

Cochlea-Based Spectral Decomposition of Sensor Signals for Resource-Constrained Sensor Networks
Abstract
The biological nervous system represents a highly efficient method of sensing and transmitting information from the external world through a condensed format. Additionally, the system is capable of aggregating information from various sources and actuating based on the perceived external stimulus. This entire process is performed instantaneously, allowing for real time sensing and actuation. Of particular interest in this study is the succinct sensing methods exhibited by the human ear, which can comprehend a wide variety of sound frequencies, as well as a large range of decibels. To perceive sound, the ear uses a novel method of place theory and frequency decomposition. This study will focus on mimicking the techniques used in the ear for applications of signal decomposition in sensor networks. The benefit of this method can be a highly compressive form of data that requires less bandwidth and energy in resource constrained sensor networks. The accuracy of the proposed signal decomposition technique was validated using simulated sensor signals. Additionally, a prototype signal decomposition device was developed to experimentally validate the decomposition technique.