Multiantenna capacity: Myths and realities

Over the last decade, the increases in capacity promised by multiantenna communication techniques have spurred many information-theoretic analyses. Furthermore, information theory has been used as a design tool to optimize the signals fed to the transmit array and to motivate signal processing strategies at the receiver. In this chapter, we catalog a number of misconceptions that have arisen in the multiantenna literature. The focus is on information-theoretic results and their interpretations, rather than on the validity of the various modeling assumptions. After some introductory material, we address several misconceptions about optimum signaling and about the impact of model features, such as antenna correlation, line-of-sight components, and intercell interference. Particular attention is given to the low- and high-power regions. We also briefly touch upon the relationship of the capacity results with some practical transmit and receive architectures. The chapter deals mostly, but not exclusively, with coherent communication. Definitions Denoting by the number of transmit and receive antennas, respectively, we shall abide by the frequency-flat complex vector model where is a deterministic scalar that represents the average channel gain, while the random matrix H is normalized to satisfy While the distribution of H is known to both transmitter and receiver, we shall specify when its realization is known to either.