High-Order Super-Block GLRT for Non-Coherent Grassmann Codes in MIMO-OFDM Systems

We investigate non-coherent multi-input multi-output (MIMO) signal processing which requires no channel state information (CSI) at either the transmitter or the receiver. With non-coherent codes on Grassmann manifold, a receiver employing generalized likelihood ratio test (GLRT) algorithm offers the maximum-likelihood performance even without CSI. However, the conventional GLRT suffers from a severe performance degradation when the channel changes fast within a coding block duration. We propose an improved GLRT algorithm referred to as high-order super-block techniques. The super-block scheme makes effective use of correlated channels for adjacent blocks in slow fading, whereas the high-order scheme can overcome the channel fluctuation during a block in fast fading. We demonstrate that the proposed scheme significantly improves performance for MIMO-OFDM with non-coherent Grassmann space-frequency block codes (SFBC).