METRICS APPLICATION IN METROPOLITAN BROADBAND
ACCESS NETWORK SECURITY ANALYSIS
Rodrigo S. Miani, Bruno B. Zarpelão, Leonardo de Souza Mendes and Mario L. Proença Jr.
School of Electrical and Computer Engineering (FEEC), University of Campinas, Campinas, Brazil
Computer Science Departament, State University of Londrina, Londrina, Brazil
Keywords: Metropolitan networks, Security metrics, Security Analysis, Network Security.
Abstract: This work proposes the development and application of specific security metrics for metropolitan broadband
access networks that aim to measure the efficiency of security programs and support action planning against
detected problems. The approach presented in this work show metrics developed for these networks and
parameters for metrics definition. This paper also presents results achieved from application of the metrics
reported here in the metropolitan broadband access network of Pedreira, a city located in São Paulo, Brazil.
1 INTRODUCTION
Metropolitan broadband access networks (MBAN)
can be defined as the convergence of services,
applications and infrastructure to create a
community communications network of a city.
The MBAN is characterized by a variety of
services that intend to reach every sector of the
society in an universalization process of information
access. The need for data manipulation privacy and
the high number of users create new challenges
related to information security in this network’s
paradigm. Our work proposes the utilization of
security metrics as an important tool for the solution
of the questions involving the presented scenario.
CERT (2006) recommends the usage of security
metrics for dealing with the security vulnerabilities
that may occur in MBANs.
Through a combination of pre-defined
objectives, collection and data analysis, security
metrics can indicate the actual level of security we
must aim at, directing the actions that network
administrators must take to secure the network
(Patriciu et al 2006).
Weiss et. al. (2005) proposes a technique to
measure organizations security. This technique
consists in identifying several scenarios of threats
that have the same function of the metrics here
proposed, and the creation of a security index for
these scenarios. The index is calculated using the
percentage of lost assets with possible attack
scenarios. Our proposal consists in the creation of
indexes considering the current state of security
components of a metric.
The contributions of this work are: i) definition
of a security metrics model for MBANs, ii) creation
of a generic model for formulas calculation of
security metrics and standardization of names and
terms related to security metrics, iii) presentation of
five security metrics for MBANs.
This paper is organized as follows. On section 2,
we define classification concepts of MBANs. In
Section 3, we show how to build security metrics for
MBANs and the mathematical model for metric
indicator calculation. Section 4 brings a general
view of a MBAN developed in Pedreira, Brazil, and
preliminary results related to proposed metrics
utilization. In Section 5 we present the final
considerations.
2 MBAN CLASSIFICATION
A MBAN can be classified according to the three
layers model, composed by: i) Network structure, ii)
Interconnection points and iii) Services (Alexiou et
al. 2006).
The network structure of a MBAN can be based
on four technologies: optic, wireless, dedicated
access (e.g. ADSL, cable network) or hybrid.
A MBAN can also be classified according to the
type of the points connected. There are three
categories of connections: public buildings (schools,
hospitals, etc); private buildings (industries); and
473
S. Miani R., B. Zarpelão B., de Souza Mendes L. and L. Proença Jr. M. (2008).
METRICS APPLICATION IN METROPOLITAN BROADBAND ACCESS NETWORK SECURITY ANALYSIS.
In Proceedings of the International Conference on Security and Cryptography, pages 473-476
DOI: 10.5220/0001924204730476
Copyright
c
SciTePress
residences. Once the infrastructure is ready, several
services can be made available for the MBAN users,
as shown in (Ford & Koutsky, 2005). Some of these
are: Internet distribution, VoIP, E-gov systems,
information programs for education departments,
and video sharing systems.
This classification allows the treatment of
security problems separately in each of the layers of
the MBAN model. This procedure permits a better
understanding about security within each layer and,
consequently, the analysis of the respective potential
problems.
3 SECURITY METRICS APPLIED
TO MBANS
Here, we present an approach for definition and
application of security metrics that unite
methodology concepts proposed by (Payne, 2006),
(Swanson et al. 2003) and (Jaquith, 2007). The set of
attributes defined for the MBAN security metrics are
the following.
Objectives: The metrics proposed in this work
were developed to attain the following general
objective: the metrics must be able to allow an
efficient analysis of the security risks and respective
counter measures at the three levels of MBAN. This
generic objective could be used for development of
other security metrics for MBANs.
Besides the main objective above mentioned,
specific objectives for each metric may be defined.
Metrics and measures: Identification of metrics
and measures are realized from the acknowledgment
of the essential security objectives inside a MBAN
that must lead to accomplishment of these goals. The
metrics are obtained from the analysis of data
measured over the operation of the network.
Data source and frequency: The data source
includes network administrators interviews, system
logs, auditing, and tracking tools (Swanson et al.
2003). Frequency defines the periods of time that
data shall will be gathered.
