ESTIMATION OF THE SECURITY LEVEL IN A MOBILE AND
UBIQUITOUS ENVIRONMENT BASED ON THE SEMANTIC
WEB
Reijo Savola
VTT Electronics, P. O. Box 1100, FIN-90571 Oulu, Finland
Keywords: Information security, security metrics, Semantic Web, mobile ad hoc networks
Abstract: The emerging Semantic Web enables semantic discover
y and systematic maintenance of information that
can be used as reference data when estimating the security level of a network, or a part of it. Using suitable
security metrics and ontologies, nodes can estimate the level of security from both their own and the
network’s point of view. The most secure applications and communication peers can be selected based on
estimation results. In this paper we discuss security level estimation in a mobile and ubiquitous environment
based on the Semantic Web. An interdisciplinary security information framework can be built using the
Semantic Web to offer metrics and security level information for product quality, the traffic and mobility
situation, general statistical knowledge and research results having an effect on the security level.
1 INTRODUCTION
Semantic Web (Berners-Lee et al., 2001) is an
extension of the current Web, in which information
is given well-defined meaning, better enabling
computers and people to work in cooperation. The
Semantic Web provides an infrastructure that
enables not just web pages, but databases, services,
programs, sensors, personal devices, and household
appliances to both consume and produce data on the
web (Hendler et al., 2002). Semantic Web agents are
autonomous goal-directed agents that can act in
cooperation with other agents and establish their
own trust and reputation databases. The agents can
seek information according to their goals and
flexibly negotiate their interaction models with other
agents.
Another emerging paradigm, mob
ile and
ubiquitous computing, aims at providing the
technological means of offering user-friendly
information and communication services, anywhere
and anytime. The ubiquitous computing scenarios
are expected to involve a great number of small,
handheld, wireless computing devices that enable
interaction between users, environment and
computing elements. Mobile ad hoc networks
(MANETs) (IETF, 2004) have great potential for
broad use in making ubiquitous computing possible
and successful, enabling self-organization and
dynamic operation.
Ubiquitous computing can clearly benefit from
th
e Semantic Web, which provides the infrastructure
for the extensive usage of distributed knowledge.
Devices that use the Semantic Web are able to
combine information and functionality from local
and remote sources, as well as to configure
themselves in new environments.
The current security resea
rch effort for the
Semantic Web concentrates on trust – particularly
trust management and trust negotiation. Trust
negotiation is the iterative disclosure of credentials
and requests for credentials between two parties,
with the goal of establishing sufficient trust that the
parties can complete a transaction (Winslett et al.,
2002). Digital credentials on the Semantic Web are
attributes similar to those one uses in human society
to deem trust. In addition to trust negotiation and
management, the new possibilities offered by the
Semantic Web can be used to raise the overall
security of a network by estimating the security level
and selecting applications and connections based on
it. Many kinds of interdisciplinary information affect
this level, e.g. product quality, human factors, trust
management, cryptographic strength, and chosen
algorithms. Statistical security level information can
be stored in databases and systematically updated
using trusted searches in the Semantic Web. The
databases can be at individual nodes’ disposal to
256
Savola R. (2005).
ESTIMATION OF THE SECURITY LEVEL IN A MOBILE AND UBIQUITOUS ENVIRONMENT BASED ON THE SEMANTIC WEB.
In Proceedings of the Seventh International Conference on Enterprise Information Systems, pages 256-262
DOI: 10.5220/0002551802560262
Copyright
c
SciTePress
support their self-organized security level
estimation.
The main contributions of this work are in the
introduction of a mechanism that uses databases on
the Semantic Web to carry out security level
estimation, and in the identification of the type of
component metrics that are needed for an example
case of mobile ad hoc networks, the connectivity
basis of ubiquitous computing. However, the same
estimation mechanism can be used in general on the
Semantic Web.
The rest of the paper is organized in the
following way. Section 2 gives overviews of
security metrics and the security concerns of mobile
ad hoc networks. Section 3 introduces our proposal
for estimation of the security level. Finally, Section
4 represents conclusions and directions for further
research. The related work consists of security
research work in MANETs and the information
management solutions on the Semantic Web.
2 BACKGROUND
2.1 Security Metrics
There is often considerable controversy when the
term “metrics” is used. The difference between
measurements and metrics is the following.
