A Hybrid Model of Adaptive Video Streaming Control Systems G. - - PowerPoint PPT Presentation

▶
a hybrid model of adaptive video streaming control systems
SMART_READER_LITE
LIVE PREVIEW

A Hybrid Model of Adaptive Video Streaming Control Systems G. - - PowerPoint PPT Presentation

Feb 11, 2024 358 likes •575 views

A Hybrid Model of Adaptive Video Streaming Control Systems G. Cofano, Luca De Cicco , S. Mascolo Politecnico di Bari, Italy 55th IEEE Conference on Decision and Control - Las Vegas, USA 12-14 December 2016 Introduction Video will represent

slide-1
SLIDE 1

A Hybrid Model of Adaptive Video Streaming Control Systems

  • G. Cofano, Luca De Cicco, S. Mascolo

Politecnico di Bari, Italy 55th IEEE Conference on Decision and Control - Las Vegas, USA

12-14 December 2016

slide-2
SLIDE 2

Introduction

Video will represent ∼80% of Internet global traffic by 2018, up from ∼60% in 2013 (source Cisco) Users are moving from traditional TV broadcasting to Internet based video services

(Source statista.com)

  • L. De Cicco (luca.decicco@poliba.it)

A Hybrid Model of AVS Control Systems IEEE CDC 2016 1 / 16

slide-3
SLIDE 3

The Challenge

Provide an experience at least as good as traditional TVs on a heterogeneous set of devices (tablets, smartphones, smart TVs) Playback interruptions due to buffer depletion must be avoided (highly detrimental for QoE) Video bitrate should be as high as possible to increase visual quality Legacy systems (until ∼2010) encoded video at fixed bitrate Modern video streaming systems can vary the video bitrate and resolution to adapt to:

◮ Time-varying Internet available bandwidth ◮ Device resolution

  • L. De Cicco (luca.decicco@poliba.it)

A Hybrid Model of AVS Control Systems IEEE CDC 2016 2 / 16

slide-4
SLIDE 4

The mainstream Multi-Bitrate Approach

ˆ τ bitrate Si,3 Si,2 Si,1 Si,s bitrate li video time video time l1 S1,1 S1,2 S1,3 S1,s ˆ τ

Encoder Video bitrate nominal nominal bitrate

The video is encoded into a number of versions, the video levels, at different (nominal) bitrates li Each video level is temporally divided into segments of fixed duration ˆ τ Si,s is the actual size produced by the encoder for segment s and level i. Actual bitrate is Si,s/ˆ τ

  • L. De Cicco (luca.decicco@poliba.it)

A Hybrid Model of AVS Control Systems IEEE CDC 2016 3 / 16

slide-5
SLIDE 5

A video streaming control streaming in a nutshell

Playout buffer Internet dr fr r q s (l, w) L Client (li, s) r GET (li, s) Downloader Player Controller Server HTTP

The client fetches video segments at a video bitrate li decided by the controller from an HTTP server

  • L. De Cicco (luca.decicco@poliba.it)

A Hybrid Model of AVS Control Systems IEEE CDC 2016 4 / 16

slide-6
SLIDE 6

The Controller

Playout buffer Internet dr fr r q s L Client (li, s) r GET (li, s) (i, w) Downloader Player Controller Server HTTP

The controller’s output has two components:

1

The index i ∈ {1, . . . , N} of the video level to be downloaded

2

The idle time w to be waited between two consecutive segments download

Control actions can be actuated only when a new segment download is triggered (event driven)

  • L. De Cicco (luca.decicco@poliba.it)

A Hybrid Model of AVS Control Systems IEEE CDC 2016 5 / 16

slide-7
SLIDE 7

The Downloader

download of segm. s is started download ends

  • f segment s

download time τs OFF idle time

  • f segm. s
  • downl. of segm.

s + 1 is started ts ws ts + τs + ws ts+1 = ON The downloader serves as the actuator and fetches video segments from the HTTP server It pushed video segments as soon as they are downloaded (at time ts + τs)

  • L. De Cicco (luca.decicco@poliba.it)

