Performance analysis of energy-efficient adaptive modulation.

摘要

The focus of this dissertation is on the development and performance analysis of techniques that use perfect channel state information (CSI) at the transmitter to configure the system to operate in the most energy-efficient manner. In order to perform a mathematical optimization of the various problems considered, we define and motivate a metric called energy-per-goodbit (EPG), which represents the total energy cost to convey one bit of information without error at the receiver. Although the methods developed are quite general and can be extended to several communication scenarios of interest, it is convenient to classify the work in to two main categories: (a) scalar channel (b) parallel channel. First, we consider a single-carrier QAM system (scalar channel), and find the optimum constellation size (spectral efficiency) and transmit power that minimizes the EPG. The aspect that complicates this problem is that the transmitter power backoff (to avoid nonlinear region of PA) is a function of the constellation peak-to-average power ratio (PAR). We characterize the statistical performance of the solution under assumption of a frequency-flat, long-term static Rayleigh fading channel. Second, we consider a single-carrier MIMO-MRC system (scalar channel) and find the optimum transmit power that minimizes EPG as a function of the number of antennas and the channel fade. Due to the simplifications that result from assuming a constant backoff, unlike the first problem, we derive the closed-form CDF of the optimum EPG. Using this result, we find the optimum number of antennas from an energy-efficiency point of view. We consider Rayleigh and Rician fading between the antennas. Third, we consider minimizing the EPG of an OFDM system in a frequency-selective channel (parallel channel). We derive an energy-efficient water-filling (power allocation) algorithm. The energy efficiency formulation subsumes both the classical maximum rate (MR) and maximum margin (MM) problems as specific cases. As a result, the energy efficiency viewpoint provides a convenient and unified perspective of the various water-filling solutions. Finally, we extend the parallel channel formulation to a MIMO-SVD system and MIMO-OFDM-SVD systems (parallel channel). Assuming Rayleigh and Rician fading channels, we characterize (using simulation) the optimum number of antennas and also the optimum average number of spatial streams that are utilized as a function of transmit cost and fading conditions. We also study the impact that frequency diversity has on MIMO-OFDM EPG performance. We conclude by suggesting some areas for further research.