Measurements provide a one-time view of specific
measurable parameters and are represented by
numbers, weights or binary statements. On the other
hand, metrics are produced by taking measurements
over time and comparing two or more measurements
with predefined baselines, thus providing a means
for interpretation of the collected data (Jelen, 2000).
Synonyms for metric are, e.g., measure, score,
rating, rank, or assessment result (Henning, 2001).
The wide majority of the available security metrics
approaches have been developed for evaluating the
maturity of security engineering processes. The most
widely used of these maturity models is the Systems
Security Engineering Capability Maturity Model
SSE-CMM (ISO/IEC 21827, 2002). Another well-
known model, Trusted Computer Security
Evaluation Criteria TCSEC – “The Orange Book” –
(U.S. Department of Defense, 1985), expresses the
security engineering process using classes and
divisions as evaluation levels. Although a high level
of security engineering may tend to give a higher
level of technical security, it cannot be guaranteed.
Although it is essential to measure the security
engineering process, we here focus on technical
security metrics. The object to be measured in
technical metrics is the actual system, not the
associated processes. Technical security metrics can
be used to describe, and hence compare, technical
objects – e.g. algorithms, specifications,
architectures and alternative designs, products, and
as-implemented systems at different stages of the
system’s lifecycle. In general, metrics are found
most useful when they can be used proactively
predicting or trying to understand the future
situation.
(Jonsson, 2003) sorts the methods of
security measurement into the following:
Risk analysis is an estimation of the probability
of specific threats and vulnerabilities and their
consequences and costs;
Certification is the classification of the system
in classes based on the design characteristics and
security mechanisms; and
Measures of the intrusion process is a
statistical measurement of a system based on the
effort it takes to make an intrusion.
Technical security metrics can be used in the
following ways:
Goal establishment;
Prediction before implementation or in an
implemented system;
Comparison of the security level of technical
objects;
Monitoring or scanning the security level of an
object; and
Enabling analysis: e.g., metrics enable analysis
in fault injection testing.
2.2 Security Metrics for Mobile Ad
Hoc Networks
As mobile ad hoc networks have the potential to
offer the underlying connectivity for the mobile and
ubiquitous environment, we here investigate their
security concerns. The ultimate goal of the security
solutions for MANETs is to provide services for the
desired security needs, mainly confidentiality,
integrity, availability, authentication and non-
repudiation, at the desired security level. Table 1
presents the typically needed security services and
attack types in MANETs.
In general, the research has noted that traditional
security solutions, such as public key infrastructures
or authentication mechanisms, also have potential
for MANETs, but in many cases they are not
sufficient by themselves. Overviews of the research
efforts can be found in (Hubaux et al., 2001), (Yang
et al., 2004) and (Zhou & Haas, 1999). The nature of
the basic mechanisms of the ad hoc paradigm causes
vulnerabilities, e.g.:
ESTIMATION OF THE SECURITY LEVEL IN A MOBILE AND UBIQUITOUS ENVIRONMENT BASED ON THE
SEMANTIC WEB
257
Lack of central administration,
Routing: routing mechanisms are more
vulnerable than in conventional networks
because each node can act as a relay;
Co-operation: if a node does not respect the co-
operation rules – i.e. it is selfish – the
performance of the network can be severely
affected;
Variation in memory and computation
resources: many of the nodes are expected to be
low-priced consumer electronics with cheap and
slow computation capability and limited storage
size, and
Energy constrained operation: many of the
nodes are expected to operate on battery power.
Sleep or standby modes are used to conserve
energy, during which they may not be reachable.
Sleep deprivation torture is used by attackers.
Table 2 lists some component security metrics areas
(Savola, 2004), a composition of which forms the
basis for estimation of the overall security level in
mobile ad hoc networks. The most critical
component metrics emphasize trusted information
distribution in a mobile ad hoc network. In this
context, distribution means the location of the
critical information in the network with regard to
time. Trusted information includes key, routing,
mobile entity identity and packet forwarding
information. Technical challenges, such as the
trusted information distribution, dominate the
overall security level in the first stages of the
technical evolution of MANETS. As technology
matures, aspects such as product quality become
more emphasized. To some extent, the component
metrics structure of MANETs shown in Table 2 is
the similar to that in other types of networks. Mobile
ad hoc networks set very challenging security
requirements because of their infrastructure-less
structure.
