Information Technology Reference
In-Depth Information
number. This is achieved through using a population of artificial DCs to per-
form aggregate sampling and data processing. Using multiple DCs means that
multiple
data items in the form of antigen are sampled
multiple times
.Ifasin-
gle DC presents incorrect information, it becomes inconsequential provided that
the majority of DCs derive the correct context. The sampling of data is com-
bined with context information received during the antigen collection process.
Different combinations of input signals result in two different antigen contexts.
Semi-mature antigen context implies antigen data was collected under normal
conditions, whereas a mature antigen context signifies a potentially anomalous
data item. The nature of the response is determined by measuring the number of
DCs that are fully mature, represented by a value,
MCAV
-the
mature context
antigen value
. If the DCA functions as intended, the closer this value is to 1,
the greater the probability that the antigen is anomalous. The MCAV value is
used to assess the degree of anomaly of a given antigen. By applying thresh-
olds at various levels, analysis can be performed to assess the anomaly detection
capabilities of the algorithm.
The DCA has three stages:
initialisation, update and aggregation
. Initialisation
involves setting various parameters and is followed by the update stage. The
update stage can be decomposed into
tissue update
and
cell cycle
.Boththe
tissue update and cell cycle form the
libtissue
tissue server. Signal data is fed
from the data-source to the tissue server through the tissue client.
The tissue update is a continuous process, whereby the values of the tissue
data structures are refreshed. This occurs on an event-driven basis, with values
for signals and antigen updated each time new data appears in the system.
Antigen data enters tissue update in the same, event driven manner. The updated
signals provide the input signals for the population of DCs.
The cell cycle is a discrete process occurring at a user defined rate. In this pa-
per, 1 cell cycle is performed per second. Signal and antigen from the tissue data
structures are accessed by the DCs during the cell cycle. This includes an update
of every DC in the system with new signal values and antigen. The cell cycle and
update of tissue continues until a stopping criteria is reached, usually until all
antigen data is processed. Finally, the aggregation stage is initiated, where all
collected antigen are subsequently analysed and the MCAV per antigen derived.
3.3 Parameters and Structures
The algorithm is described using the following terms.
-
Indices:
i
=0
, ..., I
input signal index;
j
=0
, ..., J
input signal category index;
k
=0
, ..., K
tissue antigen index;
l
=0
, ..., L
DC cycle index;
m
=0
, ..., M
DC index;
n
=0
, ..., N
DC antigen index;
p
=0
, ..., P
DC output signal index.
