Extrinsic LLR of Soft-Demapper under Presence of Channel Estimation Errors

Abstract

This paper proposes a novel extrinsic log likelihood ratio (LLR) of soft demapper to be conveyed to maximum a-posteriori probability (MAP) decoder under the presence of channel estimation errors. In principle, the extrinsic LLR should obey Gaussian distribution for the sake of reducing computational complexity in the MAP decoder. However, channel estimation errors deteriorate Gaussianity of the LLR derived by conventional definition. In order to guarantee the Gaussianity, this paper defines alternative extrinsic LLR according to mean square error (MSE) of channel estimates. The derivation of proposed LLR is extended to an iterative channel estimation scenario. We will demonstrate that the proposed extrinsic LLR is capable of improving bit error rate (BER) performances.