An intelligent autoencoder for reducing peak to average power in advanced optical 5G radio network for 64-QAM transmission schemes

Abstract In this study, we present an intelligent autoencoder-based method to address the advanced optical 5G radio networks’ peak-to-average power ratio (PAPR) problem. Effective PAPR reduction strategies are necessary because high PAPR in multicarrier communication systems, like OFDM, causes power inefficiencies and signal distortion. By utilizing autoencoders’ potent representation learning capabilities, our approach decreases PAPR by encoding and decoding signals with less amplitude fluctuations. The autoencoder is trained using a sizable dataset of multicarrier signals, which helps the model learn how to compress and rebuild signals with as few peak excursions as possible. The suggested method is tested in a 64-QAM and 256 subcarrier 5G system that is representative of metropolitan settings with substantial line-of-sight components and operates in a Rician fading channel environment. The intelligent autoencoder considerably lowers PAPR while preserving signal integrity and overall system performance, according to simulation data. By comparing the suggested method’s performance to that of conventional PAPR reduction strategies, it is shown to be more effective in terms of both bit error rate (BER) and PAPR reduction. This study demonstrates how sophisticated machine learning approaches can be incorporated into next-generation optical 5G communication systems to enable more dependable and efficient 5G networks.