Deep Learning Based Diversity Combining for Generic Noise and Interference
Imtiaz Ahmed, Evan Allen · 2020
In this work, we develop a deep learning (DL) based robust data detection algorithm for a receive-diversity system, where the data is corrupted by Gaussian and different non-Gaussian noise and interference in different diversity branches. A fully-connected deep neural network (DNN) is designed for this purpose and is trained with different noisy datasets offline. The developed DNN is then applied in real-time for data detection process. We emphasize that the detector does not require the noise distribution to be known for its operation. Furthermore, by the use of simulations, we show that the proposed DL-based detector performs much better than the conventional maximal ratio combining (MRC) detector for non-Gaussian noise. Moreover, simulation results point out that the proposed detector is robust enough to operate in time-varying noisy environment.