The invention relates to a 5G channel
estimation time-frequency
smoothing method adopting an amplitude-phase
separation technology, and the method comprises the steps: carrying out the LS channel
estimation of a received
signal, correcting the non-ideal factors of a channel, carrying out the amplitude-phase separation expansion in a
frequency domain, separately conducting transforming to form time domains through IFFT (Inverse
Fast Fourier Transform), selecting a raised cosine window with a proper window bilateral trailing length for
smoothing, finally conducting transforming to a
frequency domain through FFT (
Fast Fourier Transform), and finishing time-frequency
smoothing calculation of channel
estimation after amplitude and phase combination. According to the method, by correcting non-ideal factors of a channel, searching of a window position can be reduced, the center position of a
time domain window can be fixed at a zero point, and amplitude and phase separation and two-path independent
processing are adopted, so that the amplitude and phase of the channel can obtain good denoising performance; and effective
data expansion can reduce
jitter at two ends of a channel, and better smoothness performance is obtained. Therefore, the problem that the channel estimation accuracy is reduced due to
energy leakage and
jitter of the two ends of the channel useful
signal can be solved.