Robust Network Compressive Sensing Lili Qiu UT Austin NSF Workshop Nov. 12, 2014.

Robust Network Compressive Sensing

Lili QiuUT Austin

NSF WorkshopNov. 12, 2014

Network Matrices and Applications

• Network matrices– Traffic matrix– Loss matrix– Delay matrix– Channel State Information (CSI) matrix– RSS matrix

2

3

Q: How to fill in missing values in a matrix?

1

3

2router

flow 1

flow 3

flow 2

link 2link 1

link 3

flow 1

flow 2

flow 3

time 1 time 2 …

• Applications need complete network matrices– Traffic engineering– Spectrum sensing– Channel estimation– Localization– Multi-access channel design– Network coding, wireless video coding– Anomaly detection– Data aggregation– …

Missing Values: Why Bother?4

subcarrier 1

subcarrier 2

subcarrier 3

time 1 time 2 …

Vacant

freq ,loc1

freq 2, loc1

freq 3, loc1

time 1 time 2 …

5

The Problem

6,3

6,2

6,1

5,3

5,2

5,1

4,13,32,3

4,13,22,2

4,13,12,1

1,3

1,2

1,1

x

x

x

x

x

x

xxx

xxx

xxx

x

x

x

X

Interpolation: fill in missing values from incomplete, erroneous, and/or indirect measurements

Anomaly FutureMissing

x1,3

State of the Art

• Existing works exploit low-rank nature of network matrices

• Many factors contribute to network matrices– Anomalies, measurement errors, and noise– These factors may destroy low-rank

structure and spatio-temporal locality– Limit the effectiveness of existing works

6

Network Matrices

Network Date Duration Size (flows/links x #timeslot)

3G traffic 11/2010 1 day 472 x 144

WiFi traffic 1/2013 1 day 50 x 118

Abilene traffic 4/2003 1 week 121 x 1008

GEANT traffic 4/2005 1 week 529 x 672

1 channel CSI 2/2009 15 min. 90 x 9000

Multi. Channel CSI

2/2014 15 min. 270 x 5000

Cister RSSI 11/2010 4 hours 16 x 10000

CU RSSI 8/2007 500 frames 895 x 500

Umich RSS 4/2006 30 min. 182 x 3127

UCSB Meshnet 4/2006 3 days 425 x 1527

7

Rank Analysis8

Adding anomalies increases ranks in all cases.

Without anomalies With anomalies

LENS Decomposition: Basic Formulation

9

= +

y1,30…

…

…y3,n

0 0 0000 00000 0

000 0 0

+

[Input] D:Original matrix

[Output] X:A low rank matrix (r «

m,n)

[Output] Y:A sparse anomaly

matrix

[Output] Z:A small noise

matrix

d1,3d1,2

…

…

…

d2,n

dm,n

d3,n

d1,4

d2,1 d2,2 d2,3

d3,1 d3,4

dm,2 dm,4

…

xm,r

xr,n

…

x1,1

x2,1

x3,1

xm,1

x3,r

x2,r

x1,r

x1,1

xr,1 xr,2

x1,2

xr,3

x1,3 x1,n… …

…

…

…

LENS Decomposition: Basic Formulation

• Formulate it as a convex opt. problem:

10

min:

subject to:

= + +d1,3

d1,2 d1,4

[Input] D:Original matrix

x1,2 x1,4

[Output] X:A low rank

matrix

0 0 y1,3 0

0 0 0 0

0 0 0 0

[Output] Y:A sparse anomaly matrix

[Output] Z:A small noise

matrix

α βσ

LENS Decomposition: Support Indirect Measurement

• The matrix of interest may not be directly observable (e.g., traffic matrices)– AX + BY + CZ + W = D

• A: routing matrix• B: an over-complete anomaly profile matrix• C: noise profile matrix

11

,t,t,t xxy 321 1

3

2router

flow 1

flow 3

flow 2

link 2link 1

link 3

LENS Decomposition: Account for Domain Knowledge

• Domain Knowledge– Temporal stability– Spatial locality– Initial solution

12

min:

subject to:

Optimization Algorithm

• One of many challenges in optimization:– X and Y appear in multiple places in the objective

and constraints– Coupling makes optimization hard

• Reformulation for optimization by introducing auxiliary variables

•

13

min:

subject to:

Optimization Algorithm• Alternating Direction Method (ADM)

– For each iteration, alternate among the optimization of the augmented Lagrangian function by varying each one of X, Xk, Y, Y0, Z, W, M, Mk, N while fixing the other variables

– Improve efficiency through approximate SVD

14

Setting Parameters

• • •

where (mX,nX) is the size of X, (mY,nY) is the size of Y, η(D) is the fraction of entriesneither missing or erroneous, θ is a control parameter that limitscontamination of dense measurement noise

15

min:α βσ σ

Setting Parameters (Cont.)

• ϒ reflects the importance of domain knowledge– e.g. temporal-stability varies across traces

• Self-tuning algorithm– Drop additional entries in the matrix– Quantify the error of the entries that were

present in the matrix but dropped intentionally during the search

– Pick ϒ that gives lowest error

16

min:σ

γ

17

Algorithms Compared

Algorithm Description

Baseline Baseline estimate via rank-2 approximation

SVD-base SRSVD with baseline removal

SVD-base +KNN Apply KNN after SVD-base

SRMF [SIGCOMM09] Sparsity Regularized Matrix Factorization

SRMF+KNN [SIGCOMM09]

Hybrid of SRMF and KNN

LENS Robust network compressive sensing

Self Learned ϒ18

Best ϒ = 0 Best ϒ = 1 Best ϒ = 10

No single ϒ works for all traces.Self tuning allows us to automatically select the best ϒ.

Interpolation under anomalies19

CU RSSI

LENS performs the best under anomalies.

Interpolation without anomalies20

CU RSSI

LENS performs the best even without anomalies.

Conclusion

• Main contributions– Important impact of anomalies in matrix

interpolation– Decompose a matrix into

• a low-rank matrix, • a sparse anomaly matrix, • a dense but small noise matrix

– An efficient optimization algorithm– A self-learning algorithm to automatically tune the

parameters• Future work

– Applying it to spectrum sensing, channel estimation, localization, etc.

21

22

Thank you!

Evaluation Methodology

• Metric– Normalized Mean Absolute Error for missing values

• Report the average of 10 random runs• Anomaly generation

– Inject anomalies to a varying fraction of entries with varying sizes

• Different dropping models

23

Summary of Other Results

• The improvement of LENS increases with anomaly sizes and # anomalies.

• LENS consistently performs the best under different dropping modes.

• LENS yields the lowest prediction error.

• LENS achieves higher anomaly detection accuracy.

24