Optimization Of Multi-Carrier Systems With Nonliear Distortion

Nonlinear distortion caused by High Power Amplifier is an inevitable issue in any practical wireless communication systems. It will introduce in-band distortion and out-band interference, and lead to system performance degradation. For multicarrier systems, this problem becomes more serious because the PAPR (Peak-to-Average Power Rate) of multicarrier systems is much higher comparing to single carrier systems. This thesis focuses on the optimization of multi-carrier systems with the effect of PA nonlinearity.To have a more comprehensive understanding to the subject, at the beginning of this thesis, we reviewed the research status of nonlinear distortion optimization methods for multicarrier systems in recent years, analyzed the existing problems, summarized research directions which worth further study and obtained a series of results.First, we analyzed measure criterion of receive signal quality of multicarrier systems with nonlinear distortion. This is the key difference between nonlinear distortion environment and normal ideal environment, therefore is also the theoretical basis of the whole thesis. By analyzing the relationship between SNDR (Signal to Noise plus Distortion Rate) and SER (Symbol Error Rate), we learned that instead of SNR (Signal to Noise Rate), SNDR can be used to evaluate signal quality of multicarrier systems with nonlinear distortion.Based on the above analyze, SNDR is used to deriving system capacity expression of the simplest nonlinear distortion environment: single antenna system. It can be found from the capacity expression that different from ideal environment, the increasing of PA input power will not only cause the increasing of useful signal power, but also bring on the increasing of nonlinear distortion power. When the power of PA input signal is large enough, signal distortion becomes not neglectable and may worsen system capacity. Therefore, there exists an optimal PA input power which maximizes the system capacity.Then we derived the expression of this optimal PA input power using capacity expression obtained above. But this expression is too complicated to be used in a practical system. To solve this problem, we proposed two approximate numerical methods: look up table and curve fitting. These two methods have advantages and disadvantages respectively, and can be chosen according to specific scenarios. Numerical simulation proved that optimal PA input power can be found by using these methods and capacity can be maximized.After that, we studied how to extend the above conclusions to MIMO (Multi-Input Multi-Output) systems. Similarly, we first derived the system capacity expression of MIMO systems with nonlinear distortion, and then studied power allocation problems under this environment. Unlike single antenna systems, the aim of power allocation for MIMO systems is not to optimize the capacity of one single antenna, but to optimize the sum of all antennas under total power constraints. By using the new capacity expression, we first modified traditional greedy algorithm to be suitable in nonlinear distortion environment. Then, we model power allocation problems as a standard optimization problem and obtained its optimal solution by using convex optimization theory. Simulation results confirm that the new method can better reduce signal distortion and maximize system capacity.Unfortunately, the same problem exists in the above solution: it is too complicated to be used in a practical system. To solve this problem, we combined the optimal algorithm with traditional water-filling algorithm, and found a new suboptimal algorithm -- cuvette-filling algorithm. The new algorithm has the advantage of both low complexity and high performance, and is suitable for practical use.At last, we turned our attention to nonlinear distortion suppression and compensation. Nonlinear distortion suppression is mainly used in transmitter before signal enters PA, while nonlinear distortion compensation is mainly used in receiver after the distortion is generated. There are two main methods to suppress non-linear distortion at the transmitter: minimum the probability of high PAPR and minimum the affects of distortion. The relationship between these two methods is studied and the equivalency of minimum PAPR probability and minimum distortion is proved. Besides, a class of new non-linear distortion suppression algorithm is proposed based on maximum SNDR criterion. Simulation results also show that distortion suppression methods using maximum SNDR criterion and minimum PAPR criterion have the same performance.In the aspect of nonlinear distortion compensation at the receiver, a new distortion cancellation method for OFDM systems is presented in this thesis. Nonlinear distortion is modeled as parabolic arcs with random parameters and is estimated in frequency domain. The estimated distortion is then used to compensate the real distortion, thus the performance of OFDM system is improved. Simulation results also approved that the performance of proposed method improved about 2 dB in contrast to the traditional one.