Estimates, Based On Pseudo-random Code Dsss Signal Of The Neural Network

The estimation of PN code of DS/SS signals is a hot research field recently.There are a few methods for estimating the PN(pseudo-random) code of DS/SS(direct sequence spread spectrum) signals, such as eigenanalysis and phase adding methods. There are also neural networks and adaptive FI R filter methods, which were proposed recently based on the eigenanalysis method.Blind estimation methods of the PN sequence of the DS/SS signal based on the neural networks method are mainly studied in this dissertation for the parallel computation ability of artificial neural networks. The methods are analyzed theoretically and can be used to estimate the PN sequence under a low S/N ratio situation through computer simulation. The main contributions are as follows:1. The eigenanalysis method is introduced and simulated. Because of the phase indetermination of this method, two improved methods, wide-window eigenanalysis and blind-synchronization eigenanalysis, are introduced. The computer simulation showsthat these two improved methods are performed well in estimating the PN sequence.2. A neural network using CHA is introduced and analyzed. It's proved that this method can be used to estimate the PN code of DS/SS signals by computer simulations. Aslo a kind of neural network using GHA is analyzed and this type of neural network's synaptic weights will converge to the PN code after iterative computation.3. A neural network using APEX algorithm are used to estimate the PN code. This type of neural network has both feed-forward connections and feedback connections which make the algorithm more stable. By introducing the generalized eigen-value decomposition of a matrix, a generalized PCA method is introduced. Then, a class of linear PCA neural network can be drawed from this GPCA method, which also can be used to estimate the PN code theoretically.4. A neural network using error back-propagation algorithm is analyzed to estimate...