ABSTRACT
The approach of applying a cascaded network consisting of radial basis function nodes and least square error minimization block to Compressed Sensing for recovery of sparse signals is analyzed in this paper to improve the computation time and convergence of an existing ANN based recovery algorithm. The proposed radial basis function-least square error projection cascade network for sparse signal Recovery (RASR) utilizes the smoothed L0 norm optimization, L2 least square error projection and feedback network model to improve the signal recovery performance over the existing CSIANN algorithm. The use of ANN architecture in the recovery algorithm gives a marginal reduction in computational time compared to an existing L0 relaxation based algorithm SL0. The simulation results and experimental evaluation of the algorithm performance are presented here.