Table 1: Security needs and attack types for MANETs
DIMENSION GOAL
Confidentiality Critical information never disclosed to unauthorized entities
Integrity A message being transferred is never corrupted
Availability Prevention of / survivability from denial of service attacks
Authentication
Ensuring correct identity of a node
Non-repudiation
Origin of a message cannot deny having sent the message
ATTACK TYPE EXPLANATION
Passive eavesdropping
Discovery of desired information by listening to routing data. Detection of this type of
attack is challenging.
Denial of service Produced either by unintentional failure or malicious action.
Impersonation Nodes joining the network undetectably, or sending false routing information (black hole
and wormhole attacks
Disclosure Disclosure of critical information (data in nodes, routing data)
It should be noted that the current insufficient
knowledge of the nature of security hinders the
research community from finding rigorous and
objective solutions to the component metrics
contributing to the overall security.
3 PROPOSED APPROACH
In this section we present the principle of a security
level estimation mechanism that can be used in a
ubiquitous and mobile computing environment that
is based on the Semantic Web. Since mobile ad hoc
networks have great potential as a technology for
ubiquitous environments, we use them as an
example. However, the same approach can be used
in other connectivity platforms of the Semantic Web
– only the required component security metrics vary
depending on the platform technology.
In our approach, the estimation of security level
is based on information gathered by agents in the
Semantic Web and maintained in trusted databases.
This information can be accessed by trusted
measurement agents in the network’s nodes.
The approach is self-organized with one
exception: a hierarchy of trusted voting and
countermeasure entities is required. If individual
trusted nodes volunteer for these roles, the approach
is self-organized. The objectives for the mechanism
include:
Local monitoring in each node,
Utilization of statistical knowledge of the
security level,
Measurements are independent of the routing
mechanism, and
ICEIS 2005 - SOFTWARE AGENTS AND INTERNET COMPUTING
258
Decision mechanism to revocate the trust of
suspicious nodes based on the observations of
more than one node.
Clearly, there are two separate goals in the
estimation process: estimation of the security level
of a node and estimation of the security level of the
network (or part of the network).
3.1 Information Gathering
Table 2: Some component metrics areas in MANETs
Component
Sub-component Heuristic claim
Initial trust The better the assumed initial trustworthiness corresponds to actual
trustworthiness, the more secure the system.
Trust and
key
management
Operational trust The better the assumed operational trustworthiness corresponds to actual
trustworthiness, the more secure the system.
Routing
Routing
information
The better the distribution of routing information in the network
corresponds to the best possible distribution, the more secure the system.
Identity
information
The better the distribution of mobile entity identity information
corresponds to the best possible distribution, the more secure the system.
Mobility
Packet forwarding
information
The better the distribution of packet forwarding information corresponds to
the best possible distribution, the more secure the system.
Usability The more usable the system is, the more secure it is.
Performance The better the system performs, the more secure it is.
Security awareness The more security-aware users are, the more secure the system.
Social engineering The more resistant the system is to social engineering, the more secure it is.
Human factors
Freedom of use The more freedom is offered, the more vulnerable the system is.
Cryptographic
algorithms
Cryptographic
strength
The better the cryptographic strength of the used cryptosystems, the more
secure the network.
Listening The harder it is for a listener to demodulate and decode the radio signal
sent in the wireless environment, the higher the security level.
Wireless-ness
Interference The harder it is for an attacker to cause interference to the radio signals
sent in a wireless network, the higher the security level in that network.
Scale of size The bigger the network, the more vulnerable it is. Scale
Scale of use The more popular the network, the more vulnerable it is.
HW tamper resistance The more tamper-resistant HW is used in a node, the more secure the
network.
SW tamper resistance The more tamper-resistant SW is used in a node, the more secure the
network.
Physical
protection
Location
of node
The more the physical environment is protected from attackers, the more
secure the network.
Functionality The more functional the system is, the more secure it is.
Reliability The more reliable the system is, the more secure it is.
Usability The more usable the system is, the more secure it is.
Efficiency The more efficient the system is, the more secure it is.
Maintainability The more maintainable the system is, the more secure it is.
Product quality
Portability The more portable the system is, the more secure it is.
