Widely Linear Minimum Variance Channel Estimation for MC-CDMA Systems Using Real Modulation

We propose a novel widely linear (WL) minimum variance (MV) channel estimation algorithm for multicarrier code-division multiple access (MC-CDMA) systems. The presented algorithm outperforms the regular MV channel estimation algorithms when the modulation scheme is real-valued (such as BPSK). We build on the results of [1] to find a highly accurate closed form expression the asymptotic bias resulting from the additive noise. We show via simulations that the asymptotic bias and the MSE are significantly reduced when widely linear MV estimation is utilized.