MLSE for DSTBC-OFDM Detection with Channel Estimation by Blind Linear Prediction and Subcarriers Interpolation

This paper proposes low-complexity blind detection for orthogonal frequency division multiplexing (OFDM) systems with the differential space-time block code (DSTBC) under time-varying frequency-selective Rayleigh fading. The detector employs the maximum likelihood sequence estimation (MLSE) in cooperation with the blind linear prediction (BLP), of which prediction coefficients are determined by the method of Lagrange multipliers. Interpolation of channel frequency responses is also applied to the detector in order to reduce the complexity. A complexity analysis and computer simulations demonstrate that the proposed detector can reduce the complexity to about a half, although it suffers a loss of 2.5 dB in E_b/N₀ at BER of 10^-3 owing to the complexity reduction when the prediction order and the degree of polynomial approximation are 2 and 1, respectively, with the normalized Doppler frequency of 1.6 times 10^-2.