Metrics classification: A metric can be
classified accordingly to the MBAN layers presented
in section 2. It means that each metric will be related
with one ore more MBAN components
Formula: Formula (Swanson et al. 2003)
describes the calculus that must be done for metric
quantification in a numeric expression. Formula’s
input data is obtained from realized measures. From
the result of the formula, it is obtained a value or
indicator for the metric that varies between 0 and 1,
with 0 for the lowest and 1 for the higher value.
3.1 Formula Modeling
Despite some interpretation flaws (Box et al. 1978),
we define the formula using the arithmetic mean.
Our goal is the creation of a indicative that
represents the security level of a metric.
Consider a metric M, which is composed by
components denoted by a
1
, a
2
, ..., a
n
and a
i
*
,
being
*
the non-negative real numbers set. The
formula F of a metric can be defined as a
relationship between the components a
1
, a
2,
..., a
n
satisfying the condition 0
F 1.
Take the set of the components a
1
, a
2
, ..., a
n
. For
each a
i
, let a
t
be the maximum value that this
measure assumes. For example, if a
1
represents the
number of infected computers, the respective
maximum value a
t
will be the total number of
computers.
Definition 1: a component a
n
is said insecure if,
when its value increases, the risks of security
problems of the correspondent metric objectives
increases too.
Definition 2: a component a
n
is said secure if,
when its value increases, the risks of security
problems of the metric objectives decrease.
Consider an insecure component a
i
. Let a
t
be its
maximum value. We denominate CI =
100
100
t
i
a
a
as
normalized insecure component. Analogously, for a
secure component a
i
with a
t
as its maximum value,
we can define CS as normalized secure component.
Let X be a set of normalized secure components,
Y a set of normalized insecure component and a
metric M. Consider the possible cases about M:
Case 1 – M is composed only by secure
components.
The formula of M will be given by the mean of
normalized secure components:
=
M
F
X
Case 2 – M is composed only by insecure
components.
The same as the case 1, but to preserve the
relationship 0 = low security and 1 = high security,
the formula will be modified as:
(
)
YF
M
= 1
Case 3 – M is composed by insecure and secure
components.
In this case, the formula will be given by the
mean between the secure and insecure components.
SECRYPT 2008 - International Conference on Security and Cryptography
474
3.2 Metrics Description
Next, we present five security metrics for MBANs.
1
) Metric: Buildings connected using
information security technologies.
Objective: To analyze and to increase security
level among MBAN buildings.
Data source: Interviews with network
administrators and network equipment auditing.
Frequency: Monthly or every two months.
Classification: Network structure.
Measures: a
t
= total number of buildings, a
1
=
number of buildings with firewall resources, a
2
=
number of buildings with secure connections.
Formula:
=
1
F
X
with X =
,,
21
tt
a
a
a
a
(1)
2) Metric: Ratio between administrator users
and desktops.
Objective: To reduce the number of desktop
users with administrative privileges in the MBAN.
Data source: Auditing tools such as Network
Management Suite (2008).
Frequency: Monthly for baseline establishment;
then, every three months.
Classification: Interconnection point.
Measures: a
t
= total number of users and a
1
=
total number of administrative users.
Formula:
(
)
YF = 1
2
with
=
t
a
a
Y
1
(2)
3) Metric: Ratio of wireless connections with
security protocols enabled.
Objective: To increase security level of wireless
connections.
.Data source: Equipment auditing, scanner tools
like AirSnort (2004).
Frequency: Monthly.
Classification: Network structure.
Measures: a
t
= total number of APs (Access
Points), a
1
= number of APs without security
protocols, including WEP, a
2
= number of APs with
default passwords
Formula:
(
)
YF = 1
3
with
=
tt
a
a
a
a
Y
21
,
(3)
4) Metric: Ratio of servers with backup and
redundancy services. Servers’ ratio uptime.
Objective: To increase availability and
reliability of MBAN servers.
Data source: Monitoring and auditing in the
servers.
Frequency: Monthly.
Metric classification: Services.
Measures: a
t1
= number of servers, a
t2
= number
of hours, a
1
= number of servers with redundancy, a
2
= number of servers in the backup program, a
3
=
number of servers with remote backup, a
4
= mean of
servers uptime.
Formula:
XF =
4
with X =
2
4
1
3
1
2
1
1
,,,
tttt
a
a
a
a
a
a
a
a
(4)
5) Metric: Internet utilization rate of MBAN.
Objective: Baseline creation for helping the
detection of: traffic, abusive use and outliers.
Data source: Traffic and bandwidth analysis
tool and auditing in the network management.
Frequency: monthly.
Metric classification: Network structure and
Services.
Measures: a
t1
= Internet access bandwidth size,
a
t2
= number of desktops that access Internet, a
1
=
Internet access average bandwidth and a
2
= number
of desktops that access Internet through external
links to the MBAN.
Formula:
(
)
YF = 1
5
with Y =
2
2
1
1
,
tt
a
a
a
a
(5)
4 CASE STUDY: MBAN OF
PEDREIRA
The MBAN of Pedreira is a project that has being
developed by the University of Campinas
(UNICAMP) and by the government of the city of
Pedreira. The project started in 2005 and officially
launched in 2007. The network infrastructure is
constituted by an optical backbone and wireless
access points, assembled in the form of wireless
microcells. The current available services running
over the MBAN are: Internet distribution, VoIP, E-
mail and City surveillance.
