Bit-Interleaved Coded Modulation in the Presence of Unknown Impulsive Noise

Bit-interleaved coded modulation (BICM) is a bandwidth efficient coding scheme and has been extensively studied for Gaussian distributed additive channel noise. In many real-world communication systems, the additive noise is non-Gaussian impulsive. In such situations, the conventional squared Euclidean distance based decoding metric suffers from severe metric mismatch. The optimal metric in maximum likelihood sense may not be applicable because the exact noise probability density function is unavailable at the receiver due to the dynamically changing nature of the impulsive noise (IN). It is therefore important to design decoding algorithms that are robust to the IN distributions. In this work, we propose two novel decoding schemes for BICM in the presence of unknown IN. The proposed decoding schemes alleviate metric mismatch by automatically ignoring the impulse-corrupted symbols in decoding. It is demonstrated that the proposed decoding schemes offer great performance improvement over the conventional decoder counterparts.