Implementation of Compressive Sensing in Hardware

Abstract

The Shannon-Nyquist theorem enables signal acquisition with sampling frequency greater than or equal to twice the maximum signal bandwidth. Compressed Sensing does sampling and compression at the same time and original signal, sparse in any domain can be reconstructed by fewer number of samples than those required in conventional sampling theory. The compressed sensing relies on two basic processes: Compressive Sampling and signal reconstruction. Compressive sampling of signals can be accomplished using Random Modulation Pre-Integrator (RMPI). In this work, the RMPI is implemented on the Field Programmable Analog Array (FPAA) to demonstrate the performance enhancement. Anadigm Dual Apex Development board with simulation software Anadigm Designer2 is used in this work. FPAA is an integrated circuit which can be reconfigured to implement different analog circuits. The functional blocks of the RMPI are realized by using available configurable analog Modules (CAMs) of the FPAA. A MATLAB/SIMULINK model of RMPI is also developed and basic sinusoidal signals are taken as input to the RMPI model as test signals. Output of the FPAA implementation is imported to MATLAB and reconstructed using Orthogonal Matching Pursuit (OMP) algorithm. Output of the OMP algorithm is multiplied by the Fourier dictionary to reconstruct the original signal. FPAA implementation based reconstruction output is compared with SIMULINK model based reconstruction output. RMSE value of is achieved in this work.