Privacy
Legislation
Commercial
Cultural
Other factors
Force majeure
scenarios
The information needed for security level estimation
can be stored in databases. The purpose of them is to
offer correct and up-to-date component metrics and
reference information contributing to the security
level. The following kinds of metrics and security
level reference information are useful for estimating
the security level of MANETs (compare to Table 2):
ESTIMATION OF THE SECURITY LEVEL IN A MOBILE AND UBIQUITOUS ENVIRONMENT BASED ON THE
SEMANTIC WEB
259
Trust management: digital credential
information can be obtained using credential
management techniques, see e.g. (Winslett et al.,
2002);
Routing: traffic information and information on
recent attacks can be gathered in trusted traffic
control databases; maintained by devoted
Semantic Web agents;
Mobility: up-to-date mobility information on
nodes can also be maintained in trusted traffic
control databases;
Human factors: statistical databases of human
factors can be used to depict research results of
typical human behavior in different kinds of
applications, and take cultural and group-
dependent factors into account;
Cryptography: databases with information on
the cryptographic strength of different kinds of
cryptosystems can be maintained by devoted
agents;
Wireless-ness: up-to-date research information
on the security of wireless devices can be
maintained in trusted wireless security research
databases;
Scale: a network can assign an agent to keep
track of the size of the network, and the
popularity of network types can be tracked in
traffic control databases;
Physical protection: manufacturers can be
released physical protection certificates of their
products as part of the digital product quality
certificates that can be digital credential
information;
Product quality: the devices used in the
network can be certified with the level of
product quality information attached to the
digital certificates. Certificates can be obtained
from trusted certification pages.
Privacy: A trade-off database can be used to
estimate the effects of privacy requirements on
the security level; and
Legislative, commercial, and force majeure
issues have their own databases.
3.2 Trust Establishment and
Management
The most critical part of the security level estimation
is the trust establishment and management between
the database maintainers and their users. The agents
that are gathering information into the databases also
need to establish their own trusted connections. It is
important to note that trust management is not static
– access rights can be delegated and revocated
dynamically.
Distributed trust models (Blaze et al., 1996)
developed for the Semantic Web can be used in
establishing the trust between different agents
residing in different nodes in the network or in
another network. Examples of trust management
systems assuming a priori knowledge of authority
include PolicyMaker (Blaze et al., 1996), KeyNote
(Blaze et al., 1999), SPKI/SDSI (Simple Public Key
Infrastucture / Simple Distributed Security
Infrastructure) (Ellison et al., 1999), and Delegation
Logic (Li et al., 2003). (Winslett et al., 2002)
introduce TrustBuilder, which supports automated
trust negotiation between strangers on the Web.
(Kagal et al., 2003) propose a policy-based
framework for pervasive computing environments
that extends SPKI and role-based access control. In
the latter approach trust distribution depends on,
e.g., domain, delegation chain and policies.
In the MANET research community, network-
level (rather than application-level) trust distribution
mechanisms have also been proposed. (Zhou &
Haas, 1999) introduce the idea of distributing a CA
(Certification Authority) throughout the network, in
a threshold fashion, at the time of network
formation. In their threshold cryptography-based
approach, the duties of CA (issuing, revoking, and
storing of certificates) are distributed among the
nodes. More recent proposals include (Čapkun et al.,
2003) and (Luo et al., 2002). More state-of-the-art
references can be found in (Hubaux et al., 2001).
Suitable ontologies, i.e. taxonomies with a set of
inference rules, see (Chandrasekaran et al., 1999) for
more information, for information gathering of
different classes of metrics and reference levels are
needed. Using these ontologies in connection with
trust establishment and management, automatic
updating of the trusted security level databases is
possible.
3.3 Key Elements in the Estimation
Process
In our estimation approach the key elements of the
architecture are a Measurement Agent (MA)
attached to each node of a MANET, and a Voting
Agent (VA). A Countermeasure Agent (CMA) is
also used for the Intrusion Detection functionality.
The estimation is carried out in a mobile ad hoc
network by co-operation between MAs and VAs.
Each MA in the network maintains a private metrics
repository with the following information for each
metric:
Metric objects: a collection of measurable
objects to be measured, e.g. a property in routing
information messages;
ICEIS 2005 - SOFTWARE AGENTS AND INTERNET COMPUTING
260
Metric methods: methods associated with the
metrics; and
Metric measuring rod: a database associated
with the metrics that consists of reference
information classified according to the level of
security. The measuring rod database can include
security level data that is either generally known
or gathered from statistical data. The
classification in the reference information may
be based on quantitative or qualitative (using
thresholds) reasoning.
