부산대학교 도서관

When we consider a large number of substreams and constellations in channel-coded space division multiplexing (SDM) in a MIMO channel, computational complexity of a soft MIMO decoder based on the maximum likelihood (ML) detection becomes enormous. Even in complexity-reduced processing such as sphere decoding, the MIMO detector must search several non-ML points to calculate a log-likelihood ratio (LLR) of each bit for a soft-input channel decoder. In the paper, we propose a simplified iterative processing method for the soft MIMO detector and soft-input/soft-output (SISO) channel decoder to decrease the number of Euclidean distance calculations of non-ML points. To update the non-ML points to be searched, we monitor the LLR value of each bit returned from the SISO channel decoder. The computer simulation assuming 4tx-4rx turbo-coded SDM of 16-QAM shows that the proposed method provides the gain of 0.7 dB at FER of 0.01 against the non-iterative method using 65536 points, where the number of searched points per iteration at average E/sub b//N/sub 0/ = 1 dB was 17.8 in average and 624 at most in the proposed method.