Performance of Decision-Directed Channel Estimation Using Low-Rate Turbo Codes for DFT-Precoded OFDMA

Abstract

A low-rate turbo code can reduce the required average received signal-to-noise power ratio (SNR) that satisfies the target block error rate (BLER) due to its high coding gain. When a low-rate turbo code is used however, channel estimation (CE) error becomes a bottleneck in bringing out the potential coding gain. Therefore, this paper presents the average BLER performance using decision-directed channel estimation (DDCE) for a low-rate turbo code in uplink Discrete Fourier transform (DFT)-precoded orthogonal frequency division multiple access (OFDMA). In the proposed DDCE, we employ soft-decision symbol estimation based on the log-likelihood ratio (LLR) of an extrinsic probability at the Max-Log-MAP (maximum a posteriori probability) decoder output in addition to reference signals (RSs). The DDCE works synergistically with the low-rate turbo code, since the extrinsic LLR at the Max-Log-MAP decoder becomes more reliable due to the increasing coding gain. Computer simulation results show that when the turbo code with R = 1/15 (1/9) and the constraint length of K = 4 bits is used, the required average received SNR at the average BLER of 10-2 using the DDCE is decreased by approximately 0.6 (0.4) dB compared to that for the RS based CE. As a result, the required average received SNR using the DDCE for R = 1/15 and 1/9 is decreased by approximately 7.0 and 5.0 dB, respectively, compared to that for R = 1/3.