الفهرس 
IntroductionOnly 14 pages are availabe for public view
 |  | from | 143 |  |  |
| | | from | 143 | | |
Abstract
Communication networks employing multiuser detection undergo a revolution. The need for efficient use of channel bandwidth and transmission power is felt most acutely in wireless communications . Multiuser system employing code-division multiple access (CDMA) technique has recently received a considerable attention . In CDMA system , several packets of information are transmitted simultaneously over a common channel using preassigned code sequences . The transmitter could send the user signal in a synchronous or in an asynchronous fashion depending on the time offsets between user signals . In this thesis , the problem of asynchronous CDMA signals detection in the presence of additive white Gaussian noise (AWGN) will be considered . Therefore , several linear detectors depending on different blind adaptive algorithms and nonlinear detector employing neural network will be explained . The performance of adaptive Minimum-Mean Square Error (MMSE) blind algorithms such as Minimum Output Energy (MOE) , Constrained Minimum Output Energy (CMOE) , Griffiths Algorithm (GA) , Variable Step Size Griffiths Algorithm (VSS-GA) , Model-Based Griffiths Algorithm (MBGA) , And Variable Step Size Model-Based Griffiths Algorithm (VSS-MBGA) are evaluated using computer simulation , since the probabilistic models for such asynchronous adaptive detectors are intractable . Learning characteristics as well as performance comparison for the different algorithms will be considered in details . The results show that the VSS-MBGA is capable of achieving a considerable improvement over other algorithms , especially in case of time delay mismatch at the receiver side . The performance of neural networks detector based on Self Organized Mapping (SOM) adaptive algorithm is also compared with the adaptive MMSE blind algorithms . The neural network detector shows its superiority over the adaptive MMSE algorithms . However , its computation complexity grows with the number of users .