A Novel Framework for Cluster-based Sensor Fusion

Sensor fusion is a significant aspect for wireless sensor networks. Due to random deployment of sensor nodes, clustering and access sequence of sensor nodes will influence energy consumption and time delay of data fusion. In this paper, an energy-time-efficient framework for cluster-based data fusion is proposed for improving the performance and reducing the congestion of wireless link, which adopts maximum entropy clustering to partition wireless sensor network and engages ant colony optimization to arrange the schedule of access sequence for all sensor nodes inside clusters. In the proposed framework, a mobile agent firstly fuses local information progressively within each cluster; then central service node performs a global fusion from local results. The simulation results verify that the optimized method for cluster-based data fusion can improve the global precision. It is analyzed that the clustering can influence the efficiency and precision of data fusion. Finally, the performance of cluster-based data fusion is analyzed with the energy*time metric.