The problem of estimating the carrier frequency offsets in Multiple-Input Multiple-Output (MIMO) systems with distributed transmit antennas is addressed. It is supposed that the transmit antennas are distributed while the receive antennas are still centralized, and the general case where both the time delays and the frequency offsets are possibly different for each transmit antenna is considered. The channel is supposed to be frequency flat, and the macroscopic fading is also taken into consideration. A carrier frequency offset estimator based on Maximum Likelihood (ML) is proposed, which can separately estimate the frequency offset for each transmit antenna and exploit the spatial diversity. The Cramer-Rao Bound (CRB) for synchronous MIMO (i.e., the time delays for each transmit antenna are all equal) is also derived. Simulation results are given to illustrate the per- formance of the estimator and compare it with the CRB. It is shown that the estimator can provide satisfactory frequency offset estimates and its performance is close to the CRB for the Signal-to-Noise Ratio (SNR) below 20dB.
The paper proposed an improved Ordered Successive Interference Cancellation(OSIC) detection scheme for V-BLAST systems with square/rectangular Quadrature Amplitude Modulation(QAM) modulation.It utilizes an equivalent real-valued vector expression of relationship between transmit signals and received signals,exploits the constellation's product-form structure and can eventually make the order,in which components of the transmit signals vector are extracted,more "optimal" in some sense.Thereby,it can offer an improved error probability as compared to the conventional OSIC detection scheme.In addition,the paper also proposes an efficient projection al-gorithm to calculate nulling vectors in a simple recursive fashion in order to avoid the vast increase of complexity,which is due to the fact that the original complex N -dimensional data vector and M×N channel matrix are transformed into a real 2N -dimensional data vector and a real 2M ×2N channel matrix respectively.A scrutinous complexity analysis shows that the complexity increases by only 33% as compared to the conventional scheme.
