On the Suitability of Compressive Sampling for LoRa Signals Classification

Abstract

Sensor networks require even more reliable and power saving communication systems, especially as regards to Internet of Things paradigm. In this scenario, LoRa technology, based on an efficient Chirp Spread Spectrum modulation, is well tailored, granting both long range distance and low energy consumption in the communication link. Even though it is a recent solution, several instrumentation manufacturers are providing test facilities and setups to carry out reliable measurements on LoRa signals in both laboratory experiments and on-field applications. Unfortunately, these solutions are usually given as upgrade of expensive instruments operating in time and/or frequency domain. To overcome this drawback, the authors propose hereinafter an innovative approach based on compressive sampling to suitably reduce hardware requirements and dependably reconstruct the signal of interest from a reduced number of samples acquired in time domain. To assess its reliability and the potential attractiveness, the proposed methodology has been exploited to extract characteristic parameters from actual LoRa signals by mean of an experimental set-up, which takes care of their generation and acquisition. The obtained results highlight the promising performance of the proposed approach.