Analysis of DFT-Based Channel Estimation for Uplink Massive MIMO Systems

To reduce the edge effect in discrete Fourier transform-based channel estimation, various schemes have been proposed to estimate the channel response at the non-allocated subcarriers. However, it is not easy to estimate the channel response accurately at the non-allocated subcarriers, particularly at low signal-to-noise ratio (SNR). Moreover, the estimation places an additional computational burden on the receiver. The analysis in this letter shows that the edge effect is not serious in uplink massive multiple-input multiple-output systems. Zero padding at the non-allocated subcarriers is asymptotically optimal in the low SNR regime. We can also use fast Fourier transform/inverse fast Fourier transform pruning to reduce the computational complexity dramatically.