Deep Learning-Based Quantization of L-Values for Gray-Coded Modulation

In this work, a deep learning-based quantization scheme for log-likelihood ratio (L-value) storage in fading channels affected by interference is introduced. We derive the number of sufficient statistics required to exactly reconstruct the set of L-values corresponding to a channel use as 3+2xNI, where NI is the number of interferers. We analyze the dependency between the average magnitudes of different L-values and show they follow a consistent ordering, regardless of the channel coefficient or interference distribution. Based on this we design a deep autoencoder that jointly compresses and separately reconstructs each L-value, allowing the use of a weighted loss function that promotes more accurate reconstruction of low magnitude inputs. Our method is shown to be competitive with state-of- the-art maximum mutual information quantization schemes, reducing the required memory footprint by a ratio of up to two and achieving a loss of performance lower than 0.1 dB with less than two effective bits per L-value and lower than 0.04 dB with 2.25 effective bits. We demonstrate that the same network can be reused without further training on various channel models and error- correcting codes while preserving the same performance benefits.