Subspace-based blind channel identification for noncircular multicarrier transmissions

We propose a new blind channel identification method for multicarrier (MC) transmission systems employing noncircular modulation formats. To exploit the noncircularity property of the signal constellation, the proposed algorithm relies on the subspace decomposition of the augmented vector obtained by stacking the received data vector and its complex-conjugate version. Unlike recently proposed subspace-based techniques, this method allows one to successfully estimate the channel impulse response under certain identifiability conditions, without requiring the introduction of additional redundancy in transmission (e.g., virtual carriers) and without resorting to oversampling and/or multiple antennas at the receiver. Moreover, it is shown that blind channel identification can be performed satisfactorily also in the presence of narrowband interference (NBI), requiring only approximate rank determination of the NBI augmented autocorrelation matrix. Finally, numerical simulations are carried out for comparing the performance of the proposed algorithm with that of an existing subspace-based method.