Channel Estimation And Optimal Pilot Sequence Design For MIMO OFDM Systems

Orthogonal Frequency Division Multiplexing (OFDM) is an effective method for high data rate transmission, and is being pursued as a potential candidate for the future mobile communication systems due to its robustness against frequency-selective fading channels, bandwidth efficiency and simple implementation. Channel estimation is one of the key techniques for OFDM systems. Compared with the differential demodulationn, coherent demodulation with accurate channel information can achieve 3-4 dB gain in signal-to-noise ratio (SNR). Therefore, in the fourth generation (4G) systems which require high data rate, high quality and high mobility, channel estimation is an indispensable part. This thesis focuses on this topic and the corresponding design issue of the optimal pilot sequence.Since channel estimation has attracted a lot of attention, there have been many literatures on this issue. This thesis summarizes some classical channel estimation methods. After analyzing the Multiple Input Multiple Output (MIMO) OFDM systems, some new conclusions are then obtained. In order to avoid solving underdetermined equations for MIMO OFDM systems, an improvement algorithm which combines successively transmitted pilot sequences is proposed. If pilot sequences are sufficient, more than one estimate can be obtained. Consequently, a combined way with consideration of the estimation mean square error (MSE) is given.In practical OFDM systems, virtual subcarriers are usually used to avoid the transmitted data being distorted by the low-pass filter on the transmitter side. Thus, with the consideration of these virtual subcarriers, we can find that the conventional uniformly spaced approach will not be applicable anymore in some situations. It is, therefore, necessary to derive the optimal pilot sequences with the nonuniform pilot tone placement. Based on the least square (LS) channel estimation method, the optimal pilot sequences of MIMO OFDM systems are derived in this thesis to minimize the estimation MSE. It can be shown that such a design issue is equivalent to solving a system of linear inequalities. For such inequalities, three solving algorithms are given, which are solving an equivalent constrained programming problem, solving an equivalent unconstrained optimization problem, and directly using Gram-Schmidt orthogonalization, respectively.Usually, channel can be modeled as a finite impulse response (FIR) filter. But for sparse multipath channel, the parametric model can approximate the actual environment better. Since there are only path delays and path gains to be estimated in the parametric model, the estimation dimension can be reduced a lot for sparse channel, and consequently, the estimation performance can be improve. Based on parametric channel model, a maximum likelihood (ML) estimation scheme is proposed for sparse channel in this thesis. First, the number of channel path are detected by minimum-description-length (MDL) criterion and channel path delays are initialized by Estimation of signal-parameters-by-rotational-invariance-techniques (ESPRIT). Then the path delays and path gains are obtained by proposed ML approach. To reduce the...