11-10-2012, 11:23 AM
A Novel Algorithm for Mining Association Rules in Wireless Ad Hoc Sensor Networks
A Novel Algorithm.ppt (Size: 809 KB / Downloads: 28)
Abstract
Comprehensive framework for mining AR in WASNs
Main goal
Generate rules that improve the WASN’s Quality of Service
Introduction
Introduce a data mining solution to extract behavioral patterns from WASNs
Main objective of sensor AR
Capture the temporal relations between sensor nodes base on common intervals of activities.
Mining AR in WASNs
The sensor AR mining problem
Data extraction methodologies
Direct reporting
Distributed extraction
Positional lexicographic tree
Mining AR in WASNs
Positional lexicographic tree
Used to store the behavioral data extracted from WSANs and generate the frequent patterns
Definition
Rank(s)
Maps each sensor node to a unique integer number
Pos(j) = Rank(j) – Rank(i)
Maps each sensor node in the lexicographic tree
Path(S)
List of all possible paths from root to any other nodes
V(P) = {pos(s1) , pos(s2) , … , pos(sk)}
Position vector of the pattern P
Conclusion
Improve the Quality of Service of the WASNs by identifying the correlated sensor
Received validity message from 50 to 90% by compared with direct reporting.
PLT outperforms FP-Growth in term of CPU time and memory usage by 30 to 50%