Comparative Analysis of Reconstruction Algorithm on Compressive Sensing for Tone Using Orthogonal Matching Pursuit

This paper investigates the use of advanced technology in enhancing information exchange through data compression. Data compression reduces the size of data to be transmitted, which accelerates the exchange process. Audio compression, in particular, has become a widely adopted solution to address storage limitations. A novel approach to compression is Compressive Sensing (CS), which combines signal acquisition and compression into a single process. In this study, Compressive Sensing is applied to audio tones using the Orthogonal Matching Pursuit (OMP) reconstruction algorithm. The system is designed with two main processes: compressing and reconstructing the audio signal. The compression process involves transforming the audio signal from the time domain to the frequency domain using the Discrete Cosine Transform (DCT). In the reconstruction phase, the compressed signal is reverted to its original form using the OMP algorithm. For tone recognition, feature extraction is performed using the Fast Fourier Transform (FFT), and classification is achieved with the Euclidean Distance Matrix (EDM). The results demonstrate that compressing tones using CS and reconstructing them with the OMP algorithm—specifically with a compression ratio of 12.5, a segment size of 1024, and thresholds of 0.1 and 50—achieves an average accuracy rate above ${9 0 \%}$. However, lower compression ratios and segment sizes, along with higher thresholds, lead to decreased accuracy rates.