The high cost of communication vis-a-vis computation cost necessitates efficient aggregation in wireless sensor networks. This paper proposes and evaluates two different clustering techniques, FAMNN and DBA. The first approach based on adaptive resonance theory whereas the second approach is based on density based clustering of data. The first technique, FAMNN is based on fuzzy ART and fuzzy ARTMAP neural network. Fuzzy ART works in offline phase with stored data while ARTMAP is employed in the cluster heads to segregate outliers or form new clusters whenever new data arrives from the sensors. The second method works on DBA algorithm which is based on density of points within the clusters. It uses two parameters, minimum points for the clusters and distance between the data within clusters. The small number of data clusters eliminates the need to transfer large amounts of data. The testing of both approaches was done on synthetic sensor data with different parameters generated through tool close to real world sensor data. Simulation results indicate that FAMNN is able to identify the natural clusters and map new data to existing clusters or form new clusters to drastically reduce the amount of data required to be sent to the sink. DBA is able to generate the initial clusters but the number of data points that are marked as noise is significant. FAMNN is able to cluster form fresh clusters as per the requirements of sink whereas, DBA does not cluster outliers till their number is high escalating the communication cost.
[1]
Hans-Peter Kriegel,et al.
Multi-step density-based clustering
,
2005,
Knowledge and Information Systems.
[2]
Stephen Grossberg,et al.
Fuzzy ARTMAP: A neural network architecture for incremental supervised learning of analog multidimensional maps
,
1992,
IEEE Trans. Neural Networks.
[3]
Stephen Grossberg,et al.
Fuzzy ART: Fast stable learning and categorization of analog patterns by an adaptive resonance system
,
1991,
Neural Networks.
[4]
Hans-Peter Kriegel,et al.
Density-based clustering of uncertain data
,
2005,
KDD '05.
[5]
Sajal K. Das,et al.
Information-intensive wireless sensor networks: potential and challenges
,
2006,
IEEE Communications Magazine.
[6]
Gianluigi Ferrari,et al.
Optimal Transmit Power in Wireless Sensor Networks
,
2006,
IEEE Transactions on Mobile Computing.
[7]
Hans-Peter Kriegel,et al.
Density-Based Clustering in Spatial Databases: The Algorithm GDBSCAN and Its Applications
,
1998,
Data Mining and Knowledge Discovery.
[8]
Konstantinos Psounis,et al.
Modeling spatially correlated data in sensor networks
,
2006,
TOSN.
[9]
Ian F. Akyildiz,et al.
Wireless sensor networks: a survey
,
2002,
Comput. Networks.
[10]
KriegelHans-Peter,et al.
Density-Based Clustering in Spatial Databases
,
1998
.