A Neural Network Aided Approach for LDPC Coded DCO-OFDM with Clipping Distortion

In visible light communications (VLC), a neural network-aided bit-interleaved coded modulation (NN-BICM) receiver is designed to mitigate the clipping distortion for the LDPC coded DC biased optical orthogonal frequency division multiplexing (DCO-OFDM) systems. Taking the cross-entropy as loss function, the feed forward network is trained by backprop-agation algorithm to output the condition probability through the softmax activation function, thereby assisting in a modified log-likelihood ratio (LLR) improvement. This feed-forward network simplifies the input layer with a single symbol and the corresponding Gaussian variance instead of focusing on the inter-carrier interference over multiple subcarriers. To achieve a further gain, a staked NN-BICM (S-NN-BICM) iterative receiver is proposed by calculating the condition probability with the aid of apriori probability that derived from the extrinsic LLRs in the LDPC decoder at the previous iteration. Both NN-BICM and S-NN-BICM schemes achieve significant performance gains over the existing counterparts, especially more than 0.8 and 1.4 dB in the case of 16-QAM and 511 occupied subcarriers.