with an equal

time diﬀerence (i.e., 1/FPS second) in between. Assume

that TCP is used, once a frame is sent from the uploader

side, the uploader will not consider it anymore but will

only take care of later frames. The uploader then starts

transmitting some newest possible frame as buﬀer time

arXiv:2210.01652v1 [cs.MM] 16 Apr 2022

文档加载中……请稍候！
如果长时间未打开，您也可以点击刷新试试。

下载文档到电脑，查找使用更方便

10 玖币 0人已下载

立即下载

摘要：

AConditional-Probability-DistributionModelforBandwidthEstimationwithApplicationinLiveVideoStreamingWeijiaZheng11DepartmentofInformationEngineering,CUHK1DepartmentofMathematics,CUHK1wjzheng@link.cuhk.edu.hkAbstractExperienceoflivevideostreamingcanbeimprovedifthevideouploaderhavemoreaccurateknowledgea...

展开>> 收起<<

A Conditional-Probability-Distribution Model for Bandwidth Estimation with Application in Live Video Streaming Weijia Zheng1.pdf

共5页,预览1页

还剩页未读，继续阅读

声明：本站为文档C2C交易模式，即用户上传的文档直接被用户下载，本站只是中间服务平台，本站所有文档下载所得的收益归上传人(含作者)所有。玖贝云文库仅提供信息存储空间，仅对用户上传内容的表现方式做保护处理，对上载内容本身不做任何修改或编辑。若文档所含内容侵犯了您的版权或隐私，请立即通知玖贝云文库，我们立即给予删除！

A Conditional-Probability-Distribution Model for Bandwidth Estimation with Application in Live Video Streaming Weijia Zheng1

相关推荐

开通VIP享超值会员特权

作者详情

相关内容

热门标签

举报选择: