Top Banner
Mezić Research Group SIAM Conference on Applications of Dynamical Systems May 22-26, 2011 Snowbird, Utah Ryan Mohr and Igor Mezić LOW-DIMENSIONAL MODELS FOR TCP-LIKE NETWORKS USING THE KOOPMAN OPERATOR
39

R. Mohr: Low-dimensional Models for TCP-like Networks Using the Koopman Operator

Mar 13, 2015

Download

Documents

Presentation delivered at SIAM DS11 in May, 2011 at Snowbird, UT.
Welcome message from author
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Page 1: R. Mohr: Low-dimensional Models for TCP-like Networks Using the Koopman Operator

Mezić Research Group

SIAM Conference on Applications of Dynamical SystemsMay 22-26, 2011 Snowbird, Utah

Ryan Mohr and Igor Mezić

LOW-DIMENSIONAL MODELS FOR

TCP-LIKE NETWORKS USING THE

KOOPMAN OPERATOR

Page 2: R. Mohr: Low-dimensional Models for TCP-like Networks Using the Koopman Operator

Monday, May 23, 2011 Can change this on the Master Slide

Mezić Research Group

Overview

2

Page 3: R. Mohr: Low-dimensional Models for TCP-like Networks Using the Koopman Operator

Monday, May 23, 2011 Can change this on the Master Slide

Mezić Research Group

Overview• Interested in characterizing network dynamics

Periodicity

Global coherent behavior etc.

2

Page 4: R. Mohr: Low-dimensional Models for TCP-like Networks Using the Koopman Operator

Monday, May 23, 2011 Can change this on the Master Slide

Mezić Research Group

Overview• Interested in characterizing network dynamics

Periodicity

Global coherent behavior etc.

• Build a low-dimensional model

2

Page 5: R. Mohr: Low-dimensional Models for TCP-like Networks Using the Koopman Operator

Monday, May 23, 2011 Can change this on the Master Slide

Mezić Research Group

Overview• Interested in characterizing network dynamics

Periodicity

Global coherent behavior etc.

• Build a low-dimensional model

• What is the appropriate way to compare model and data?

2

Page 6: R. Mohr: Low-dimensional Models for TCP-like Networks Using the Koopman Operator

Monday, May 23, 2011 Can change this on the Master Slide

Mezić Research Group

Overview• Interested in characterizing network dynamics

Periodicity

Global coherent behavior etc.

• Build a low-dimensional model

• What is the appropriate way to compare model and data?

2

Koopman Operator

Spectral Analysis

Page 7: R. Mohr: Low-dimensional Models for TCP-like Networks Using the Koopman Operator

Monday, May 23, 2011 Can change this on the Master Slide

Mezić Research Group

Overview• Interested in characterizing network dynamics

Periodicity

Global coherent behavior etc.

• Build a low-dimensional model

• What is the appropriate way to compare model and data?

2

Page 8: R. Mohr: Low-dimensional Models for TCP-like Networks Using the Koopman Operator

Monday, May 23, 2011 Can change this on the Master Slide

Mezić Research Group

Overview• Interested in characterizing network dynamics

Periodicity

Global coherent behavior etc.

• Build a low-dimensional model

• What is the appropriate way to compare model and data?

2

Koopman Delay Embeddings

Pseudo-metrics

Page 9: R. Mohr: Low-dimensional Models for TCP-like Networks Using the Koopman Operator

Monday, May 23, 2011 Can change this on the Master Slide

Mezić Research Group

Assumptions on the system• Discrete-time dynamical system

3

(X, T ) T : X → X

• Compact state space Variables for each router• buffer size

• congestion window

• roundtrip timer

• etc,

• Continuous Transformation TCP protocol

• On an attractor

• Moderately size network = large state space

Page 10: R. Mohr: Low-dimensional Models for TCP-like Networks Using the Koopman Operator

Monday, May 23, 2011 Can change this on the Master Slide

Mezić Research Group

Observables• Introduce observable on the system

assume observable is continuous, measurable

4

g : X → R

X

g

R

Page 11: R. Mohr: Low-dimensional Models for TCP-like Networks Using the Koopman Operator

