Study on Channel Prediction in IRS-Assisted Wireless Com-munication Systems

The use of intelligent reflecting surface (IRS) is being investigated for wireless communication in high frequency bands. By appropriately controlling the reflection coefficient of each IRS element, high quality communication can be achieved even in non-line-of-sight environment. The phase control of the IRS requires the determination of the optimal phase pattern based on the estimated channel, which is then fed back to the IRS through the control channel. To reduce channel estimation overhead, we propose a channel prediction method based on Gaussian process regression. We evaluate the proposed method performance on the points of signal-to-noise ratio (SNR) and confirm that the proposed approach can mitigate SNR degradation caused by terminal movement compared to the system without prediction.