Layered Space-Time Structure for MIMO-OFDM Systems
MetadataShow full item record
The low complexity of layered processing makes the layered structure a promising candidate for MIMO systems with a large number of transmit antennas and higher order modulation. For broadband systems, orthogonal frequency division multiplexing (OFDM) appears promising for its immunity against delay spread. In addition, OFDM is especially suitable for frequency selective MIMO systems since the introduction of orthogonal subcarriers makes system design and implementation as simple as those for flat fading channels. Therefore, the combination of layered structure with OFDM is a promising technique for high-speed wireless data transmission. The proposed research is focused on the layered structure for MIMO-OFDM systems, where several techniques are proposed for performance enhancement, namely, channel estimation based on subspace tracking, parallel detection of group-wise space-time codes by predictive soft interference cancellation, quasi-block diagonal low-density parity-check codes (LDPC) coding and statistical data rate allocation for layered systems. For MIMO-OFDM systems, rank reduction by some linear transform matrix is necessary for channel estimation. In the proposed research, we propose a channel estimation algorithm for MIMO-OFDM systems, which uses the optimum low-rank channel approximation obtained by tracking the frequency autocorrelation matrix of the channel response. Then parallel detection algorithm is proposed for a modified layered system with group-wise space-time coding, where the structure of particular component space-time code trellises is exploited using partial information from the Viterbi decoder of the simultaneously decoded interfering component codes. Next we incorporate the layered structure with LDPC to develop a quasi-block diagonal LDPC space-time structure. The lower triangular structure of the parity check matrix introduces correlation between layers. Each layer, as a part of the whole codeword, can be decoded while taking information from other undetected layers to improve the decoding performance. In the end, a modified layered structure is proposed where the layer detection order is fixed and the data rate for each layer is allocated based on the detection order and channel statistics. With Gaussian approximation of layer capacities, we derive the optimum data rate allocation.