Predicting Network Performance Characteristics - Introducing Flow - - PowerPoint PPT Presentation

Predicting Network Performance Characteristics

• Introducing Flow Field Forecasting -

Michael Frey Kyle Caudle Bucknell University South Dakota School mfrey@bucknell.edu

• f Mines & Technology

e-Weather Center Project DOE ASCR Program

• Phil DeMar
• Thomas Ndousse-Fetter
• Brian Tierney

Summer er 201 201 E ESCC/Inter ernet et2 J 2 Joint T Tec echs 1.

Outline

• The NPC forecasting problem
• Standard forecasting tools
• Flow field forecasting
• Demonstrations
• Present status

Summer er 2 201 01 ESCC/Inter ernet et2 J 2 Joint T Techs

• 2. .

The problem

Given: A sequence of observations of an NPC along a network path with given observation times

 

?

?

t1 t2 t3t4 t5t6t7 tN1tN tF tF

NPC Goal: Predict the NPC reliably at time F t

in the near future

Extrapolate plausibly at F t beyond the near future

Summer er 2 201 01 ESCC/Inter ernet et2 J 2 Joint T Techs

• 3. .

Beyond near future

Stable underlying mechanism Non-stable underlying mechanism Reliable prediction beyond near future Unreliable prediction beyond near future

2010 2000 1990 1980 1970 1960

400 390 380 370 360 350 340 330 320 310

Year Atmospheric CO2 (ppm)

David Keeling CO2 Data - Mauna Loa, Hawaii

Measured monthly

1/1/11 1/1/09 1/1/07 1/1/05 1/1/03 1/1/01

$1.60 $1.50 $1.40 $1.30 $1.20 $1.10 $1.00 $0.90 $0.80

Euro price

Euro - US Dollar Exchange Rate

Recorded daily, at close

• f EU

currency Sole

2000 1950 1900 1850 1800 1750 1700

200 150 100 50

Year Count

Sunspot cycle

Atmospheric Administration U.S. National Oceanic and noon 04:00 20:00 noon 04:00 20:00 noon 4 3 2 1

Fri 24 Jun 2011 12:00:00 EDT - Sun 26 Jun 2011 12:00:00 EDT Los Angeles Inbound (Gbits/sec)

Channel Untilization - Los Angeles/Houston Backbone

12-HOUS-LOSA-10GE-05581 BACKBONE: LOSA-HOUS 1 | rt r.losa.int ernet 2.edu--ge-6/ 1/ 0.0

Summer er 2 201 01 ESCC/Inter ernet et2 J 2 Joi

• int

nt T Techs hs

• 4. .

Desiderata

Error estimate – reliably estimates near-future prediction error Plausible – plausibly extrapolates beyond near future Robust – accepts non-uniformly spaced

• bservation times

Autonomous – no human guidance Fast – computationally efficient; e.g., no multi- dimensional numerical optimizations Accommodative – capable of exploiting “parallel” data

Summer er 2 201 01 ESCC/Inter ernet et2 J 2 Joint T Techs

• 5. .

Available Tools

Moving averages Traditional regression ARIMA forecasting Neural networks Gaussian process regression Semiparametric regression Spectral/wavelet methods

???

Error estimate X √ √ X √ √ X Plausible X X X X √ X √ Robust X √ X √ √ √ X Autonomous √ X X √ X √ √ Fast √ √ X X X √ √ Accommodative X √ X √ X X X

Summer er 2 201 01 ESCC/Inter ernet et2 J 2 Joint T Techs

• 6. .

New Forecasting Method

Moving averages Traditional regression ARIMA forecasting Neural networks Gaussian process regression Semiparametric regression Spectral/wavelet methods Flow field forecasting Error estimate X √ √ X √ √ X √ Plausible X X X X √ X √ √ Robust X √ X √ √ √ X √ Autonomous √ X X √ X √ √ √ Fast √ √ X X X √ √ √ Accommodative X √ X √ X X X √

Summer er 2 201 01 ESCC/Inter ernet et2 J 2 Joint T Techs

• 7. .

Flow Field Forecasting - 1

Step 1: Extract skeleton

n n n

S Y ε + = Skeleton           =

K K K

s s s s s δ δ δ δ δ κ κ κ κ   

3 2 1 3 2 1 3 2 1

• Semi-parametric regression
• Use only skeleton for forecast
• Data reduction
• Original time spacing not

relevant Data = Noise + Skeleton

                         

Summer er 2 201 01 ESCC/Inter ernet et2 J 2 Joint T Techs

• 8. .

Flow Field Forecasting - 2

Step 2: Interpolate flow field

Summer er 2 201 01 ESCC/Inter ernet et2 J 2 Joint T Techs 9

Flow Field Forecasting - 3

Step 3: Build to the future

0 1 2



K s0 s1 s2



sK d0 d1 d2



dK sK1 K1





 

K1 sK2 K2





 

K2 sK3 K3





 

K3 

 

Skeleton

  flowfield interpolation

Summer er 2 201 01 ESCC/Inter ernet et2 J 2 Joint T Techs 10

• 10. .

Demonstration 1

• Flow field contains much applicable information
• Prediction error builds slowly

Su Summer 201 ESC ESCC/Inter ernet et2 J 2 Joint Tec echs 11

• 11. .

Demonstration 2

• Flow field contains little applicable information
• Prediction error builds quickly

Summer er 2 201 01 ESCC/Inter ernet et2 J 2 Joint Tec echs

• 12. .

Current Status

• Python code undergoing final testing

For a copy  mfrey@bucknell.edu

• Integration with e-Weather Center

can begin in September 2011

• “Introducing Flow Field Forecasting”

by Frey and Caudle For a copy  mfrey@bucknell.edu

Summer er 2 201 01 ESCC/Inter ernet et2 J 2 Joint Tec echs

• 13. .