Dissertation > Excellent graduate degree dissertation topics show
Research on Target Signal Separation Technology
Author: ShiWenBin
Tutor: WangDaMing
School: PLA Information Engineering University
Course: Military Communication
Keywords: wavelet denoising MDL estimate improved whitening PSOCF target signalextraction
CLC: TN911.7
Type: Master's thesis
Year: 2012
Downloads: 52
Quote: 0
Read: Download Dissertation
Abstract
It is very difficult to obtain the target signal due to the lack of prior information such assource signal and channel in noncooperative communication. The blind signal separation(BSS),which only use mixed observed signal to separate and reconstruct target signal, has become oneof effective solution methods on target signal separation in noncooperative communicationbecause of its independence on source signal. Therefore, aiming at how to achieve target signalseparation efficiency in noncooperative communication, the main contents of the paper are asfollows:1. Considering the poor performance caused by signal with noise blind separation, thispaper proposed an adaptive blind signal separation algorithm based on JADE and waveletdenoising. The algorithm introduced the wavelet denoising and consists of observed signalspredenoising and separated source signal postdenoising, in order to use the differentadaptability of these two styles according to different Gaussian source, added SNRpreestimation and realized the adaptive selection, therefore improved separation algorithmadaptability to different SNR.2. Without the knowledge of the amount of source signal, due to the inaccurately estimatingthe number of source signals through PCA, so the performance of signal separation wasdegraded. The improved whitening MDLFastICA algorithm was proposed aimed at solving thisproblem. The algorithm estimated the number of source signal by the MDL to accuratelyseparate the signal and noise subspace, and based on that, noise average variance was estimatedby the eigenvalue of signal and in turn modify the main eigenvalue of signal, suppress noiseeffect. Simulation result shows that amount of source signal is more accurate and separationperformance is improved obviously in the proposed algorithm compared with traditional FastICAalgorithm in different condition of Gaussian source.3. The probability density function estimation is foundation of many blind signal separationalgorithm, but the algorithm separation performance will be severely declined when selectingunsuitable activation function. Aiming at this problem, this paper proposed the blind signalseparation algorithm based on normalized fourthorder cumulant. Considering maximizing theabsolute value of normalized fourthorder cumulant as the objective function, the algorithmcould avoid the activation function selection. The PSOCF method was adopted to solve theoptimization problem, in which the whole situation optimum solution of whole situation and theprominent merit of low algorithm complexity could be acquired. Simulation shows that thesignal of superGaussian, subGaussian and mixed Gaussian is successfully separated in theproposed algorithm and the strong applicability, fast convergence speed and great separationperformance could be achieved.4. Aiming at the actual application requirement of extracting finite target signal frommultisource mixed signal, a limited blind target signal extract algorithm based on PSOCF wasproposed, in which the efficiency of blind signal separation algorithm based on normalized fourthorder cumulant could be further increased. By estimating error between separate signaland expectative signal, the reference independent component analysis was introduced to obtainthe evaluation of separate efficiency as the convergence indication of separate algorithm.inwhich the target signal separation was achieved by least time cost. Simulation result shows thatthe algorithm has great separation performance, low complexity and high run efficiency.

Related Dissertations
 Research on Electromagnetic Acoustic Transducer Detection System Based on FPGA,TH878.2
 Design of Testing Equipment for Electronic Products,TN06
 Research of Weak Heart Electric Signal Processing Based on Wavelet Transform,TN911.7
 Research on Fault Diagnose Method of Pipeline Leakage Based on Hopfield Neural Network,U178
 Highspeed Train Detection Technology between Pantograph and Catenary Current Collection Based on Traction System,TM922.5
 Battle tanks in tank bottom detection technology,TE972
 Thickness of plate electromagnetic ultrasonic detection and data processing technology research,TG115.28
 Research of Ultrawideband Radar Signal Processing Based on Wavelet Theory,TN957.51
 The Study of Cough Signal Detection Based on Gaussian Mixture Model,TN911.23
 The Key Technology Research on Signal Measurement of Measurement While Drilling System,TN911.23
 The Flaw Signal Extraction of Wire Rope Based on independent Component Analysis and Wavelet Transform,TG115.284
 Study on Modified Sagnac Interferometer’s Position Detection Signal’s Recognition and Processing Technology,TN929.11
 Research on Low SignalToNoise Ratio Signal Detection Based on NonLinear Theory,TN911.23
 Application of Wavelet Analysis in Digital Image Processing,TP391.41
 Research on Feature Extraction Technique in Fault Signal of Rolling Bearing,TH165.3
 Denoising method based on wavelet transform and independent component analysis study,TN911.4
 Design of Tool Condition Monitoring Signal Processing Platform Based on ARM & WinCE,TG11
 Feature Parameters Extraction of Metal Crack Acoustic Emission Signals Based on DSP,TG115.28
 Fiber grating fence alarm system and experimental study,TN253
 Millimeter wave detector signal processing system design and implementation,TP722.6
 The Design and Accuracy Analysis of Wide Frequency Range in Adjusting Speed Control System,TM921.5
CLC: > Industrial Technology > Radio electronics, telecommunications technology > Communicate > Communication theory > Signal processing
© 2012 www.DissertationTopic.Net Mobile
