TABLE OF CONTENTS As seen in Chapter 1, many signal processing techniques have been developed with very simple assumptions regarding physical wireless channels, including multi-antenna channels. In particular, we have so far considered that the elements of the MIMO channel matrix are uncorrelated variables. In practical scenarios, these assumptions may be far from realistic. Yet the characteristics of radio propagation environments dictate the ultimate performance of wireless systems, irrespective of the assumptions used to design the system. Therefore, the goal of Chapters 2 and 3 is to investigate realistic models and analytical representations of wireless channels for MIMO systems.
When dealing with MIMO channels, space comes as an additional dimension, which needs to be modeled on its own in the same way as time and frequency variations have been modeled for wideband SISO systems. This multi-dimensionality should always be kept in mind when modeling MIMO channels or when designing space-time codes. As an example, let us consider that the angular distribution of the energy has to be modeled at both the transmit and receive sides. This leads to a so-called double-directional description:
the term directional indicates that the channel description includes a model of the angular distribution of the energy at the antennas, by contrast to a non-directional model, which deals only with the temporal spreading;
the term double means that the spatial description concerns both sides, i.e. that there are multiple antennas at transmit and receive sides.
Naturally, the more usual notions, such as the channel stationarity and...
