An Efficient Approach to Categorize Data Using Improved Dendritic Cell Algorithm with Dempster Belief Theory - Proceedings of All India Seminar on Biomedical Engineering 2012 (AISOBE-2012) - page 161

interpret this information for T-cells of the adaptive immune system, to induce
appropriate immune responses toward perceived threats. Therefore, DCs can be
seen as detectors for different policing sites of the body as well as mediators for
inducing a variety of immune responses. Signals are normally precategorized as
''PAMP'', ''Danger'' or ''Safe'' in the DCA.
The co-occurrence of antigen and high/low signal values forms the basis of
categorization for the antigen data [ 4 ].
The primary components of a DC-based algorithm are as follows:

• Individual DCs with the capability to perform multisignal processing.
• Antigen collection and presentation.

Dendritic Cell Algorithm (DCA) Based on Belief Theory

The basic idea of the Dendritic Cell Algorithm (DCA) based on Belief theory is
to solve the problem of correlation and resolve the problem of unknown and
rapidly evolving harmful attacks [ 5 ]. The basic steps of the algorithm are given
below:

Step1: With the help of Dendritic Cell Algorithm we categorized data, whether
the data are normal or affected with anomaly or in other words, abnormal.
Step2: Dempster-Belief theory is used to compute the probability of evidences
that indicate support to the attack or normal class. The use of Dempster
Belief theory steadily spreads out, mostly because it is used to cope with
large amounts of uncertainties that are inherent of natural environments.
This new approach considers sets of propositions and assigns to each of
them an interval [Belief, Plausibility][ 8 ].
Step3: After the classification we calculate the entropy of the attack treated as
signal. For the calculation of entropy let us consider a set having possible
event Now measured entropy, represented by H(x) which is calculated with
the help of the given formula [ 9 ]:
H ð x Þ¼ X

Bit = message

N

1
pi

P ð xi Þ log ð 1 Þ

i ¼ 1

where p(xi) the probability of event.

Experimental Setup and Result

The experiment's data comes from the KDD CUP competition data set. These data
are part of the data collected from the MIT Lincoln Labs 1998 DARPA Intrusion
Detection Evaluation Program and are considered benchmark data for evaluating
intrusion detection system.