You are on page 1of 1

IEEE SIGNAL PROCESSING MAGAZINE [175] JULY 2013

ERRATA
REFERENCE
[1] G. Mateos and K. Rajawat, Dynamic network cartography, IEEE Signal Processing Mag., vol. 30, no. 3, pp. 129143, May 2013.
P
a
t
h

I
n
d
e
x
t
20 40 60 80 100
40
35
30
25
20
15
10
5
10
20
30
40
50
60
0
P
a
t
h

I
n
d
e
x
10
20
30
40
50
60
40
35
30
25
20
15
10
5
0
(ms)
t
20 40 60 80 100
(ms)
P
a
t
h

I
n
d
e
x
t
20 40 60 80 100
40
35
30
25
20
15
10
5
10
20
30
40
50
60
0
(ms)
P
a
t
h

I
n
d
e
x
t
(a)
(c)
(b)
(d)
20 40 60 80 100
40
35
30
25
20
15
10
5
10
20
30
40
50
60
0
(ms)
False-Alarm Probability
(a) (b)
D
e
t
e
c
t
i
o
n

P
r
o
b
a
b
i
l
i
t
y
0
0
0.1
0.1
0.2
0.2
0.3
0.3
0.4
0.4
0.5
0.5
0.6
0.6
0.7
0.7
0.8
0.8
0.9
0.9
1
1
PCA, r = 1
Estimator (10)
PCA, r = 2
PCA, r = 3
Spatial Anomgraphy, r = 1
Spatial Anomgraphy, r = 2
Spatial Anomgraphy, r = 3
0
200
400
0
50
100
0
2
4
6
T
im
e
In
d
e
x
(t
)
F
lo
w

I
n
d
e
x

(
f
)
A
n
o
m
a
l
y

A
m
p
l
i
t
u
d
e
In the article Dynamic Network Cartography by G. Mateos
and K. Rajawat [1], IEEE Signal Processing Magazine,
vol. 30, no. 3, pp. 129143, Figures 3 and 6 were incorrect
due to a production error. The subfigures within Figure 3
were misplaced. Part (a) should be swapped with (c), and
(b) should be swapped with (d). In the legend of Figure 6(a),
the fourth row should read Estimator (10) instead of
Estimator (17). The correct figures are printed below.
We apologize for the errors and any confusion they may
have caused.
[FIG3] True and predicted delay map for 62 paths in the Internet-2 data set [1] over an interval of 100 min.
(a) True delays. (b) Network kriging [18]. (c) Difussion wavelets [19]. (d) KKF [46]. Delays of several paths
change slightly around , t 80 = but this change is only discernible from the delay predictions offered by
KKF. Delay maps summarize the network state and are useful tools aiding operational decision in
network monitoring and control stations [46]. (Figure used with permission from [46].)
[FIG6] Unveiling anomalies from Internet-2 data [1]. (a) ROC curve comparison between (10) and the PCA
methods in [34] and [55], for different values of : ( ) S dim r n = . Leveraging sparsity and low rank jointly
leads to improved performance. (b) In red, the estimated anomaly map A
t
obtained via (10) superimposed
to the true anomalies shown in blue [38]. (Figure used with permission from [38].)
Digital Object Identifier 10.1109/MSP.2013.2258696
Date of publication: 12 June 2013

You might also like