Adaptive Low-complexity Orthogonal Matching Pursuit Channel Estimation for DCO-OFDM Systems

We propose an adaptive low-complexity orthogonal matching pursuit (ALOMP) channel estimation approach for direct current biased optical-orthogonal frequency division multi-plexing (DCO-OFDM) systems in optical wireless communication (OWC), requiring a single DCO-OFDM block. This is the first investigation to utilize light-emitting diode (LED)'s limited bandwidth to reduce the high complexity of the traditional OMP method in OWC, while maintaining good channel estimation performance. A number of termination factors are designed to reduce the number of iterations, by exploring correlations between channel impulse responses (CIRs) in time domain caused by LED's limited bandwidth. The proposed ALOMP approach provides complexity close to the minimum mean square error (MMSE) method, and requires just two iterations, significantly less than the traditional OMP method with a large number of iterations. We employ OMP to estimate two CIRs in time domain, and formulate two equations. The line-of-sight (LoS) channel and LED's limited bandwidth are estimated separately via the formulated equations rather than jointly in all CIRs in time domain or on all sub carriers in frequency domain in previous works. Also, we derive the lower bound of the proposed ALOMP approach, and the theoretical bit error rate (BER) including channel estimation errors. Simulation results verify the proposed ALOMP approach.