Monday, May 23, 2011 Can change this on the Master Slide

Mezić Research Group

Observables• Introduce observable on the system

assume observable is continuous, measurable

4

g : X → R

• Choose functions that are relevant link loads for a subset of paths

load on a routers Traffic matrices

X

g

R

Page 12: R. Mohr: Low-dimensional Models for TCP-like Networks Using the Koopman Operator

Monday, May 23, 2011 Can change this on the Master Slide

Mezić Research Group

Koopman Operator, U• Interested in how the transformation drives the dynamics on

the real line

• Introduce the Koopman operator

Properties: Linear & Bounded (when integrating against a preserved measure)

• Interpretation1) Operator acting on space of functions = evolves observables, or2) For each point in phase space, get a time series in :

5

(Ukg)(x) := g(T kx)

Ryk = (Ukg)(x)

Page 13: R. Mohr: Low-dimensional Models for TCP-like Networks Using the Koopman Operator

Monday, May 23, 2011 Can change this on the Master Slide

Mezić Research Group

Spectral Decomposition of U• Want to decompose operator as action of projections onto

eigenspaces

• Let

6

U =�

j

λjPj

Uϕj(x) = λjϕj(x)

g ∈ span{ϕj} =⇒ g(x) =�

j

cjϕj(x)

=⇒ Ukg(x) =�

j

λkj cjϕj(x)

Page 14: R. Mohr: Low-dimensional Models for TCP-like Networks Using the Koopman Operator

Monday, May 23, 2011 Can change this on the Master Slide

Mezić Research Group

Spectral Decomposition of U• Want to decompose operator as action of projections onto

eigenspaces

• Let

• How do we compute the projection?

7

U =�

j

λjPj

Uϕj(x) = λjϕj(x)

g ∈ span{ϕj} =⇒ g(x) =�

j

cjϕj(x)

=⇒ Ukg(x) =�

j

λkj cjϕj(x)� �� �Pjg(x)

Page 15: R. Mohr: Low-dimensional Models for TCP-like Networks Using the Koopman Operator

Monday, May 23, 2011 Can change this on the Master Slide

Mezić Research Group

Harmonic Projections• Define class of operators

• The limit is an eigenfunction of the Koopman operator :

8

Ug∗ω(x) = ei2πkωg∗ω(x)

Uk =�

ω

ei2πkωPωT

PωT g(x) := lim

n→∞

1n

n−1�

k=0

e−i2πkωUkT g(x) =: g∗ω(x)

Page 16: R. Mohr: Low-dimensional Models for TCP-like Networks Using the Koopman Operator

Monday, May 23, 2011 Can change this on the Master Slide

Mezić Research Group

Abilene Data Set*

9

* http://www.cs.utexas.edu/~zhang/research/AbileneTM

Page 17: R. Mohr: Low-dimensional Models for TCP-like Networks Using the Koopman Operator

Monday, May 23, 2011 Can change this on the Master Slide

Mezić Research Group

Abilene Data Set*

• Data collected over 24 weeks

9

* http://www.cs.utexas.edu/~zhang/research/AbileneTM

Page 18: R. Mohr: Low-dimensional Models for TCP-like Networks Using the Koopman Operator

Monday, May 23, 2011 Can change this on the Master Slide

Mezić Research Group

Abilene Data Set*

• Data collected over 24 weeks

• Sampling period of 5 minutes = 48,384 observations

9

* http://www.cs.utexas.edu/~zhang/research/AbileneTM

Page 19: R. Mohr: Low-dimensional Models for TCP-like Networks Using the Koopman Operator

Monday, May 23, 2011 Can change this on the Master Slide

Mezić Research Group

Abilene Data Set*

• Data collected over 24 weeks

• Sampling period of 5 minutes = 48,384 observations

• Observable = Traffic Matrices

9

* http://www.cs.utexas.edu/~zhang/research/AbileneTM

Page 20: R. Mohr: Low-dimensional Models for TCP-like Networks Using the Koopman Operator

Monday, May 23, 2011 Can change this on the Master Slide

Mezić Research Group

Abilene Data Set*

• Data collected over 24 weeks

• Sampling period of 5 minutes = 48,384 observations

• Observable = Traffic Matrices Measures the amount of traffic between input node and output node

9

* http://www.cs.utexas.edu/~zhang/research/AbileneTM

Page 21: R. Mohr: Low-dimensional Models for TCP-like Networks Using the Koopman Operator

