DOA Estimation Algorithms For Noncircular Emitter Signals

The Binary phase shift keying(BPSK) and Mary amplitude shift keying(MASK) are widely used in radar systems, their elliptic covariance do not equal zero, are noncircular signals. Improve the performance of DOA algorithms by utilising covariance matrix and elliptic covariance matrix of noncircular signals is a prosperous area. The tradition DOA algorithms and DOA algorithms of noncircular signals need eigenvalue or singular value decomposition, there are large amount of calculation. Therefore, several algorithms are researched to analyze the noncircularity of emitter signals and reduce the computational complexity. The main research work and contributions are detailed as follows.1. The noncircularity of emitter signals are anaysed, and elaborated the basic theory of sources number estimation algorithms. A new EDC algorithm for real-value signals, called REDC, is presented in this paper. Simulation experiments show that REDC can handle more sources than sensors, and is able to improve the performance of sources number estimation.2. Researched the MUSIC algorithm and spatial smoothing techniques. The MUSIC algorithm does not use the noncircularity of signals, and can't estimate the coherent signals. A new conjugate MUSIC algorithm, called CMUSIC, is presented in this paper. And combined the thoughts of spatial smoothing techniques, a novel real-valued spatial smoothing algorithm is presented. Simulation experiments show:CMUSIC is able to handle more sources than sensors, can estimate the coherent signal source'DOA, and have better angular precision than MUSIC; RSS can increase the dimensions of covariance matrix, also increase the amount of calculation, but inobtains higher estimation successful rates and better angular precision than the spatial smoothing techniques.3. Elaborated the basic theory of ESPRIT algorithms. Using the features of noncircular signals. A new conjugate ESPRIT algorithm, called CESPRIT, is presented in this paper. Simulation experiments show that CESPRIT also increase the calculations, but is able to handle more sources than sensors, and its resolution and angular precision are better than EPSRIT algorithm.4. To reduce the computational complexity of DOA estimation, a new algorithm with rotational invariance propagator for noncircular signals, called NCRIP, is presented in this paper. Simulation experiments show that its performance of DOA estimation is better than EPM, approaching the performance of NC-ESPRIT, and can handle more sources than sensors. NCRIP not only needn't eigenvalue or singular value decomposition, but also needn't search the peak of spectrum, greatly reduce the computational complexity. This work was supported by the National Natural Science Foundation of China (60702026) and the Scientific and Technological Funds for Young Scientists of Sichuan (09ZQ026-040).