Concatenated codes for the multiple-input multiple-output quasi-static fading channel

The use of multiple antennas at the transmitter and/or the receiver promises greatly increased capacity. This can be useful to meet the ever growing demand of wireless connectivity, provided we can find techniques to efficiently exploit the advantages of the Multiple-Input Multiple-Output (MIMO) system. This work explores the MIMO system in a flat quasi-static fading scenario. Such a channel occurs, for example, in packet data systems, where the channel fade is constant for the duration of a codeword and changes independently from one transmission to another. We first show why it is hard to compute the true constrained modulation outage capacity. As an alternative, we present achievable lower bounds to this capacity based on existing space-time codes. The bounds we compute are the fundamental limits to the performance of these space-time codes under maximum-likelihood decoding, optimal outer codes and asymptotically long lengths. These bounds also indicate that MIMO systems have different behavior under Gaussian signaling (unconstrained input) and under the finite alphabet setting. Our results naturally suggest the use of concatenated codes to approach near-capacity performance. However, we show that a system utilizing an iterative decoder has a fundamental limit ? it cannot be universal and therefore it cannot perform arbitrarily close to its outage limit. Next, we propose two different transceiver structures that have good performance. The first structure is based on a novel BCJR-decision feedback decoder which results in performance within a dB of the outage limit. The second structure is based on recursive realizations of space-time trellis codes and uses iterative decoding at the receiver. This recursive structure has impressive performance even when the channel has time diversity. Thus, it forms the basis of a very flexible and robust MIMO transceiver structure.