Monday, May 23, 2011 Can change this on the Master Slide

Mezić Research Group

Abilene Data Set*

• Data collected over 24 weeks

• Sampling period of 5 minutes = 48,384 observations

• Observable = Traffic Matrices Measures the amount of traffic between input node and output node

Quantities computed via some estimated procedure

9

* http://www.cs.utexas.edu/~zhang/research/AbileneTM

Page 22: R. Mohr: Low-dimensional Models for TCP-like Networks Using the Koopman Operator

Monday, May 23, 2011 Can change this on the Master Slide

Mezić Research Group

Abilene Data Set*

• Data collected over 24 weeks

• Sampling period of 5 minutes = 48,384 observations

• Observable = Traffic Matrices Measures the amount of traffic between input node and output node

Quantities computed via some estimated procedure Unfortunately, data for only 1 initial condition

9

* http://www.cs.utexas.edu/~zhang/research/AbileneTM

Page 23: R. Mohr: Low-dimensional Models for TCP-like Networks Using the Koopman Operator

Monday, May 23, 2011 Can change this on the Master Slide

Mezić Research Group

Abilene Data Set*

• Data collected over 24 weeks

• Sampling period of 5 minutes = 48,384 observations

• Observable = Traffic Matrices Measures the amount of traffic between input node and output node

Quantities computed via some estimated procedure Unfortunately, data for only 1 initial condition

Have 144 different observables

9

* http://www.cs.utexas.edu/~zhang/research/AbileneTM

Page 24: R. Mohr: Low-dimensional Models for TCP-like Networks Using the Koopman Operator

Can change this on the Master Slide

Mezić Research Group

Monday, May 23, 2011 10

• Horizontal streaks = source/destination pair having continuous spectrum

• Vertical Streaks = majority of the network has a component at that frequency; globally relevant frequency

Page 25: R. Mohr: Low-dimensional Models for TCP-like Networks Using the Koopman Operator

Can change this on the Master Slide

Mezić Research Group

Monday, May 23, 2011 11

Page 26: R. Mohr: Low-dimensional Models for TCP-like Networks Using the Koopman Operator

gi(x)

Pωgi(x)

{ωj , Aj , θj}

yi(k) = A0 + 2D�

j=1

Aj cos(2πωjk + θj)

Monday, May 23, 2011 Can change this on the Master Slide

Mezić Research Group

12

Building a Low-Dimensional Model

Harmonic Projections

Pω ∀ω ∈ [−0.5.0.5]

Cluster&

Order Modes

Extract first D modes

ω ∈ (0, 0.5)

Low-dimensional deterministic “skeleton”

Page 27: R. Mohr: Low-dimensional Models for TCP-like Networks Using the Koopman Operator

Monday, May 23, 2011 Can change this on the Master Slide

Mezić Research Group

Comparing Data and ModelTo Measure Closeness : Introduce family of pseudo-metrics to compare discretized measures in a delay space

13

Page 28: R. Mohr: Low-dimensional Models for TCP-like Networks Using the Koopman Operator

Monday, May 23, 2011 Can change this on the Master Slide

Mezić Research Group

Comparing Data and ModelTo Measure Closeness : Introduce family of pseudo-metrics to compare discretized measures in a delay space

13

Re

{yi(0), yi(1), . . . }πe

{

yi(0)

...yi(e− 1)

,

yi(1)

...yi(e)

, · · · }

Delay Embedding

• Delay Space:

Page 29: R. Mohr: Low-dimensional Models for TCP-like Networks Using the Koopman Operator

≡ limk→∞

1n

n�

k=0

UkT (IBi ◦ πe ◦ g(x))

px,T (Bi) := P 0T (IBi ◦ πe ◦ g(x))

Monday, May 23, 2011 Can change this on the Master Slide

Mezić Research Group

Comparing Data and Model• Computing measures in delay space :

Define a regular grid in delay space

Let be indicator function on a box in grid Empirical measure :

14

≡ limk→∞

1n

n�

k=0

IBi(

y(k)

...y(k + e− 1)

)

IBi

Page 30: R. Mohr: Low-dimensional Models for TCP-like Networks Using the Koopman Operator

≡ limk→∞

1n

n�

k=0

e−i2πωkUkT (IBi ◦ πe ◦ g(x))