The application of the security metrics presented
in section 3.2 to the MBAN of Pedreira was
executed in the period between January and April of
2008. Table 1 summarizes some results, such as
components and formula of each metric.
In metric 1, all the buildings have logical access
control between the connections (access lists).
METRICS APPLICATION IN METROPOLITAN BROADBAND ACCESS NETWORK SECURITY ANALYSIS
475
However, the WEP protocol is used by the eight
buildings with secure connections.
Table 1: Components and Formula.
Metric Components Formula
1) a
t
=18; a
1
=18; a
2
=8
F
1
= 0.7222
2)
a
t
= 116; a
1
= 111 F
2
= 0.0431
3)
a
t
=18; a
1
=18; a
2
= 0 F
3
= 0.5
4)
a
t1
= 3; a
t2
= 768 hours; a
1
= 0;
a
2
= 3; a
3
= 2; a
4
= 767 hours
and 34 minutes.
F
4
=
0.66652
5)
a
t1
=8 mbits/s; a
t2
=214;
a
1
=0.87844; a
2
= 6
F
5
=
0.93107
The low indicator of metric 2 shows that there is
not a defined policy for users creation in Pedreira’s
network. In metric 3, no default password was
found, but all APs use WEP.
In metric 4, the final result was affected by the
lack of redundancy component. The metric 5 result
reveals that the use of Internet was not abusive
because, on the average only 0.1098% of download
bandwidth and 0.06915% of upload bandwidth is in
use.
The proposed security controls from the results
of security metrics application are: design of a test
suite for the switches access lists, cryptography
implementation in the connections, deployment of a
Domain Controller, to change WEP by WPA,
passwords auditing in the APs, RAID
implementation for data redundancy and Heartbeat
(Hearbeat, 2007) for services.
5 CONCLUSIONS
The successful deployment of MBANs it depends on
the reliability of the systems that constitute them.
Warranty of this reliability can be obtained through
well formed criteria of information security. The
security metrics are tools that can accomplish such
objectives when properly developed and applied.
The metrics applications presented in this work
allow the visualization of security critical areas on
the MBAN of Pedreira. Security controls were
proposed, following the obtained results for this
network. It is important to note that for a complete
network analysis, a larger set of security metrics
must be developed and implemented.
Future work includes the development of new
security metrics for MBANs using the template
proposed here. Another topic is the development of a
framework for data analysis in security metrics.
ACKNOWLEDGEMENTS
The work presented here has been developed under
the umbrella of the projects “Municipal Infovia – An
Open Access Network for Cities” and “SIGM – An
Integrated e-Gov Environment for Cities”. These
projects have been supported in part by the
governments of the cities of São José do Rio Preto,
Pedreira, Penápolis, and Campinas, São Paulo State,
Brazil.
Bruno Bogaz Zarpelão’s work is supported by
the State of São Paulo Research Foundation
(FAPESP).
REFERENCES
AirSnort. 2004. [online] [Accessed 15th April 2008]
Available from World Wide Web
<http://airsnort.shmoo.com/>
Alexiou, A., Bouras, C., Primpas, D., 2006. Design
Aspects of open municipal broadband networks. In
Access Nets ’06, Proceedings of the 1st international
conference on Access networks. ACM Press.
Box, G., Hunter, W., Hunter, J., 1978. Statistcs for
Experimenters. Wiley Series in Probability and
Mathematical Statistics.
CERT. (Unpublished, 2006). Potential Vulnerabilities in
Municipal Communications Network. Report dated
May 2006.
Ford, G., Koutsky, T., 2005. Broadband and economic
development: a municipal case study from Florida.
Review of Urban & Regional Development Studies.
HeartBeat. 2008. [online] [Accessed 15th April 2008]
Available from World Wide Web http://www.linux-
ha.org/Heartbeat
Jaquith, A., 2007. Security Metrics – Replacing Fear,
Uncertainty and Doubt. Addison-Wesley.
Network Management Suite. 2008. [online] [Accessed 5th
April 2008] Available from World Wide Web:
<http://www.mishelpers.com/network_management/>
Patriciu, V., Priescu, I., Nicolaescu, S., 2006. Security
metrics for enterprise information systems. Journal of
Applied Quantitative Methods.
Payne, S., 2006. A Guide to Security Metrics. SANS
Security Essentials Version 1.2e.
Swanson, M., Bartol N., Sabato, J., Hash., J. Graffo, L.,
2003. Security Metrics Guide for Information
Technology Systems. NIST Special Publication 800-5.
Weiss, S., Weissmann, O., Dressler, F., 2005. A
Comprehensive and Comparative Metric for
Information Security. In ICTSM2005, Proceedings of
IFIP International Conference on Telecommunication
Systems, Modeling and Analysis
SECRYPT 2008 - International Conference on Security and Cryptography
476