Channel Estimation and Pilot Overhead Reduction in OFDM Systems Using Compressed Sensing Dynamic Mode Decomposition

This work investigates the potential of employing the approach Compressed Sensing Dynamic Mode Decomposition (CS-DMD) in the context of time-varying wireless channels. To the best of the authors' knowledge, this marks the first instance of utilizing CS-DMD for pilot-based channel estimation in Orthogonal Frequency Division Multiplexing (OFDM) systems. The effectiveness of this method is compared with two advanced deep learning-based channel estimation techniques: Interpolation-ResNet and Learned Approximate Message Passing (LAMP). Furthermore, we leverage the advantageous characteristics of DMD in analyzing complex nonlinear dynamic systems to predict the future state of the channel, thereby reducing the required pilot signals. Simulation results show that utilizing CS-DMD can achieve superior channel estimation performance with less pilot overhead.