الملخص
Accurate knowledge of channel state information (CSI) is one of the main challenges in Massive MIMO, especially in the case of multiple users in the system, where accurate estimation of channel state becomes a fundamental pillar of good performance.
Users log in to the base station via the uplink and the base station sends its data to the user via the downlink. The channel state is usually estimated at the uplink due to the computational superiority of the base station over the user equipment UE.
Deep learning technique was used to estimate the state of the channel on the downlink assuming several nonlinear models in the radio interfaces and analyze their performance and compare them with the linear model
The numerical results confirm the merit of the proposed solution, and show that effective performance can be obtained compared to well-known conventional solutions that assume linear behavior of channel components, or methods that rely on least squares.