Using Equalization-Induced Noise Coloring to Improve Error-Correcting Decoding
Jane A. Millward, Ken R. Duffy, Muralidhar Rangaswamy, Muriel Médard · 2024
The effects of inter-symbol interference (ISI) are known to degrade communications performance. In this paper we show that by accounting for the effects of inter-symbol interference as a form of colored noise in Guessing Random Additive Noise Decoding (GRAND) we are able to obtain performance improvements in terms of bit and block error rate. We use a soft-information variant of GRAND called Ordered Reliability Bit Guesing Random Additive Noise Decoding - Approximate Independence (ORBGRAND-AI) which accounts for both soft information and statistical structure present in the noise for the purpose of investigating decoding performance in ISI channels. We show that over 4 dB gain can be obtained by using ORBGRAND-AI decoding in channels which are subject to ISI over systems which do not employ any forward error correction. We generate the ISI channel coefficients using data from RFView, the state-of-the-art high fidelity, site-specific physics-based RF modelling and simulation tool.