Real-World Applications - Immunological Computation: Theory and Applications - page 187

Figure 7.10 The performance of MILD when tested with “full-tail failure” data,
where this type of fault is manifested in pitch error rate (starting at the 1200th
time step). The graph also shows the number of detectors activated (lower-bar
chart) as signifi cant deviations in data patterns appear. The bar chart shows the
arrangement of the detectors with increased radius.
h is helped in classifying the residuals into distinct patterns, denoting diff erent
faulty situations.
An algorithm was proposed by Pinto et al. (2005) for detecting faults in telephone
systems based on DT in immunology, which is guided by the principle that the pres-
ence or absence of secondary signals determines responsiveness or tolerance. Each
call in this fault-detection system is represented by an antigen composed of linear
attributes: origin, destination, duration of calls, and a nominal attribute. Two signal
levels were identifi ed: signal 1 for perceiving the presence of the antigen and signal 2
for costimulation by using the noncompleted call rate. Signal 2 was responsible for
alarming a danger situation. Detector death, detector deactivation, detector popula-
tion renewal, and a voting routine were signifi cantly employed in this work.
Guzella et al. (2007) presented an immune-inspired approach for fault detec-
tion called dynamic eff ector regulatory algorithm (DERA). h e proposed approach
integrates the role of regulatory T cells in control and signaling between cells. In
DERA, new components of the immune system such as cytokines and regulatory
cells are incorporated in the model. DERA uses a population of regulatory and