Iterative Equalization With Soft Feedback With a Subsequent Stage Employing Error Search and Correction

In this paper, a second stage for algorithms applying iterative soft decision interference cancellation (ISDIC) for channel equalization is proposed, which performs error search and correction. Analysis by simulations shows that matched filter (MF)-based ISDIC (MF ISDIC) with subsequent second stage can outperform more complex minimum-mean-square-error (MMSE) based ISDIC. Furthermore, the latter can be improved by the proposed second stage by up to 2 dB. It is shown that the ISDIC schemes with subsequent second stage can reach the MF bound up to a fraction of a decibel for all analyzed channels and 4-ary quadrature amplitude modulation (QAM) transmission, and up to 1 dB for 16-ary QAM transmission. In the work of Gerstacker , it has been reported that MF ISDIC performs very well for highly dispersive channels, e.g., with 60 symbol-spaced taps, but for channels of moderate dispersion, an error floor occurs. By utilizing performance bounds for delayed decision-feedback sequence estimation (DDFSE), we show that now for channels of moderate length, e.g., 20 symbol-spaced taps, even low-complexity MF ISDIC with the proposed subsequent second stage outperforms a DDFSE approach with states.