- Inherent freedom due to lack of central authority in selforganized mobile ad hoc networks introduces challenges to security and trust management. Arguably, trust management is the most critical security issue in mobile ad hoc networks. If nodes do not have any prior knowledge of each other, the trust establishment becomes complicated. In this kind of situations, the nodes themselves should be responsible of their own security. We propose a model for security management in self-organizing mobile ad hoc networks that is based on nodes' own responsibility of their security and node-level security monitoring.
Dynamical communication is essential when attempting to
monitor security critical situations dangerous conditions as
early as possible. Mobile ad hoc networks (MANETs) [
Among all the research issues, information security in MANETs is particularly challenging due to the highly dynamic network topology, the lack of central authority, the shared wireless medium, and memory and performance resource constraints. Despite the advances in the field, the research is lacking proposals to measure the overall security level of mobile ad hoc networks.
The main contributions of this work are in the analysis of basis for identification of suitable high-level information security metrics for mobile ad hoc networks and in the introduction of an on-the-fly security level estimation mechanism for MANETs to support both node-level and network-level decisions.
The nature of ad hoc networks makes them vulnerable to a number of attacks, such as denial-of-service, interference, impersonation, eavesdropping, information leakage and data tampering. In the following, we review the security goals and security environment of MANETs, discuss the scope of security, and provide an overview of security metrics.
The ultimate goal of the security solutions for mobile ad hoc networks is to provide services for the desired security needs, mainly confidentiality, integrity, availability, authentication and non-repudiation, at the desired security level. In general, the research has noted that traditional security solutions, such as public key infrastructures, or authentication mechanisms, are potential also for ad hoc networks, but in many cases they are not sufficient by themselves.
Technical security metrics can be used to describe, and hence
compare, technical objects. This includes algorithms,
specifications, architectures and alternative designs, products,
and as-implemented systems at different stages of the system
lifecycle. Design vulnerabilities can result from an insecure
design, whereas implementation vulnerabilities are connected
to poor implementation of a product. Thus the former term
typically refers to lower technology maturity. Security
metrics model consists of three components: the object being
measured, the security objectives (i.e. the “measuring rod”
the object is being measured against), and the method of
measurement. The security objectives typically consist of
security requirements, such as specifications or standards,
e.g. Common Criteria (CC) [
III. BASIS OF SECURITY METRICS FOR MANETS
A compositional approach can be used to define security
metrics for MANETs, with the following, possibly iterative,
steps [
Critical control information distribution in a mobile ad hoc network means the location of the critical control information in that network with respect to time. The following main types of critical control information distribution can be identified: o Trust information (e.g. keys, certicates, signatures), o Routing information, o Mobile entity identity information, and o Packet forwarding information.
Security metrics can be developed from heuristics that compare the actual critical control information distribution to a posteriori knowledge of the most secure distribution of such information. The most secure distribution is the distribution the implements best the security goals. It is possible to develop predictive methods based on a posteriori knowledge.
In mobile ad hoc networks the cryptographic strength has
tight cross-relationships with trust management, and often
with other critical information management. There are
various ways of describing cryptographic algorithm metrics,
e.g. [
Generally, the most unexplored and the most critical field in security is the human user behaviour. Human factors have an enormous impact on the global security level of mobile ad hoc networks too.
An important consideration from the human user point of
view is user acceptance, or, from a reverse perspective, user
resistance to the systems with which they must interact. The
user resistance manifests itself in various ways, including
improper use of the security mechanisms [
When the technological solutions for routing. mobility and trust management become more mature, the effect of product quality will have more emphasis on the overall security level of mobile ad hoc networks. With mature technology we can investigate implementation vulnerabilities. Product quality metrics can be seen as a general framework for measuring mature solutions.
In this section we propose an on-the-fly security level estimation mechanism for a networked monitoring system based on mobile ad hoc networks. The approach is selforganized with one exception: a hierarchy of trusted voting and countermeasure entities is required. If individual trusted nodes volunteer for these roles, the approach is selforganized. The objectives for the mechanism include the following: o No central database can be used, o Local monitoring in each node, o Statistical knowledge of the security level is utilized, o Measurement should be independent of the routing mechanism, and o 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: o Estimation of the security level of a node, and o Estimation of the security level of the network.
In our estimation approach the key elements of the architecture are a Measurement Entity (ME) attached to each node, and a Voting Entity (VE). A Countermeasure Entity (CME) is also used for the Intrusion Detection functionality. The estimation is carried out in a mobile ad hoc network by co-operation between MEs and VEs.
A Voting Entity (VE) contains the same functionality as ME. In addition, it has an organizer role in case that that several MEs are going to make decisions concerning the security level and trustworthiness of a node. In an ad hoc network, certain trusted nodes can act as VEs.
A Countermeasure Entity (CME) acts on the results obtained from the voting process. Certain trusted nodes can act as CMEs.
Because critical information is distributed among MEs, VEs, and CMEs, a trust establishment and distribution mechanism is needed to enable the estimation and voting processes.
The basic node-level estimation process is carried out continuously by the ME of the node. The ME uses the data stored in its metrics and reputation repository to estimate the current level of security from its own node point of view.
The critical information is updated in the reputation repositories of the MEs to support their estimation of the security level in the network. A VE can obtain update information from other VEs located in different parts of the network. General-level security updates to the MEs’ metrics repositories can also be delivered using the VEs as a communication link.
At node level, MEs support the decision processes of the node, that use the security level information as an input. For example, the trustworthiness of a service may be assessed using the security level monitoring of ME.
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 ME detects a node with suspicious activity in the vicinity, voting can be used to justify countermeasures.
An ME can also inform a VE about its own security level estimates of an object. A voting process can be used to compare other MEs’ observations of the same object.
Mobile ad hoc networks are intrinsically resourceconstrained, 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 entities and countermeasure entities is also a problem in cases where complete selforganization of the network is a goal. Suitable trust establishment procedures are needed to select these trusted entities from a group of nodes. Trust management is also needed to enable the communication between the VEs, MEs and CMEs.
Suitable estimation algorithms should be developed for the metrics framework. This is a challenging task and requires a rigorous analysis of the metrics to be used.
As a long-time goal, general-level statistical knowledge has to be collected on: security algorithms, network products, user behaviour, applications, experiences from virus and worm attacks, etc. – about all critical issues contributing to the overall level of security.
The security level estimation mechanism presented here is
closely related to the Intrusion Detection System (IDS)
approaches proposed for mobile ad hoc networks. Mishra et
al. [
We have discussed the problem of self-measuring the information security level in a mobile ad hoc network used as a dynamical communication basis for critical monitoring applications. Solving this problem clearly requires a multidisciplinary effort. The current limited knowledge of the nature of security is hindering us from finding a rigorous solution. In this paper we have identified some major basis components that contribute to the security level of MANETs. The optimum management of critical control information distribution in time and placement in a mobile ad hoc network is seen as the major technical goal. Critical information includes, e.g., keys and certificates, routing information, identity information and packet forwarding control information.
Moreover, we have presented a security level estimation mechanism that can be used in a networked monitoring system, where a node has a lot of responsibility of itself and its neighbours.
Our future work will include further exploration of component metric areas and identification of the dependencies between them.