The diversity of STF-coded MIMO-OFDM systems with a general correlation model

Owing to insufficient antenna spaces, mobile scenarios, and multipaths in practice, transmission correlations in space, time, and frequency domains are inevitable in wireless communications. This paper studies the effect of general spatial, temporal, and frequency/path correlations on the performance of space-time-frequency (STF)-coded multiple-input, multiple-output orthogonal frequency division multiplexing (MIMO-OFDM) systems over frequency-selective block-fading channels. Specifically, we first derive an upper bound on the maximum achievable diversity by Hadamard and tensor products and analyze the effect of general spatial, temporal, and frequency/path correlations on it using rank properties of block matrices. We then address STF code designs and give two examples, one traditional STF code and another quasi-SF code, to show that our upper bound on the maximum diversity is achievable. The decoding complexity is considered in the MIMO system with arbitrary correlated fading channels using the traditional STF code. We also identify the newly developed statistical channel models for MIMO LTE and 802.11n as special cases of our STF-coded MIMO-OFDM system by showing that our theoretical diversity results match those simulated from these statistical channel models. Finally, we show that our general diversity result recovers various maximum diversity gains for different special correlation scenarios that have appeared in the literature.