Compressive sampling–based data loss recovery for wireless sensor networks used in civil structural health monitoring

Abstract

In a wireless sensor network, data loss often occurs during the data transmission between the wireless sensor nodes and the base station. In the wireless sensor network applications for civil structural health monitoring, the errors caused by data loss inevitably affect the data analysis of the structure and subsequent decision making. This article explores a novel application of compressive sampling to recover the lost data in a wireless sensor network used in structural health monitoring. The main idea in this approach is to first perform a linear projection of the transmitted data x onto y by a random matrix and subsequently to transmit the data y to the base station. The original data x are then reconstructed on the base station from the data y using the compressive sampling method. The acceleration time series collected by the field test on the Jinzhou West Bridge and the Structural Health Monitoring System on the National Aquatics Center in Beijing are employed to validate the accuracy of the proposed data loss recovery approach. The results indicate that good recovery accuracy can be obtained if the original data have a sparse characteristic in some orthonormal basis, whereas the recovery accuracy is degraded when the original data are not sparse in the orthonormal basis.