Information Technology Reference
In-Depth Information
In Fig. 3, when a detector matches a state transition string from the FSM, the first
signal will be sent to the controller, and the parity checking result from CED module
will be sent to the controller as the second signal (co-stimulation). According to the first
and second signals, the controller will update the detector set
R
if necessary. In the
updating process of
R
, the state transition string causing false positive will be collected
by the self set
S
as a new self string, and
R
will be filtered with
S
. The results record
module is used for collecting the performance datum of the system.
Fig. 3.
Simulation experiment system. Different from [8, 9], the self set
S
is kept by the system, it
is used for the filtration process of detector set
R
in detector set updating steps.
The length of the bit-string is 16, and then the size of string space
O
is 2
16
. The
“r-continuous-bits” matching rule is adopted here and the initial value of matching
length threshold
r
is fixed to 12.
The detector set updating strategy of ASTA-CED in [8] is temporarily denoted as
strategy 0 in these experiments. For convenience, in all tables about experimental
results, number 0 denotes strategy 0, I denotes strategy I, and II denotes strategy II.
In an independent run, three strategies have identical initial detector set, which is
generated randomly and has
r
=12. The number of self strings
N
S
is fixed at 3000. The
self set of every independent run is generated randomly, and is identical for both the
three strategies. The parameter
a
, which is the proportion of self strings already known
in advance among the complete set of self strings, is set to {0.9, 0.8, …, 0.1} for
observing the change of the results against it.
Every state transition of the whole string space appears once in an independent run,
in which the detector set is updated according to different strategies. The process for
generating a new detector will be performed for only 1 cycle when an old detector is
deleted from current detector set
R
in its updating process.
The results take the average values over 10 independent runs for every value of
a
.
There are two parameters taken to make the comparisons:
N
R
and
△
C
.
N
R
is the number
of mature detectors.
C
is the number of total invalid state transitions covered by the
detector set, which represents the non-self coverage degree. And
△
C
is the variation
