Utilizing the cochlea as a bio-inspired compressive sensing technique

Abstract

Structural monitoring for civil infrastructure is a rapidly developing field that has made significant advancements over the last decade. However, a number of performance bottlenecks remain including challenges with cost-effectively scaling monitoring systems up to large nodal counts. Due to the many parallels between biological sensory systems and engineered sensing systems, the biological nervous system can offer potential solutions to the current deficiencies of structural monitoring systems. The nervous system is capable of real-time processing and data transmission of external stimuli through an extremely condensed format with very basic processing units. This study explores the mammalian auditory system for inspiration because it achieves efficient data acquisition processes that outperform existing engineered sensing systems. Specifically, the auditory system realizes this through three steps: (1) real-time decomposition of a convoluted time-based signal into frequency components, (2) information compression for each component, and (3) efficient high-speed data transmission to the auditory cortex. In this paper, these three main mechanisms are explored and a bio-inspired structural monitoring system is proposed. The functionality of the proposed system is compared to traditional data compression techniques (wavelet transforms and compressed sensing) on various vibratory signals. While the wavelet transform is able to outperform the proposed sensor by minimizing signal reconstruction errors, the proposed bio-inspired sensor achieves similar compression rates but, unlike the others, does so using real-time processing.

Compressive sensing-based lost data recovery of fast-moving wireless sensing for structural health monitoring

“…Compressive sensing (CS), which is also called compressive sampling, provides a new sampling theory to reduce data acquisition, namely, that sparse or compressible signals can be exactly reconstructed from highly incomplete random sets of measurements [20][21][22][23]. The potential of CS for the SHM has been investigated, and many applications of CS have been presented [24][25][26][27][28][29][30][31][32][33]. This paper used a CS-based wireless data transmission approach for lost data recovery.…”

Dynamic Characteristics Estimation of Cable-Stayed Bridge Using Artificial Filter Bank (AFB)

“…The closer RE is to zero, the more effective reconstruction becomes. At this point, Lynch [7] introduced a statistical concept for determination of reconstruction error, applying a residual sum of squares (RSS be universally applicable by dividing the difference between original and reconstructed signals by original ones, as suggested in (1). Next, peak-picking algorithm is developed to re-sample peak values based on time information of only peak values picked from reconstruction signal determined by (1), where peak value is determined only when there is a sign change while calculating the slope of each sample signal.…”

“…Dynamic response of a structure using WSNs will contain more data compared to that of the static response, and may result in data loss due to bottleneck during wireless transmission of the dynamic data and also cost increase from integrating and managing large database (DB) accordingly. Therefore, in order to obtain and manage dynamic responses of infrastructure using WSNs in real time with reliability, a technical alternative needs to be proposed to filter valid dynamic responses so as to efficiently obtain and transmit only necessary data [7].…”

Robust data transmission and recovery of images by compressed sensing for structural health diagnosis

“…Motivated and improved upon the traditional transform coding, compressed sensing [32][33][34] directly acquires a random, seemingly incomplete, small subset of relevant compressed measurements, from which the original, sparse, signal is recovered by solving a well-defined optimization program. Applications are spread in vast fields such as channel coding [35], medical imaging [36], computer vision [37], etc, recently also in system identification and damage detection of civil structures [38][39][40][41][42][43].…”