SC-EEDC: Similarity Based Clustering for Energy Efficient Data Collection in WSN

Wireless sensor networks (WSN) are widely used in various situations. Energy saving is one of the most important issues because of the limited power. Communication is the mainly energy consuming part of sensor nodes. Reducing the size of transmitted data can conserve the energy of nodes. Spatial and temporal correlation is ubiquitous in wireless sensor networks. By exploiting spatial and temporal correlation, only a subset of data need to be transmitted and the rest of the data can be estimated. Data aware clustering is an effective way to exploit spatial correlation among sensor nodes. In our SC-EEDC framework, fuzzy ART artificial neural network is used to measure the shape similarity among data sequences and the magnitude similarity is estimated by a magnitude similarity model. And two corresponding estimation algorithm are proposed. We propose Weighting based K-means clustering algorithm (WK-means), which considers multiple clustering factors in addition to the data similarity. The K-means algorithm structure is introduced in clustering algorithm to search the more energy-saving clustering topology. Anchor node based data collection strategy is proposed to measure the spatial correlation in real time and suppress the transmission of spatial redundant data at source node. Sleeping scheduling is introduced to dynamically adjust the spatial sampling rate and the temporal redundancy is reduced using Length Encoding. The cluster maintenance scheme keeps the cluster structure’s efficiency over the network lifetime. Simulation results show that our SC-EEDC achieves significant data reduction without affecting the accuracy of collected data, reduces the energy consumption in each round of data collection and effectively prolongs the network lifetime.