Abstract :
Single-carrier frequency-domain turbo equalization (SC-FDTE) is investigated, with the focus on the design of soft-decision frequency-domain equalizer (SD-FDE) under minimum mean square error (MMSE) criteria. Existing SD-FDEs perform equalization on each frequency bin individually, which however only achieves suboptimal performance, and with the availability of the a priori input, block-wise operation is required to achieve optimal performance. In this paper, optimal MMSE SD-FDE designs with three soft interference cancellation (SIC) schemes are introduced. The first SIC scheme, SIC-I, utilizes the a priori knowledge from the channel decoder, the second SIC scheme, SIC-II, relies on the decoder a priori knowledge as well as the symbol-wise a posteriori soft-decision feedback of the SD-FDE itself, and the third SIC scheme, SIC-III, is comparable to the SIC-II while with a block-wise soft-decision feedback. The optimal SD-FDEs incur high computation complexity due to block-wise processing, then a suboptimal bin-wise SD-FDE with SIC-III is proposed to achieve complexity-performance tradeoff. It is shown that the suboptimal SD-FDE with SIC-III considerably outperforms existing suboptimal SD-FDE with SIC-I, while at a similar magnitude of complexity. It also achieves better performance than the optimal SD-FDE with either SIC-I or SIC-II. Compared to the optimal SD-FDE with SIC-III, the performance loss is marginal.
