الفهرس 
IntroductionOnly 14 pages are availabe for public view
 |  | from | 107 |  |  |
| | | from | 107 | | |
Abstract
The thesis aims to improve the perfomance of blind equalizers used in bandwidth¬ efficient digital transmission that employ high-order QAM modulation, where a training sequence is not feasible or not desirable. Two categories of algorithms are discussed, namely, the algorithms explicitly based on Higher order statistics (HOS), and those implicitly based on HOS.
The first category is the family of algorithms that are explicitly based on HOS. The thesis presents a modified approach for enhancing the performance of the Eigenvector Approach (EV A) [34], based on the cross-cumulants criterion, for blindly equalizing non¬constant modulus signals. The EV A algorithm [3,4] depends only on the known statistical
properties of the data source. The modified approach presented in this thesis takes into account the extra information about the known constellation alphabet of the data source through utilizing the Decision Directed (DD) algorithm. It is shown with simulation that this extra information will dramatically enhance the performance for high-order signal constellations. Next, the use of blind space-time equalization based on the EV A [4] was suggested for use in wireless communication systems that suffer from fading multipath propagation. The extra information due to space processing was shown to improve the system performance for denser constellations that might be used in high-data rate applications.
The second category discussed is the family of algorithms that are implicitly based on HOS. Blind equalization using the concurrent operation of both the constant modulus algorithm (CMA) and the decision directed (DD) algorithm has been recently introduced [11]. This scheme outperforms the conventional CMA for high-order signal constellations. However, general filter design guidelines [12] were not taken into considerations. in [11]. In this thesis, we show through simulations that applying these filter design guidelines can enhance the performance of the equalizer, in terms of reducing computational requirements or increasing the convergence rate. Aiming to evaluate the performance of the equalizers in realistic situations, channel models obtained from experimental measurements were used in the simulations. Examples of these channels are cable TV channels and digital microwave radio channels, obtained from the database administered by Rice University and known as the Signal Processing and Information Base (SPIB) [20]. The digital satellite video broadcasting (DVB-S) channel is also used as a case study [36].