16.01.2013 Views

Journal of Computers - Academy Publisher

Journal of Computers - Academy Publisher

Journal of Computers - Academy Publisher

SHOW MORE
SHOW LESS

Create successful ePaper yourself

Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.

1976 JOURNAL OF COMPUTERS, VOL. 6, NO. 9, SEPTEMBER 2011<br />

A New Method <strong>of</strong> Detecting Multi-component<br />

LFM Signals Based on Blind Signal Processing<br />

1 Qiang Guo<br />

1 Space Control and Inertial Technology Research Center,Harbin Institute <strong>of</strong> Technology, P.R. China<br />

Email: guoqiang292004@163.com<br />

2 Yajun Li and 1 Changhong Wang<br />

2 College <strong>of</strong> Information and Communication Engineering, Harbin Engineering University,P.R.China<br />

Email: liyajun1985happy@163.com<br />

Abstract—To effectively detect and recognize multicomponent<br />

Linear Frequency-Modulated (LFM) emitter<br />

signals, a multi-component LFM emitter signal analysis<br />

method based on the complex Independent Component<br />

Analysis(ICA) which was combined with the Fractional<br />

Fourier Transform(FRFT) was proposed. The idea which<br />

was adopted to this method was the time-domain separation<br />

and then time-frequency analysis, and in the low SNR cases,<br />

the problem which is generally plagued by noised <strong>of</strong> feature<br />

extraction <strong>of</strong> multi-component LFM signal based on FRFT<br />

is overcame. Compared to the traditional method <strong>of</strong> timefrequency<br />

analysis, the computer simulation results show<br />

that the proposed method for the multi-component LFM<br />

signals separation and feature extraction was better.<br />

Index Terms—multi-component LFM emitter signals, timefrequency<br />

analysis, feature extraction,ICA<br />

I. INTRODUCTION<br />

Radar emitter signal detection is a key problem which<br />

is demanded to be resolved in modern electronic<br />

reconnaissance system. With large new complex radar<br />

systems in practice, a large number <strong>of</strong> pulses overlap and<br />

form the multi-component emitter signals(MCES)[1].<br />

MCES analysis is a prerequisite and primary task for<br />

detecting and identifying emitter signals. Multicomponent<br />

Linear Frequency Modulated (LFM) emitter<br />

signals is a non-stationary signal which is commonly<br />

used in active sonar, radar imaging, fuse <strong>of</strong> missile and so<br />

on. As a new time-frequency analysis tool, FRFT is a<br />

generalization <strong>of</strong> the Fourier transform (FT). It not only<br />

has a natural link in classical FT, but also provides some<br />

characteristics which FT do not have. So FRFT is<br />

specially suitable for processing LFM class (chirp-like)<br />

signal. At present, regardless <strong>of</strong> the traditional parameter<br />

estimation or detection methods <strong>of</strong> multi-component<br />

LFM signal, most <strong>of</strong> them are based on time-frequency<br />

analysis or all finds <strong>of</strong> FT method[2]. The parameter<br />

estimation methods mainly through two-dimensional<br />

Manuscript received January 2, 2011; revised February 1, 2011;<br />

accepted February 28, 2011.<br />

Qiang Guo, Yajun Li,Changhong Wang.<br />

© 2011 ACADEMY PUBLISHER<br />

doi:10.4304/jcp.6.9.1976-1982<br />

object function, and combined with two-dimensional<br />

search to estimate, such as the maximum likelihood<br />

method, the time-frequency analysis methods, FRFT,<br />

match fourier transform(DCFT)[3,4] ,S-method and so<br />

on. But when the low SNR, together with the existence <strong>of</strong><br />

weak signal, the traditional detection method is <strong>of</strong>ten<br />

difficult to effectively detect the MCES, even lead to<br />

misjudgment <strong>of</strong> signal and noise ,so the result <strong>of</strong> crossterm<br />

suppression is not good. Above-mentioned issue has<br />

been a relatively difficult problem. In this paper, a multicomponent<br />

LFM emitter signal analysis method based on<br />

complex FastICA which was combined with FRFT was<br />

proposed. Firstly, complex ICA algorithm was used as<br />

time-domain separation for multi-component LFM<br />

emitter signals with noise. Secondly, determine the signal<br />

and noise by the automatic identification method <strong>of</strong><br />

second central moment <strong>of</strong> FRFT. Lastly, the noise was<br />

removed and the LFM signals were detected by FRFT.<br />

The effect <strong>of</strong> noise can be greatly reduced.<br />

Simultaneously, the cross-terms are effectively deduced<br />

with higher time-frequency resolution. It is a good<br />

method for the multi-component LFM signals.<br />

Simulation results verify the effectiveness <strong>of</strong> this new<br />

method.<br />

II. MODEL OF MULTI-COMPONENT LFM EMITTER SIGNALS<br />

In modern electronic reconnaissance system, receiver<br />

<strong>of</strong>ten intercepted to pulses which emitted by multiple<br />

sources at the same time. A stream <strong>of</strong> pulses was formed<br />

through these pulses interleaved together. As the pulse<br />

signals density increases, the pulse formed MCES<br />

x() t ,the signal model is expressed as follows.<br />

k −1<br />

∑<br />

i=<br />

0<br />

i<br />

2<br />

j2 π ( fit+ ( µ it<br />

/2)) , −∆t/2 ≤ t ≤ ∆ t/2<br />

x() t = Ae + n() t<br />

(1)<br />

Where Ai is the amplitude <strong>of</strong> each signal, fi is initial<br />

frequency and µ i is chirp rate. nG() t is White Gauss<br />

Noise with zero mean and variance 2<br />

σ .<br />

III. MULTI-COMPONENT LFM EMITTER SIGNALS<br />

ANALYSIS

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!