Stacked Denoising Autoencoder Enhanced Polar Codes Over Rayleigh Fading Channels

Polar codes, with low encoding/decoding complexity and capacity-achieving potential, have drawn much attention recently. It is very critical to study the impact of fading on polar codes implementation into wireless communications. Existing research works on polar codes over the fading channel use simplified fading channel models while further reducing the frame error rate is highly needed. In this letter, using the structure of polar codes we propose the deep learning based scheme for polar codes over Rayleigh fading channels. Simulation results verify that our proposed scheme can significantly decrease the frame error rate for polar codes over Rayleigh fading channels.