≡ limk→∞

1n

n�

k=0

e−i2πωkIBi(

y(k)

...y(k + e− 1)

)

px,ω,T (Bi) := PωT (IBi ◦ πe ◦ g(x))

Monday, May 23, 2011 Can change this on the Master Slide

Mezić Research Group

Comparing Data and Model• Computing measures in delay space :

Spectral measures :

15

Page 31: R. Mohr: Low-dimensional Models for TCP-like Networks Using the Koopman Operator

Monday, May 23, 2011 Can change this on the Master Slide

Mezić Research Group

Comparing Data and Model• Pseudo-metrics :

For each relevant frequency

16

dx,ω(Td, Tm) =b�

i=1

|px,ω,Td(Bi)− px,ω,Tm(Bi)|

Dx(Td, Tm) = κ0dx,0(Td, Tm) +�

j

κjdx,ωj (Td, Tm)

{κj} = weights for each frequency

Page 32: R. Mohr: Low-dimensional Models for TCP-like Networks Using the Koopman Operator

x(k) = 40 cos(2π

√2

20k + π/4) + 5 cos(2π

110

k)

e = 2 d0 = 0.5388

d0.0707 = 1.1766

D(Td, Tm) = 0.8573

Modes = 2

Can change this on the Master Slide

Mezić Research Group

Monday, May 23, 2011 17

50 40 30 20 10 0 10 20 30 40 5050

40

30

20

10

0

10

20

30

40

50Delay Space Trajectories

y(k)

y(k+

1)

DataModel

Toy Example

Page 33: R. Mohr: Low-dimensional Models for TCP-like Networks Using the Koopman Operator

d0.0707 = 1.0933d0 = 0.0161

d0.1 = 0.443

D(Td, Tm) = 0.5175

Modes = 3

Can change this on the Master Slide

Mezić Research Group

Monday, May 23, 2011 18

50 40 30 20 10 0 10 20 30 40 5050

40

30

20

10

0

10

20

30

40

50Delay Space Trajectories

y(k)

y(k+

1)

DataModel

Page 34: R. Mohr: Low-dimensional Models for TCP-like Networks Using the Koopman Operator

D(Td, Tm) = 0.8388

e = 2

Modes = 3

Can change this on the Master Slide

Mezić Research Group

Monday, May 23, 2011 19

Abilene

Page 35: R. Mohr: Low-dimensional Models for TCP-like Networks Using the Koopman Operator

e = 2

Modes = 14

D(Td, Tm) = 0.2124

Can change this on the Master Slide

Mezić Research Group

Monday, May 23, 2011 20

Abilene

Page 36: R. Mohr: Low-dimensional Models for TCP-like Networks Using the Koopman Operator

Monday, May 23, 2011 Can change this on the Master Slide

Mezić Research Group

21

A few comments

Page 37: R. Mohr: Low-dimensional Models for TCP-like Networks Using the Koopman Operator

Monday, May 23, 2011 Can change this on the Master Slide

Mezić Research Group

21

A few comments1. We presented the theory for deterministic dynamical systems;

everything carries through for the stochastic case.

(a) Define a random dynamical system and the stochastic Koopman operator.

Page 38: R. Mohr: Low-dimensional Models for TCP-like Networks Using the Koopman Operator

Monday, May 23, 2011 Can change this on the Master Slide

Mezić Research Group

21

A few comments1. We presented the theory for deterministic dynamical systems;

everything carries through for the stochastic case.

(a) Define a random dynamical system and the stochastic Koopman operator.

2. The analysis shows the network has deterministic period behavior, but also that the continuous portion of the spectrum is quite important to get the dynamics right.

Page 39: R. Mohr: Low-dimensional Models for TCP-like Networks Using the Koopman Operator

Monday, May 23, 2011 Can change this on the Master Slide

Mezić Research Group

21

A few comments1. We presented the theory for deterministic dynamical systems;

everything carries through for the stochastic case.

(a) Define a random dynamical system and the stochastic Koopman operator.

2. The analysis shows the network has deterministic period behavior, but also that the continuous portion of the spectrum is quite important to get the dynamics right.

3. The variational norm for the measures needs to be replaced by a more suitable metric; the norm misbehaves when the discretization is refined and can get arbitrarily bad.