A Hybrid Model of AVS Control Systems IEEE CDC 2016 6 / 16

slide-8
SLIDE 8

The Player dr = 1 Play dr = 0 Pause Buffer gets empty Buffer gets above qmin

Two possible states: Pause: video playback is paused (drain rate dr = 0) to allow the buffer to be filled Play: the player drains the buffer at rate dr = 1 and plays the video

  • L. De Cicco (luca.decicco@poliba.it)

A Hybrid Model of AVS Control Systems IEEE CDC 2016 7 / 16

slide-9
SLIDE 9

The playout buffer

ˆ τ q[s] t

It is drained continuously by the player at a rate dr It is impulsively filled by a fixed amount ˆ τ (segm. duration) on completion of a segment download Net increment (decrement) of queue length due to the download of a segment: q(ts + τs) − q(ts) = ˆ τ − ˆ ts+τs

ts

dr(ξ)dξ

  • L. De Cicco (luca.decicco@poliba.it)

A Hybrid Model of AVS Control Systems IEEE CDC 2016 8 / 16

slide-10
SLIDE 10

Hybrid Dynamical Systems modeling framework

A hybrid system can be described by four elements [HDS]: H :

  • ˙

x = f (x, u, r) (x, u, r) ∈ C, x+ = g(x, u, r) (x, u, r) ∈ D Flow set: C where the state x “flows” (evolves continuously) Jump set: D where the state x “jumps” (time discrete evolution) Flow map: f modelling time-continuous evolution Jump map: g modelling time-discrete evolution

[HDS] R. Goebel, R.G. Sanfelice, A.R. Teel, “Hybrid Dynamical, Systems: modeling, stability, and robustness”, Princeton University Press, 2012

  • L. De Cicco (luca.decicco@poliba.it)

A Hybrid Model of AVS Control Systems IEEE CDC 2016 9 / 16

slide-11
SLIDE 11

State, input, disturbance

State x =         q dr ∆ τ σ s         Playout buffer length (s) [PB] Draining rate [P] Segment downloaded bytes [D] Idle timer (s) [D] Downloader state (1=ON,0=Idle) [D] Segment index [D] Input u = i w

  • Video level i ∈ {1, . . . , N}

Idle duration (s) Disturbance r available bandwidth (bytes/s)

xc x r y C H u

  • L. De Cicco (luca.decicco@poliba.it)

A Hybrid Model of AVS Control Systems IEEE CDC 2016 10 / 16

slide-12
SLIDE 12

Jump set D

D is the union of the sets: Dfeed = {(x, u, r) ∈ X × U × R≥0 : ∆ = Si,s} Ddwnl = {(x, u, r) ∈ X × U × R≥0 : τ = w ∧ σ = 0} Dempty = {(x, u, r) ∈ X × U × R≥0 : q = 0 ∧ dr = 1} Dplay = {(x, u, r) ∈ X × U × R≥0 : q ≥ qmin ∧ dr = 0} Dfeed End of download of segment s of level li ⇒The segment is fed to the buffer Ddwnl End of an idle period ⇒Triggers the download of the next segment (s + 1) Dempty Queue gets empty ⇒ Playback is stopped (dr = 0) Dplay Queue gets filled ⇒ Playback is started (dr = 1)

  • L. De Cicco (luca.decicco@poliba.it)

A Hybrid Model of AVS Control Systems IEEE CDC 2016 11 / 16

slide-13
SLIDE 13

Continuous and discrete dynamics (f and g)

Continuous dynamics (x, u, r) ∈ C ˙ q = −dr ˙ ∆ = σr ˙ τ = 1 (other variables’ derivatives are 0) Discrete dynamics (x, u, r) ∈ D d+

r =

  • 1,

(x, u, r) ∈ Dplay 0, (x, u, r) ∈ Dempty q+ = q + ˆ τ (x, u, r) ∈ Dfeed ∆+ = 0 (x, u, r) ∈ Dfeed σ+ =

  • (x, u, r) ∈ Dfeed

1 (x, u, r) ∈ Ddwnl τ + = 0 (x, u, r) ∈ Dfeed ∪ Ddwnl s+ = s + 1 (x, u, r) ∈ Dfeed

  • L. De Cicco (luca.decicco@poliba.it)

