الفهرس 
IntroductionOnly 14 pages are availabe for public view
 |  | from | 239 |  |  |
| | | from | 239 | | |
Abstract
This thesis is divided into four parts. The first part provides the foundation to the rest of thesis. It provides the fundamentals of DS/CDMA system and formulates the multiuser detection problem. In addition, a generalized DS/CDMA system model which accounts for asynchronism, multipath propagation, near-far effect, signature mismatch, and inter-symbol-interference (ISI) is adopted. Part two of this thesis is concerned with the adaptive blind multiuser receivers based on the constrained optimization approach with multiple constraints and inverse QR decomposition (IQRD). The IQRD-based recursive least-squares (RLS) algorithm (IQRD-RLS) is very popular because, it has good numerical stability and can be mapped onto Coordinate Rotation Digital Computer (CORDIC) processorbased systolic arrays, which is suitable for very large scale integrated circuits (VLSI) architecture and real-time implementation. The method of linearly constrained IQRD was borrowed from the area of array processing and beamforming. In IQRD updating method, the least squares weight vector can be calculated without back-substitution which gives numerical stability to such algorithms. In this thesis, the blind minimum output energy (MOE) detector is implemented using linearly constrained IQRD method. Three different algorithms are introduced: the first is a direct form MOE detector based on IQRD-RLS with fixed constraints, the second is MOE implemented using partition linear interference canceller (PLIC) structure and IQRD-RLS algorithm, and the third is an optimal MOE algorithm based on combined IQRD update method and subspace tracking algorithm for tracking the channel vector. Improvement will be added to the third receiver by implementing a fast subspace tracking algorithms for tracking the channel vector based also on IQRD algorithm. Moreover, a new fast minor subspace tracking algorithm is proposed to track the channel vector. In addition, a comparative analysis between channel estimation techniques anchored on the third algorithm is performed. Furthermore, a new approach is proposed to handle noise enhancement accompanying max/min approach at low SNR using quadratic inequality constraint. A new simple solution for the QI constraint is introduced. The possibility of systolic array implementations for the proposed algorithms is introduced and a new systolic implementation for the third receiver is proposed. Finally, the performance of the proposed schemes is compared and analyzed in terms of BER, SINR, and computational complexity. Simulations are presented in a rich multipath environment with near-far effect to demonstrate the potential and fidelity of the proposed receivers. 7 In part three, a novel low-computational complexity robust adaptive blind multiuser receiver based on constrained optimization approach with multiple constraints and QI constraint on the weight vector norm is developed. Quadratic constraints have been a widespread approach to improve the robustness against mismatch errors, uncertainties in estimating the data autocovariance matrix, and random perturbations in detector parameters. A diagonal loading technique is mandatory to apply the quadratic constraint where the diagonal loading level is amended to fulfill the quadratic inequality constraint value. The Lagrange multiplier methodology is exploited to solve the QI constraint problem. A new optimal Variable Loading (VL) technique which is capable of providing robust control against uncertainties and mismatch errors with low computational complexity is adopted. In addition, the diagonal loading term is precisely computed using a simple quadratic equation. Integrating the quadratic constraint into adaptive update schemes such as least-mean-square (LMS) or RLS seems to be a moot point since there is no closed-form solution for the diagonal loading term. In this thesis, the MOE detector is implemented using a fast recursive steepest descent (RSD) adaptive algorithm based on the PLIC structure with multiple constraints and a QI constraint on the weight vector norm. The proposed RSD algorithm is adopted to avoid matrix inversion required by RLS algorithm. Additionally, an analogous to the robust MOE detector, a new robust blind receiver based on linearly constrained constant modulus algorithm (LCCMA) with multiple constraints and QI constraint on the weight vector norm is developed. Furthermore, a robust blind multiuser receiver based on block-Shanno constant modulus algorithm (BSCMA) and QI constraint on the weight vector norm is adopted in this thesis as well. The BSCMA is realized using a modified Newton’s algorithm without inverse of Hessian matrix estimation. The PLIC structure with multiple constraints is invoked to identify the multiple access interference (MAI). Simulations are presented in a loaded multipath environment with near-far effect and/or signature mismatch to demonstrate the robustness of the proposed detectors. It is shown from the computer simulations that the proposed robust detectors can efficiently handle the near-far effect under no restrictions. Additionally, a comparative analysis between constant modulus based receivers is conducted. The use of antenna array offers the possibility of utilizing the spatial characteristics of different user signals to augment the temporal discrimination provided by their signature sequences which lead us to the spatial-temporal processing. Therefore, in order to provide complete interference cancellation strategy, we have considered beamforming techniques in Part four of this thesis to suppress unsolicited interference. Recently, there has been an enormous 8 endeavoring to design robust adaptive beamforming algorithms to improve robustness against uncertainties in steering vector. These uncertainties may be caused by uncertainty in direction-of-arrival (DOA), imperfect array calibration, near-far effect, and other mismatch and modeling errors. A diagonal loading technique is obligatory to fulfill the uncertainty constraint where the diagonal loading level is adjusted to satisfy the constrained value. The foremost drawback of diagonal loading technique is that it is not clear how to get the optimal value of diagonal loading level based on the recognized level of uncertainty constraint. The contribution of Part III in this thesis is two fold. We firstly propose an alternative realization of the robust adaptive linearly constrained minimum variance (LCMV) beamforming which includes ellipsoidal uncertainty constraint on the steering vector. More specifically, the genuine steering vector is estimated from the presumed steering vector using steepest descent (SD) or conjugate gradient (CG) algorithms with the ellipsoidal constraint imposed on the estimated steering vector. The diagonal loading technique is integrated into the adaptive update schemes by means of variable loading technique rather than always diagonal loading or ad hoc techniques. We secondly enrich the proposed robust adaptive beamformer by imposing a quadratic constraint on the weight vector to improve robustness against random perturbations in detector parameters and noise enhancement at low SNR. Several numerical simulations with DOA mismatch and moving interference are carried out to explore the performance of the proposed schemes and compare its performance with other traditional and robust beamformers. Index Terms—mobile communication, wireless communications, DS/CDMA, multiuser detection, multiple access interference, multipath propagation, frequency selective, minimum output energy detection, constant modulus algorithm, Block- Shanno, Newton’s algorithm, inverse QR-decomposition, systolic arrays, subspace tracking, constrained optimization, Lagrange method, diagonal loading, quadratic inequality constraint, ellipsoidal constraint, Capon beamforming, LCMV beamforming, moving jamming, space-time processing.