An information-theoretic look at MIMO energy-efficient communications

One of the main objectives of this paper is to provide an information-theoretic answer on how to maximize energyefficiency in MIMO (multiple input multiple output) systems. In static and fast fading channels, for which arbitrarily reliable communications are possible, it is shown that the best precoding scheme (which includes power allocation) is to transmit at very low power (Q → 0). Whereas energy-efficiency is maximized in this regime, the latter also corresponds to communicating at very small transmission rates (R → 0). In slow fading or quasi-static MIMO systems (where reliability cannot be ensured), based on the proposed information-theoretic performance measure, it is proven that energy-efficiency is maximized for a non-trivial precoding scheme; in particular, transmitting at zero power or saturating the transmit power constraint is suboptimal. The determination of the best precoding scheme is shown to be a new open problem. Based on this statement, the best precoding scheme is determined in several special but useful cases. As a second step, we show how to use the proposed energy-efficiency measure to analyze the important case of distributed power allocation in MIMO multiple access channels. Simulations show the benefits brought by multiple antennas for saving energy while guaranteeing the system to reach a given transmission rate target.