The component metrics areas discussed earlier
can form the basic high-level structure for the
private metrics repository of a MANET node.
In addition to the metrics repository, an MA
maintains a private reputation repository of the
network elements of a MANET or the elements that
are visible to that particular node. The repository
contains critical reputation information as an input to
the estimation process.
A Voting Agent (VA) contains the same
functionality as MA. In addition, it has an organizer
role in case several MAs are going to make
decisions concerning the security level and
trustworthiness of a node; certain trusted nodes can
have VAs in an ad hoc network. A Countermeasure
Agent (CMA) acts on the results obtained from the
voting process. Certain trusted nodes can have
CMAs.
3.4 Estimation and Voting
The basic node-level estimation process is carried
out continuously by the node’s MA. The MA uses
the data stored in its metrics and reputation
repository to estimate the current level of security
from its own node point of view. Moreover, the VA
updates the MAs with information messages
containing critical information on the changes in
nodes and communication in the network vicinity.
The critical information is updated in the reputation
repositories of the MAs to support their estimation
of the security level in the network.
An MA can access suitable databases, depending
on the semantic guidelines it needs to estimate the
security level. At node level, MAs support the
decision processes of the nodes that use the security
level information as an input. For example, the
trustworthiness of a service may be assessed using
the security level monitoring carried out in an MA.
There are a lot of situations where democratic
voting can be used to support decisions to be made
about the security level. For instance, if an MA
detects a node with suspicious activity in the
vicinity, voting can be used to justify the
countermeasures to be carried out by a CMA. An
MA can also inform a VA about its own security
level estimates of an object. A voting process can be
used to compare other MAs’ observations of the
same object.
3.5 Challenges
Mobile ad hoc networks are intrinsically resource-
constrained, which makes our approach difficult to
implement using the current technology. However,
as the required level of security is often higher in
cases where there are better memory and
computation resources in use, the introduced
approach is possible.
The selection of Voting Agents and
Countermeasure Agents is also a problem in cases
where complete self-organization of the network is a
goal. Suitable trust establishment procedures are
needed to select these trusted entities from a group
of nodes. Trust distribution mechanisms between the
database services and its users need to be addressed
as well.
Suitable ontologies for information gathering
from different classes of component security metrics
are needed. This is a challenging task and requires a
rigorous analysis of the metrics to be used. In
addition, information gathering ontologies for the
purposes of estimation algorithms in Measurement
Agents are needed.
As a long-time goal, general-level statistical
knowledge has to be collected on: security
algorithms, network products, user behavior,
applications, experiences from virus and worm
attacks, etc. – about all critical issues contributing to
the overall level of security.
4 CONCLUSIONS AND FUTURE
WORK
The emerging Semantic Web offers powerful tools
for carrying out self-organized estimation of the
security level in mobile and ubiquitous networks and
their nodes. Semantically relevant security level
information can be gathered and maintained in
databases. Information gathering agents can gather
information on, e.g., the traffic situation in the
network, digital credentials, statistical knowledge of
critical components of security, and research results
that affect the level of security. Measurement Agents
located in the network nodes can use the databases
to estimate the security level from their point of
view. Moreover, network-level security is increased
due to the democratic voting mechanism of
ESTIMATION OF THE SECURITY LEVEL IN A MOBILE AND UBIQUITOUS ENVIRONMENT BASED ON THE
SEMANTIC WEB
261
independent measurement entities, each
independently aiming at a higher security level in
the network.
If we are able to develop intelligent and feasible
ontologies for the information gathering, we might
even learn more about the nature of security. In
today’s information technology world there is a lot
of knowledge that just has to be combined in a
suitable way to assess the overall security level, i.e.
“find the forest from the trees.” The current limited
knowledge of the nature of security is hindering us
from finding rigorous solutions to the aspects of
overall security.
Our future work will include further exploration
of component metric areas for mobile ad hoc
networks and development of ontologies for
information gathering and estimation processes. Our
initial framework of security metrics will certainly
be updated during the course of the research – we do
not know a priori the compositional hierarchy of
causalities in such a concept as security. Our future
work will also include building an experimentation
ubiquitous environment for analyzing the
measurement method presented in this paper. It will
be also possible to investigate trust establishment in
this environment. Moreover, techniques for reducing
the memory and computation resource needs of the
approach are to be investigated.
