Frequency domain measurement node based on compressive sampling for sensors networks

The paper deals with the problem of designing and implementing a measurement node based on compressive sampling (CS). The considered node is tailored for wide area sensors networks aimed to carry out measurements in frequency domain. To this aim, the node takes advantage from some known or recently proposed CS features in such a way as to outperform the nominal specification of its data acquisition module. To make the spectrum estimation feasible on the node level, a suitable strategy for input signal random sampling and an efficient CS implementation, i.e. the greedy algorithms based on the so-called match-pursuit approach, are exploited. First tests are presented, related to a cost-effective microcontroller from STMicrocontroller, namely STM32F429ZI, characterized by a data memory depth sufficient to execute the agile computational scheme of the greedy algorithm. The estimated spectra concur with those obtained through standard discrete Fourier transform-based approaches, thus highlighting the feasibility of the proposed measurement node.