Energy Efficiency-spectral Efficiency Trade-off In MU-MIMO Systems

Improving energy efficiency (EE) and spectral efficiency (SE) of wireless communication systems is one of the important objectives of5G technologies. As a conventional performance metric, SE is already widely studied due to limited bandwidth and increasing demands of users. On the other hand, the power consumption of wireless systems increases rapidly, thus green communication has become one of the important communication technologies in the future. The bit per Joule EE, as one of the performance metric used in the research of green communication, reflects the trade-off between system capacity and the power consumption. Unfortunately, maximizing EE is not always consistent with maximizing SE. In other words, there is a trade-off between these two objectives. Recognizing the EE-SE trade-off can shed light on how to balance the two metrics in future wireless systems.This thesis focuses on studying the EE-SE trade-off in multi-user multiple input multiple output (MU-MIMO) systems. MU-MIMO technologies can benefit from not only spatial diversity and spatial multiplexing gains but also cancelling the multi-user interference using the degrees of freedom (DoF) in the spatial domain. Thus, it also improves the EE-SE trade-offs in wireless systems. The closed-form approximation (CFA) of the EE-SE trade-off for MU-MIMO systems is derived to reveal the fundamental limits of the EE and the SE in MU-MIMO systems and shed light on co-design of EE and SE. Based on this, the precoding matrices, the transmit power and the active transmit antenna sets are jointly optimized at the BS to improve the EE-SE trade-off in MU-MIMO systems. The main contributions of this theisis are as follows:1) A CFA of the EE-SE trade-off is derived for multi-antenna random beamforming system according to a reversible asymptotic approximation of the SE obtained with the help of extreme value theory. Based on the approximate EE-SE trade-off, the EE optimization problems are solved efficiently with the help of Lambert W function and the optimized EE-SE curve is characterized. It is not easy to directly derive the optimal power for maximizing EE, because the rate part of EE as a function of the transmit power is very complex. Based on the CFA, the EE optimization problems can be solved efficiently. A CFA is also derived to characterize the relationshiop between the EE and individual user SE with heterogeneous users considered. Simulation results validate the accuracy of our proposed closed-form approximation.2) The EE optimization problems are considered for block diagonalization (BD) based MU-MIMO systems with non-ideal power amplifier (PA) models and per-antenna power constraints considered. In the problem formulation, a joint optimization of the transmit covariance and the active transmit antenna set at the BS is considered. Through reformulating the original problem as a sparse beamforming design problem and using successive convex approximation method, this thesis proposes an iterative algorithm to solve the EE optimization problem. The EE-SE trade-off under the non-ideal PA model is also studied for the considered system. Through solving EE maximization problems under different SE values, the optimal relationship between EE and SE can be revealed. Simulation results shows that the SE corresponding to the optimal EE under non-ideal PAs is larger than the one under ideal PAs. If the system parameters are designed according to the optimal SE under an ideal PA model to maximize the EE, both EE and SE of the realistic system will decrease. Another interesting observation is that applying antenna selection algorithm can help improve the EE-SE trade-off.3) This thesis takes the multi-cell multiple-input single-output (MISO) system as an example to study the EE-SE trade-off in multi-cell MU-MIMO systems. The Pareto boundary of the jointly achievable region of EE and SE reflects the performance limits and also characterizes the EE-SE trade-off of the multi-cell MU-MIMO systems. The Pareto boundary of the achievable EE region in multi-cell MISO systems is first studied, and the structure of the coordinated beamforming vectors is derived to achieve the EE Pareto boundary. Based on this, a decentralized algorithm is further developed to implement the multicell coordinated beamforming achieving the EE Pareto boundary, which is verified by simulations results. Finally, the Pareto boundary of the jointly achievable region of EE and SE is studied for the multi-cell MISO systems. Equivalently, the EE Pareto boundary with multiple SE constraints is characterized, and then the structure of the optimal coordinated beamforming vectors is derived.