The Fast Method For Doa Estimation Of Wideband Signals And Its Implementation On Dsp

The direction-of-arrival(DOA) estimation is a very important research direction in array signal processing field, which finds wide applications in many fields, such as radar, communication, seism,sonar,etc. The conventional DOA estimation methods which is aim at narrowband signals can not estimat the DOA of wideband signals. At present, the methods for DOA estimation of wideband signals have complex process and tremendous computational capacity, which is difficult problem for realization of highresolution DOA estimation methods of wideband signals on practical application systems in real time, and also is a difficulty and a hotspot in DOA estimation in recent years.Compared with other methods for wideband signals DOA estimation, coherent signal subspace method(CSM) draw much more attention because it need little prior knowledge, has high precision and can deal with correlative signals. In order to reduce the computational complexity of CSM, this dissertation decides to put researchful emphasis on the three aspects of fast subspace estimating,optimizing the spatial spectrum peak searching and implemention on high speed DSP. The main work and results are summarized as follows:1. The theory of coherent signal subspace method(CSM) is studied. Implemention and computational capacity of each part of Two-side correlation transformation(TCT) algorithm is analyzed, which has the best estimating performance and does not need pre-DOA estimating. The work of this chapter establishs the foundation for this dissertation.2. The fast method for subspace estimation based on multi-stage wiener filter(MSWF) is studied. Subspace leakage while using MSWF for subspace estimation is studied, and P+2 multi-stages decomposition for subspace is given. According to the requirement of that Krylov subspace is equivalent to signal subspace, a fast method for subspace estimation based on MSWF without expected signal is proposed, Combine with TCT algorithm, a fast method for wideband signals subspace DOA estimation based on MSWF without expected signal is given.3. The fast method for subspace estimation based on principle components analysis(PCA) neural network is studied. The principle that principle components analysis is equivalent to signal subspace estimation is analyzed. The effect of learning rate on PCA network convergence is analyzed, and an self-adaptive learning rate is developed. A fast method for subspace estimation based on complex SGA criteria of PCA neural network is proposed, which avoids eigenvalue decomposition and reduces computational complexity. Combined with TCT algorithm, a fast method for wideband signals subspace DOA estimation based on PCA neural network is given. 4. The optimum technology for spatial spectrum peak searching based on Niche genetic algorithm(GA) is studied. To enhance the ability of multiple peaks searching for spatial spectrum function, combined with conventional beam-direction-finding technology, a novel searching method is proposed. A hybrid peak searching method which combines Niche GA with Hill-climbing method is proposed, which overcomes the shortage that Niche GA is lack of local searching and has slow speed at the late of convergence.5. The fast DOA estimation algorithms of wideband signals is realized in the high speed DSP. The DSP program developed in VisualDSP environment is designed, optimized and debugged in modularization. Running results demonstrate the effectiveness of fast algorithms proposed in this dissertation.