<p> The goal of this thesis is the development of a computationally-efficient coded system that enables communication over the non-coherent Multiple-Input Multiple-Output (MIMO) fiat-fading wireless channel at high data rates. The proposed signalling technique applies the principles of Bit-Interleaved Coded Modulation (BICM) with Iterative Demapping and Decoding (IDD) to non-coherent MIMO communication systems. </p> <p> The principle of BICM is applied to a constellation that mimics the non-coherent capacity achieving distribution at high signal to noise ratios. The capacity achieving distribution is in the form of isotropically distributed unitary matrices, and the constellation can be represented by points on a Grassmannian manifold. A mapping technique that exploits the Grassmannian geometry is proposed. This mapping technique is based on the partitioning of the constellation into two subsets. The Grassmannian geometry also gives rise to an efficient list-based demapping algorithm that substantially reduces the computational complexity of the receiver while incurring some degradation in performance. For example, at a bit error rate (BER) of 10-4 the signal to noise ratio (SNR) performance degradation with respect to full constellation demapping is approximately 1. 75 dB. A technique by which the decoder can augment the demapping list is proposed, and it is shown that the performance degradation of the efficient algorithm can be rendered insignificant (approximately 0.2 dB at a BER of 10-4). </p> <p> Finally, the performance of the proposed BICM-IDD using the Grassmannian constellation will be compared to that of a corresponding training-based BICM-IDD scheme. These simulations show that the proposed scheme can provide better performance at high data rates; e.g., for a data rate of 5/3 bits per channel use, the performance gap is almost 1 dB at BER of 10^(-4). </p> / Thesis / Master of Applied Science (MASc)