A Hybrid Model of AVS Control Systems IEEE CDC 2016 12 / 16

slide-14
SLIDE 14

Experimental Validation

The model is implemented in Matlab through the Hybrid Equations (HyEq) Toolbox by Sanfelice Experimental data: gathered through tests on a controlled testbed in lab

storage Video

TAPAS

Player

L Server Host Apache

HTTP Server

Client Host

Video

Emulation Internet

NetShaper

An Internet bottleneck link is emulated by a tool we developed allowing the bandwidth r to be set An adaptive video streaming client is implemented using TAPAS (https://github.com/ldecicco/tapas)

  • L. De Cicco (luca.decicco@poliba.it)

A Hybrid Model of AVS Control Systems IEEE CDC 2016 13 / 16

slide-15
SLIDE 15

Experimental Validation

Employed controller [ACC15]: a controller keeping the queue length within [qL, qH] by throttling the video level between the closest ones (li , li+1) to the available bandwidth r We compare the dynamics of:

◮ A fluid flow model of the queue length [ACC15] ◮ The proposed Hybrid system model ◮ The real system

The video is the popular benchmark “Sintel” encoded at five nominal bitrates: L = {240, 500, 900, 1400, 2600}kb/s

[ACC15] G. Cofano, L. De Cicco, S. Mascolo, “Characterizing Adaptive Video Streaming Control Systems”, in Proc. of ACC, July 2015

  • L. De Cicco (luca.decicco@poliba.it)

A Hybrid Model of AVS Control Systems IEEE CDC 2016 14 / 16

slide-16
SLIDE 16

Results: step response

Fluid Model Proposed Model Real System

100 200 300 400 10 20 30 q(t) [s] 100 200 300 400 t [s] 1000 2000 l(t) [kbps] 100 200 300 400 10 20 30 q(t) [s] 100 200 300 400 t [s] 1000 2000 l(t) [kbps] 100 200 300 400 10 20 30 q(t) [s] 100 200 300 400 t [s] 1000 2000 l(t) [kbps]

Fluid Model Proposed Model Real System

100 200 300 400 10 20 30 q(t) [s] 100 200 300 400 t [s] 1000 2000 l(t) [kbps] 100 200 300 400 10 20 30 q(t) [s] 100 200 300 400 t [s] 1000 2000 l(t) [kbps] 100 200 300 400 10 20 30 q(t) [s] 100 200 300 400 t [s] 1000 2000 l(t) [kbps]

  • L. De Cicco (luca.decicco@poliba.it)

A Hybrid Model of AVS Control Systems IEEE CDC 2016 15 / 16

slide-17
SLIDE 17

Conclusions

We have proposed a hybrid system modeling adaptive video streaming control systems The model accurately predicts the dynamics of real video streaming systems The model can be used to:

◮ Design control laws and prove properties of the closed loop system

(future work)

◮ Perform simulations of video streaming systems to aid the design

phase of controllers

  • L. De Cicco (luca.decicco@poliba.it)

A Hybrid Model of AVS Control Systems IEEE CDC 2016 16 / 16

slide-18
SLIDE 18

Thank you!

Luca De Cicco luca.decicco@poliba.it

slide-19
SLIDE 19

Results: step response

Fluid Model Proposed Model Real System

100 200 300 400 10 20 30 q(t) [s] 100 200 300 400 t [s] 1000 2000 l(t) [kbps] 100 200 300 400 10 20 30 q(t) [s] 100 200 300 400 t [s] 1000 2000 l(t) [kbps] 100 200 300 400 10 20 30 q(t) [s] 100 200 300 400 t [s] 1000 2000 l(t) [kbps]

Fluid Model Proposed Model Real System

100 200 300 400 10 20 30 q(t) [s] 100 200 300 400 t [s] 1000 2000 l(t) [kbps] 100 200 300 400 10 20 30 q(t) [s] 100 200 300 400 t [s] 1000 2000 l(t) [kbps] 100 200 300 400 10 20 30 q(t) [s] 100 200 300 400 t [s] 1000 2000 l(t) [kbps]

  • L. De Cicco (luca.decicco@poliba.it)

A Hybrid Model of AVS Control Systems IEEE CDC 2016 16 / 16