Compensation of nonlinear distortion and frequency offset in OFDM systems

Orthogonal frequency division multiplexing (OFDM) systems exhibit many advantages in wireless as well as wireline environments. However, two crucial limiting issues should be overcomed for applicaiton of OFDM scheme to wireless systems: sensitivities to nonlinear distortion and frequency offset. This dissertation deals with applications of compensation schemes in OFDM systems to solve the sensitivity problems associated with nonlinear distortion and frequency offset in the systems. With regard to compensation of nonlinear distortion in OFDM systems, a Volterra-based predistorter has been applied to transmitter of OFDM systems. As a minimum mean square error (MMSE) predistorter, the Volterrabased predistorter can offer global solution with a proper learning architecture. The previous learning architecture didn’t make an efficient training for the predistorter because the learning algorithm utilized only indirect structure. For more efficient training, we propose a new learning algorithm which utilize direct as well as indirect structures in this dissertation. Our simulation shows that the new learning algorithm achieves a gain of 5 dB or more over the previous learning scheme. In this dissertation we also simplify the structure of the Volterra-based predistorter. Our simulation shows that the number of the predistorter coefficients can be reduced by 52.4% using the simplified scheme. Since the structure of the predistorter is efficiently simplified, the predistorter exhibits more rapid convergence rate that the original Volterra-based predistorter in training the coefficients. We also present a compensation scheme of inter-carrier interference (ICI) due to frequency offset in OFDM systems. For efficient cancellation of the ICI effects, we propose a nonlinear adaptive filter. In addition, we investigate the performance of the nonlinear adaptive filter when the ICI effects are magnified by some residual nonlinear distortion. Moreover, we present a blind estimation scheme of frequency offset based on the nonlinear adaptive filter for channel efficiency. Our simulation results confirm that the nonlinear adaptive filter is superior to the conventional linear filter in canceling the ICI effects and especially more suitable for the blind estimation of frequency offset under a nonlinear channel than the linear filter. Finally, we propose a new estimation scheme oftiming offset as well as frequency offset in OFDM systems. For complexity reduction we separately estimate timing and frequency offsets. For estimation of frequency offset we utilize an ambiguity function. In addition, for estimation of timing offset we propose a standard mimization method. Our simulation results show that the offsets are separately and exactly estimated using our presented schemes.