Iterative decoding and demodulation for single-antenna vector OFDM systems

Vector orthogonal frequency-division multiplexing (OFDM) for single-antenna systems generalizes OFDM in that it converts an intersymbol interference (ISI) channel into multiple "ISI-free" vector channels and involves vector channel matrices instead of channel coefficients in its one-tap equalization (demodulation). Over frequency-selective fading channels, vector OFDM with increasing vector length implicitly offers more signal space diversity advantages over OFDM. However, the complexity of the brute force demodulator increases exponentially in terms of the vector size. In this paper, iterative demodulation and decoding for convolutionally coded vector OFDM systems are studied, and in particular, low-complexity demodulation schemes that employ linear cancellation and feature linear complexity in terms of the vector length are proposed and compared. Following the introduction of demodulators with ideal linear cancellation as benchmark, demodulators with hard and soft interference cancellation and their initialization options are studied. The demodulator with soft interference cancellation dynamically updates the symbol estimates and their instantaneous reliability measure based on the code bit probabilistic information passed from preceding iterations. It alleviates error propagation that is detrimental to the demodulator with hard interference cancellation and shows performance close to that of the brute force demodulator, with great overall complexity reduction. Extensive simulation results demonstrate how the choice among the presented schemes and the vector length affect the convergence rate and the achievable performance of the iterative receivers.