Collective Classification for Spam Filtering - Computational Intelligence in Security for Information Systems - page 35

recent paper [9] proposed an algorithm for hierarchical aggregation in VANETs
based on a probabilistic approximation to data. However, in our proposal the
probabilistic approach is applied on the verification of fuzzily aggregated data.
Various proposals have been made to promote cooperation and enhance se-
curity in VANETs through the use of authentication, key management and
pseudonyms [10]. However, to the best of our knowledge no tools that allow
ensuring that the generated information is true have been proposed till now.
A quite logical initial approach to address this problem is by providing nodes
of a mechanism to verify the packet content accuracy. In this paper nodes
will be called agents from now on, according to the Artificial Intelligence
literature.
When data aggregation is possible because the vehicle has received several
packets informing about the same event, various known schemes based on clus-
ters may be used [7] [12]. In those proposals, the so-called cluster-heads are
responsible for adding the information submitted by their cluster members, and
also for forwarding it. Consequently, those schemes can be seen as proposals
based on a hierarchy, which increases the amount of necessary communications
for cluster management, especially in highly dynamic networks such as VANETs.
In this paper we propose a new scheme that replaces such a static hierarchical
point of view by a fuzzy-logic based dynamic approach.
3 Fuzzy Logic Approach
Aggregation schemes are usually based on three sequential phases. The first
phase is decision-making, when the system must allow deciding whether two
pieces of information are similar enough to add them or not. The second phase
is data aggregation. Once the system has concluded that the received packets
must be added, it has to apply an aggregation method for combining them. The
third and last step is the aggregated data delivery. After the two previous steps,
the system must allow the dissemination of the aggregated data.
In the literature we can find several proposals for the two last stages. However,
schemes for making decision are normally based on predefined clusters, trees
or fixed structures, which are not suitable for their use in VANETs. Thus, a
new aggregation scheme is here proposed where decisions regarding whether two
received pieces of information are similar enough to be added are based on the
content of such packets. Such decisions take into account possible inaccuracies or
approximations regarding space and time data to define an event. Therefore, our
proposal for the application of fuzzy logic is focused especially on the description
of the decision criteria of the first phase of the aggregation scheme.
In the proposed scheme information is aggregated based mainly on two dimen-
sions: space and time. On the one hand, the approximate location of an event in a
road or a map is a fundamental parameter. On the other hand, a time interval in
which the persistence of an announced event must be taken into account. When
applying a fuzzy approach on such parameters the best decision on aggregation
is made because it allows a flexible reasoning that takes into account all possible
values in packets announcing the same event. This enables a dynamic approach