A Subspace Approach to Sparse-Sampling-Based Multi-Attribute Data Aggregation in IoT

Abstract

The emergence of the heterogeneous Internet of Things (IoT) has realized the demand for multi-attribute data collection in response to the increasing demand for information in diverse applications. Sparse sampling has been used to reduce network energy consumption in order to extend the life of energy-constrained networks. Real-time multi-attribute data aggregation under the sparse sensing framework has become a research focus. Therefore, we proposed a sparse-sampling-based IoT data aggregation approach to reduce network energy consumption and enable real-time multi-attribute reconstruction. For data collection, a sparse sampling data collection approach is proposed that can successfully collect and transmit the multi-attribute data to the sink even while certain IoT sensor nodes are in the sleep mode. For data reconstruction, a real-time multi-attribute data reconstruction method based on subspace is proposed. The proposed method arranges multi-attribute data in a tensor form in order to further utilize the correlation of multi-attribute data. Subspaces representing the spatial distributions of the multi-attribute data can be obtained from the previously reconstructed data. Incorporating total variation constraint, the proposed method reconstructs the current time slot multi-attribute data with high precision in real time. The experimental results demonstrate the effectiveness of the proposed method in real-time multi-attribute data reconstruction.