REFERENCES
Berners-Lee, T., Hendler, J., and Lassila, O., 2001. The
Semantic Web. In Scientific American, 284(5): 34-43.
Blaze, M., Feigenbaum, J., Ioannidis, J., and Keromytis,
A. D., 1999. The KeyNote Trust Management System,
V 2. IETF RFC 2704, Available at: www.ietf.org
Blaze, M., Feigenbaum, J., and Lacey, J., 1996.
Decentralized Trust Management. In Proceedings of
IEEE Symposium on Security and Privacy, 164-173.
Čapkun, S., Buttyán, L. and Hubaux, J-P., 20 03. Self-
Organized Public-Key Management for Mobile Ad
Hoc Networks. In IEEE Transactions on Mobile
Computing, Vol. 2, No. 1, 52-64.
Chandrasekaran, B., Josephson, J.R., and Benjamins, V.
R., 1999. What Are Ontologies, and Why Do We
Need Them? In IEEE Intelligent Systems, Jan/Feb.,
20-26.
Ellison, C., Frantz, B., Lampson, B., Rivest, R., Thomas,
B., and Ylönen, T., 1999. SPKI Certificate Theory.
IETF RFC 2693, Sep. Available at: www.ietf.org
Hendler, J., Berners-Lee, T., and Miller, Er., 2002.
Integrating Applications on the Semantic Web. In
Journal of the Institute of Electrical Engineers of
Japan, Vol 122(10), October, p. 676-680.
Henning, R. (ed.), 2001. Workshop on Information
Security Scoring and Ranking. Information System
Security Attribute Quantification or Ordering.
Hubaux, J.-P., Buttyán, L., and Capkun, S., 2001. The
Quest for Security in Mobile Ad Hoc Networks. In
Proceedings of the 2
nd
ACM International Symposium
of Mobile Ad Hoc Networking and Computing
(MobiHoc), 146-155.
Internet Engineering Task Force (IETF), 2004. MANET
Working Group. Available at: www.ietf.org
ISO/IEC 21827. 2002. Information Technology – Systems
Security Engineering – Capability Maturity Model
(SSE-CMM).
Jelen, G., 2000. SSE-CMM Security Metrics. In NIST and
CSSPAB Workshop, Washington, D.C..
Jonsson, E., 2003. Dependability and Security Modelling
and Metrics, Lecture Slides, Chalmers University of
Technology, Sweden.
Kagal, L., Finin, T., and Joshi, A., 2003. A Policy
Language for a Pervasive Computing Environment. In
Proceedings of the 4
th
Int. Workshop on Policies for
Distributed Systems and Networks (POLICY’03).
Li, N., Grosof, B. N., and Feigenbaum, J., 2003.
Delegation Logic: A Logic-based Approach to
Distribution Authorization. In ACM Transactions on
Information and System Security, Vol. 6., No. 1., Feb.,
128-171.
Luo, H., Zerfos, P., Kong, J., and Zhang, L., 2002. Self-
Securing Ad Hoc Wireless Networks. In Proceedings
of the 7
th
Int. Symposium on Computers and
Communications (ISCC), 567-574.
Savola, R., 2004. Estimation of the Security Level in
Wireless E-Commerce Environment based on Ad Hoc
Networks. In Proceedings of the 5
th
European
Conference E-COMM-LINE, Bucharest, Romania,
Oct. 21-22.
U.S. Department of Defense, 1985. Trusted Computer
System Evaluation Criteria (TCSEC) “Orange Book”,
U. S. Department of Defense Standard, DoD 5200.28-
std.
Winslett, M., Yu, T., Seamons, K. E., Hess, A., Jacobson,
J., Jarvis, R., Smith, B., and Yu, L., 2002. Negotiating
Trust on the Web. In IEEE Internet Computing,
Nov/Dec, 30-37.
Yang, H., Luo, H., Ye, F., Lu, S., and Zhang, L., 2004.
Security in Mobile Ad Hoc Networks: Challenges and
Solutions. In IEEE Wireless Communications, Vol. 11,
No.1, Feb., 38-47.
Zhou, L., and Haas, Z. J., 1999. Securing Ad Hoc
Networks. In IEEE Network Magazine, Vol. 13, No. 6,
Nov/Dec, 24-30.
ICEIS 2005 - SOFTWARE AGENTS AND INTERNET COMPUTING
262