One-dimensional soft-demapping using decorrelation with interference cancellation for rotated QAM constellations

Signal space diversity (SSD) has been exploited to improve detection performance of quadrature amplitude modulation (QAM) over fading channels and recently adopted for the second generation of digital video broadcasting system (i.e., DVB-T2). To fully exploit SSD, maximum likelihood (ML) detection has been used for soft-demapping. Because of exponential complexity of the ML detector or its max-log approximated detection (i.e., full search algorithm), sub-region based soft-demapping algorithms are proposed. Those algorithms still have high complexity due to two dimensional Euclidean distance calculation. In this paper, we propose one-dimensional soft-demapping algorithms with decorrelation. By reformulating a received rotated QAM signal as two received PAM signals, the ML detection for rotated QAM is simplified to a minimum mean squared error (MMSE) decorrelation followed by per channel soft-demapping under Gaussian assumption on post detected interference. In addition, interference cancellation (IC) is considered to reduce residual interference after MMSE decorrelation. For 256 QAM with 4/5 code rate in memoryless Rayleigh channels with/without erasure events, the performance gap between the proposed one-dimensional soft-demapping with decorrelation and the full search algorithm is within 0.1dB at 10^-4 bit error rate (BER), while the complexity is less than 1% of the full search algorithm.