Channel Estimation on MIMO-OFDM Systems using Deep Learning

During communication, the transmitted signal passes through a channel, owing to which noise and distortions are added to the signal through various sources. These need to be removed for proper analysis of the received signal. To do this, the first step is to figure out the characteristics of the channel that the signal has passed through. With the help of the known characteristics, we can predict how the channels will the transmitted signal and thus allow us to employ methods to reduce the noise added to the signal due to the channel. This technique of characterizing the channel is termed channel estimation. The process of determining the channel statistic information of a channel is coined as channel estimation. Various ways of channel estimation exist. Due to the advent of newer technologies, significant improvements are seen in wireless communications that have led to 5G and beyond networks. One of the key technologies that has aided in this advancement is deep learning, which has been of great help in supplementing the dependability and diminishing the complex computations that are involved. In this paper, we will construct a MIMO-OFDM system and transmit random signals. We will examine the signal at the receiver side to determine the channel properties. Furthermore, we present a novel channel estimation approach that leverages deep learning techniques to improve on the existing channel estimations obtained using conventional deep learning methods.