Databases Reference
In-Depth Information
FIGURE 9.4: A process workflow for the DHGN distributed event detection
in a WSN. (This figure is a copyright of and reproduced with permission of
Civil-Comp Ltd. Previously published in [96].)
9.1.4.1
Pattern Matching at Sensory Level
The occurrences of abnormal events are detected using a pattern matching
approach. Sensory readings are considered to be patterns, and any significant
changes in the structure of normal patterns are classified as events or critical
events that must be reported back to the sink (or other master node). The
use of a clustered DHGN configuration maps each sensor node with a DHGN
subnet that is able to accept a number of different sensory readings as a single
subpattern. The following algorithm describes our proposed pattern matching
approach for event detection at the sensor level.
In this algorithm, the output of the pattern matching process is a signal,
which alerts the SI module of the detection of a new event. The base sta-
tion will respond by performing a spatio-temporal analysis on the readings
obtained.
9.1.5 Performance Metrics: Memory Utilization
Memory utilization estimation for the DHGN algorithm involves an analysis
of the bias array capacity for all of the GNs in the distributed architecture
