الفهرس 
SummaryOnly 14 pages are availabe for public view
 |  | from | 119 |  |  |
| | | from | 119 | | |
Abstract
communications. Adaptive array techniques offer possible solutions to these interference problems because of their capabilities for automatic null steering in the directions of undesired signals while enhancing the required one.By an adaptive array is meant a device that is self-designing in the sense that it contains a set of weights that are automatically adjusted on the basis of estimated statistical characteristics of the revelant signals. In the adaptive array problem the requirement is to find an algorithm for adjusting the array weights, based upon a selected measure of optimality (e.g. minimum mean square error at array outpu).
Two popular adaptation algorithms are the least mean squared (LMS) algorithms and the sign algorithm (SA). The first contribution of the thesis is that it provides a comparative study of these algorithms based on both analytic arguments and comparison is that the LMS algorithm has the advantage of fast adaptation while the SA has the advantage of simple implementation. To make use of the technical simplicity of SA,it is desired to increase the speed of its adaptation. The thesis provides, second contribution, two propsals for speeding-up this adaptation. The first one is to update the step size of SA on the basis of the difference between the actual setting of the array and the optimum one. The greater is this difference, the greater will be the step size and is difference, the greater will be the step size and vice versa. Estimation of the actual distance from optimlity is based on the diference between the output of the array and a pilot signal. This proposal has been simulated on a digital computer where simulation results haveexhibited its efficacy.
The second proposal is dedicated for the case when the input signal of the array is weak. Insuch a case the convergence time of SA may be too long to be tolerable. To avoid this problem we propose the use of a rapid automatic gain control (AGC) unit at the input of the array. It should be mentioned that conventional AGCunits cannot be used for the present purpose due to the fact that they are based on long time averages of input signal and, therefor, they are slow. The proposed AGCunit is based on short time averages and consequently it is rapid. It should be mentioned that this rapidity is at the expense of accuracy of the gain correction obtained. However, the role of AGCin the present application is to avoid accurence of weak input levels rather than to perform accurate gain correction. Hence accuracy can be sacrificed for rapidity in this special AGC. Computer simulation results of the proposed SA with rapid AGC have shown its superiority to conventional SA.
The thesis is organized as follows. In chapter 1, a review of adaptive antennas arrays is given. The following items are discussed : adaptive sidelobe canceler, adaptive receiving array with a pilot signal, and methods of generation of the pilot signal. Chapter 2 is devoted to the study of LMS algorithm and SA and the proposals mentioned above. Then an analytic comparison of both algorithm is given. At the end of chapter 2 the two proposala for speeding up the adaptation of SA are presented. In chapter 3, computer simulation are given. This chapter gives and comments the following simulation results : dependince of both LMS algorithm and SA on step size and input signal level, performance of the proposed SA with adaptive step size and SA with rapid AGC and comparison of these SA`s with conventional SA.