Study On Interference Suppression For CDMA Systems In Non-Gaussian Alpha-Stable Distribution Environments

In conventional communication systems, receivers are usually designed in the presence of additive Gaussian noises. In this instance, the receivers are optimal. But the numerous man-made and natural noises in many communication channels are non-Gaussian and impulsive. Such non-Gaussianity causes the optimal receivers degenerate significantly. When the error involved by Gaussian assumption cannot be tolerated, more accurate noise models and reasonable processing systems have to be considered to avoid significant performance degradation.Among many non-Gaussian noise models, the symmetric alpha-stable (SaS) probability density functions can accurately model large classes of impulsive noises. The a -stable distribution with 0 < α < 2 does not exist finite second order statistics. Consequently, many standard signal processing tools (e.g. power spectrum and the least-square techniques) which are based on assumption of finite variance will be considerably weakened and may in fact give misleading results. The pth order moment (p < α) for α-stable distribution is finite and it can be used in adaptive filtering.Based on weighted accumulated least p-norm, a momentum LMP algorithm is proposed in this thesis; Based on weighted average least p-norm, another momentum LMP algorithm is proposed. The simulation results indicate that the performances of two momentum LMP algorithms proposed in this thesis are better than LMP's. Other advantage of the two momentum LMP algorithms is that computational complexity is the same as that of LMP algorithm.Conventional direct-sequence code-division multiple-access (DS/CDMA) systems use matched filter (MF) receivers. However, the MF receiver is subject to the near-far problem. The minimum mean squared error (MMSE) receivers have showed to be near-far resistant in additive Gaussian noise channels. The MMSE receivers can be replaced by Minimum dispersion (MD) receivers in additive non-Gaussian SaS noises. The limitation of adaptive MD receiver with LMP algorithm is that the algorithm convergences slowly. In this thesis, two new momentum LMP algorithms are used to calculate the approximate MD solution. The simulation results show that the performances of both momentum LMP algorithms are analogous to LMP's with a faster convergence rate than LMP algorithm.In practice, when the processing gain N is large, and the received energy per chip is low,rapid estimation of the weighted tap of the MD may be difficult. This motivates the simple interference suppression schemes, which contain fewer adaptive taps.