Low-Complexity Iterative Soft-Demapper for Multidimensional Modulation Based on Bitwise Log Likelihood Ratio and its Demonstration in High Baud-Rate Transmission

We propose a low-complexity demapping scheme for multidimensional quadrature amplitude modulation (QAM) format using a bitwise log likelihood ratio (LLR). A multidimensional modulation technique such as eight-dimensional (8D) can improve the optical signal-to-noise ratio tolerance by enlarging the minimum Euclidean distance between multidimensional symbols consisting of the plural QAM constellations. However, the optimal soft demapper for these modulation formats increases computational complexity when the formats have higher dimensions and higher order multilevel signals are involved. The rise in computation complexity comes from the fact that these formats have large number of multidimensional symbols. The proposed algorithm reduces the complexity by using the bitwise LLR of original QAM constellation. Furthermore, the proposed soft demapper is suitable for iterative decoding fed back to soft decision forward error correction, such as binary low-density parity check code, since this algorithm uses only the bitwise LLR. We experimentally verify the decoding performance for 120 GBaud 8D-16QAM of the proposed demapper and iterative decoding and demonstrate over 3900 km wavelength division multiplexed transmission.