Databases Reference
In-Depth Information
Algorithm 6
Pattern Matching Function at the Sensor Level
1:
given n sensory readings for time t: S
t
= (s
1
, s
2
, ...s
n
)
2:
convert S
t
to a binary signature B
t
. Therefore f(binsig) : S
t
→ B
t
3:
trigger = F ALSE
4:
eventAlert.Sensor = F ALSE
5:
repeat
6:
for
i = 0 to MAXREADINGS
do
7:
{check for matched subpatterns (sensory readings) in sensor data
storage}
8:
if
new.B
t
== s[i].Sensor
then
9:
{new.B
t
: new readings, matching process is conducted using the
DHGN algorithm}
10:
exit FOR
11:
else
12:
s[MAXREADINGS + 1].Sensor = new.B
t
13:
trigger = T RUE
14:
eventAlert.Sensor = T RUE
15:
end if
16:
end for
17:
until
trigger = T RUE
18:
send eventAlert.Sensor and s[MAXREADINGS+1].Sensor to SI mod-
ule function at base station
19:
MAXREADINGS = MAXREADINGS + 1
and the storage capacity of the SI module node. A detailed analysis of the
bias array capacity for the GN-based algorithm was presented in Chapter 3.
Based on that analysis, it can be concluded that the e
cient storage/recall
mechanism of the DHGN offers e
cient memory utilization. Furthermore, it
only uses memory to store newly discovered patterns rather than storing all
pattern inputs. Figure 9.5 compares the estimated memory capacities for a
DHGN processing cluster to the maximum memory size of a typical physical
senor node (Table 9.1) as a function of the subpattern size.
As shown in the Figure 9.5, as the size of subpattern increases, the memory
space requirement increases considerably. Note that the small subpattern sizes
consume less than 1% of the total memory space available. Therefore, the
DHGN implementation is best deployed for small subpattern sizes.
9.1.6 Spatio-Temporal Analysis of Event Data
The spatio-temporal analysis is a process of observing the frequency and
distribution of events within the wireless sensor networks. It is conducted at
the base station with computational grid-enabled infrastructure, since it has
the bird's eye view of the overall network. An analysis on forest fire detection
