Databases Reference
In-Depth Information
TABLE 9.2: Examples of Simple Temperature Readings and Their Respective
Binary Signature
Temperature Threshold Range (
◦
C)
Binary Signature
0-20
10000
21-40
01000
41-60
00100
61-80
00010
81-100
00001
readings. As an example, to obtain a standardized format for the pattern in-
put from various sensory readings, the use of an adaptive threshold binary sig-
nature scheme for dimensionality reduction and standardization is considered
for multiple sensory data. The binary signature is a compact representation
that is capable of representing different types of data with different values
using a binary format [72]. Table 9.2 shows examples of temperature data
ranges that have been converted into a series of binary signatures.
9.1.4 Event Classification
The DHGN distributed event detection scheme involves a bottom-up classi-
fication technique, in which the classification of events is determined from the
sensory readings obtained through the WSN. The approach pre-processes pat-
terns using dimensionality reduction techniques, such as the adaptive thresh-
old binary signature scheme. These patterns are propagated to all available
DHGN subnets for recognition and classification purposes.
The recognition process involves finding dissimilarities between the input
patterns and previously stored patterns. Any dissimilar patterns will create a
response for further analysis, whereas similar patterns will be recalled. This
research used the supervised single-cycle learning approach in a DHGN to
perform recognition based on the stored patterns. The stored patterns in
our proposed scheme include the set of ordinary events that are translated
into normal surrounding/environmental conditions. These patterns are de-
rived from the results of an analysis conducted at the base station, which is
based on the continuous feedback from the sensor nodes. Figure 9.4 shows the
workflow for the distributed event detection.
The event detection scheme using the DHGN incorporates twos levels of
recognition: front-end recognition and back-end recognition. Front-end recog-
nition involves using the DHGN pattern matching mechanism to determine
if the sensor readings obtained by the sensor nodes indicate an extraordinary
event or a normal surrounding condition. Conversely, the spatial occurrence
detection is conducted through the back-end recognition. In this approach, the
use of signals sent by sensor nodes is considered to be a pattern for detecting
event occurrences in a specific area or location.
