現代戰爭趨勢除了講求精良的武器設備外，精密且迅速的資訊更新系統亦是重要的一環，因此電子戰著實為成為戰場中的重要手段，為了有效針對電磁波訊號進行分類、辨識、定位與分析，進而獲得不同目標的信號特徵，使得武器系統能夠發揮最大功效，本論文利用多種訊號處理方法以增強及分析期望目標的訊號。首先以獨立成份分析法(Independent Component Analysis，ICA)作為訊號處理的基礎，得以將期望目標與其他干擾訊號分離。獨立成份分析被廣泛的應用在解決未知訊號分離的問題上，本篇論文係利用各目標之雷達訊號彼此為統計獨立的特性，將各訊號源予以分離，並經由訊號的時間、頻率、空間特性，選擇適當的獨立成份(Independent Component)進行訊號重建，以達到分離並濾除背景雜訊的目的。再以線性限制最小變異空間濾波器(Linearly Constrained Minimum Variance，LCMV)空間濾波器，針對期望目標行最佳化偵測，線性限制最小變異空間濾波器則基於多通道陣列雷達對於單一方向波源會有最小濾波總能量(Minimum Covariance)的假設，能夠依照波源訊號的特性，自適性的計算出最佳的通道權重(Channel Weight)，以達成最佳化的波束寬度(Beam Width)，萃取出所需要的訊號。最後以希爾伯黃轉換(Hilbert-Huang Transform，HHT)針對信號特徵進行分析，希爾伯黃轉換對波束構成濾波器所萃取的訊號進行時頻特徵分析(Time-frequency Analysis)，利用希爾伯黃轉換的可適性訊號處理特性，進一步將待測物的雷達訊號分解為許多具有不同時頻特徵的經驗模態函數(Intrinsic Mode Function， IMF)，並建立資料庫予以分類整理。; The trends of modern warfare are not just focuses on sophisticated weapon equipments, but also on precise and rapid information update system. Therefore, electronic warfare really is to become an important means of battlefield. In order to effectively obtain electromagnetic signals for classification, identification, location and analysis, and then get different target signal characteristics, making the weapon system to maximize efficiency. This paper uses a variety of signal processing methods to enhance and analyze the desired target signal. First, an Independent Component Analysis (ICA), which has been widely used in solving the issue of the unknown signal separation, as the basis for signal processing, to separate the desired target from other interfering signals. This paper separates the multi-channel signals into a series of Independent Components (IC) in accordance with their statistical independence, and the target-related signals can then be reconstructed according to their temporal, spectral and spatial features, in order to achieve the purpose of noise removal. Then filter the desired target by a Linearly Constrained Minimum Variance (LCMV) spatial filter to optimize the signal detection. The LCMV adaptively filters the received signals based on the assumption of minimum variance achieved by unidirectional source. Since the Signal-to-Noise Ratio (SNR) has been recognized as an important parameter in designing an adaptive beam forming filter, well separation of the target signals and noises will be helpful to improve the performance of LCMV. Last, the Hilbert-Huang Transform (HHT) is then applied to resolve the temporal-frequency features of the extracted signals. The extracted target-related signals are analyzed using HHT to decompose the LCMV output signal into a series of Intrinsic Mode Functions (IMF). A database will be further established to facilitate the recognition of detected targets based on their temporal-frequency features.