Channel estimation technique for mitigating ICI in MIMO-OFDM cellular systems

Maximum-likelihood (ML) channel estimation for multiple-input-multiple-output (MIMO) channels is derived when orthogonal frequency division multiplexing (OFDM) is employed in cellular systems. While conventional single-cell ML estimation (SCMLE) disregards the effects of inter-cell interference (ICI), multi-cell ML estimation (MCMLE) exploits the knowledge of the preambles from neighboring base stations (BSs) to combat severe ICI. The mean square error (MSE) of the MCMLE is smaller than the conventional SCMLE at the cell boundaries where signal-to-interference ratio (SIR) is about 0 dB. However, the MCMLE performs worse than the SCMLE when the mobile station (MS) is near BS because the MCMLE estimates more parameters than the SCMLE. In addition, we evaluate the influence of non-ideal preambles to the performance of channel estimation because all the preambles in cells cannot be mutually orthogonal in practice. Numerical results confirm the superiority of the MCMLE to the SCMLE when SIR has a range of 0∼15 dB.