THESIS FOR THE DEGREE OF MASTER OF SCIENCE Nonlinear Compensation and Frequency Domain Equalization for Wireless Communication

i ACKNOWLEDGEMENT In writing this master thesis, we would like to gratefully acknowledge the help and support received from teachers and friends, both past and present, without whom this paper would have been a much poorer offering. First of all, our gratitude goes out to Ali Behravan, our thesis supervisor. He gave us a lot valuable suggestions and advice and spent his time giving us some tutorial when he was very busy with his own work, which made our work going smoothly. Thank Prof. Thomas Eriksson for giving us a chance to work in such an interesting field and help us to get a room to work. Chen would like to thank Prof. Erik Ström and Arne Svensson, whose lectures are of great help to the theoretical part of the thesis. In particular, Libo would like to thank Prof. Piotr Starski, Prof. Joakim Johansson and all of his lecturers for their helpful teaching at Chalmers. ii ABSTRACT OFDM arises to take the place of single carrier, when high data rate is required in broad-band communication. OFDM splits the data into multi-carriers so as to decrease the data rate on each sub-carrier, which avoids severe problem of ISI existing in SC. However, OFDM also faces another distortion problem introduced by nonlinear amplifier and large PAPR. Several compensation methods such as predistorter and nonlinear equalizer are investigated, and frequency domain equalizer (FDE) is added at the receiver to invert the fading channel. In deduction and simulation, Saleh model is used to approximate the nonlinear amplifier and Clarke's model is taken to simulate the Rayleigh Fading Channel. For the predistorter, Least-Mean-Square (LMS) algorithm is used to update the coefficients of predistorter, and here, 5th order polynomials are used. For the nonlinear equalizer (time domain equalizer), Recursive-Least-Square (RLS) algorithm, which is a fast convergence algorithm compared with others, is used to get the coefficients. Several approaches to implement the FDE are discussed and simulated: Zero-Forcing equalizer, LMS equalizer, RLS equalizer, ZF estimator and RLS estimator. After comparison, ZF algorithm has the best convergence speed and SNR performance, and RLS algorithm has almost the same performance only a little worse. Coefficients got using LMS algorithm can't converge well, so LMS is not practical for it requires a long training sequence. On the other hand, the application of FDE makes single carrier modulation a potentially valuable alternative to OFDM. In SC system, different FDE methods mentioned …