SPC04-3: Parallel Channel Estimator for Iterative MAP Receiver using Message-Passing Algorithm

This paper proposes a parallel recursive least square (P-RLS) channel estimator for the iterative maximum a posteriori (MAP) receiver. The targeted system is the low-density-parity-check (LDPC) coded multiple-input-multiple-output (MIMO) and orthogonal frequency-division multiplexing (OFDM) system. The proposed estimator, which can be theoretically derived from the message-passing algorithm, can track a fast fading channel, and allows the parallel implementation of the whole receiver. A simplified version of P-RLS (SP-RLS) that can reduce the computational complexity is also proposed. Compared to the RLS with smoothing and removing (SR-RLS), P-RLS can reduce the computational time by parallelization, and SP-RLS can further reduce the computational time and complexity. Computer simulations show that SP-RLS achieves almost the same packet error rate (PER) performance as that of P-RLS. It also achieves almost the same PER performance as that of SR-RLS with double iterations. Even in this case, the computational time and complexity of SP-RLS are smaller than those of SR-RLS.