Investigation on LLR Computation at FDE Output with Receiver Diversity Using Low-Rate Turbo Code for DFT-Precoded OFDMA

This paper presents the block error rate (BLER) performance of a posteriori log-likelihood ratio (LLR) computation methods at the output of a frequency domain equalizer (FDE) with antenna diversity reception using a low-rate turbo code for discrete Fourier transform (DFT)-precoded orthogonal frequency division multiple access (OFDMA). In the paper, we propose an a posteriori LLR computation method based on the minimum squared Euclidean distance of the signal after coherently combining the FDE output of all receiver diversity branches. We compare the BLER performance level of the proposed LLR computation method to that for the conventional LLR computation method in which the a posteriori LLR is given as a summation of the minimum squared Euclidean distances of all receiver branches. In the BLER comparison, we consider the influences of noise correlation after frequency domain equalization and the channel estimation error. Computer simulation results show that the LLR computation using the signal after combining the receiver branches is more appropriate than the LLR computation with combining the squared minimum Euclidean distances in the probability domain with a low-to-high turbo coding rate as the a posteriori LLR at the FDE output associated with receiver diversity for DFT-precoded OFDMA.