Studies On Parameter Estimation For Spatially Distributed Signal

In array signal processing, target signal source models often are based on the point sources, but the real target signals with the space distribution features are generally have more complex than the point sources, and the receiver signals in array could be equivalent of the spatially distributed sources. Distributed signal modeling is different from traditional point source, it not only contains the center angle of the target's information, but also has a additional spatial extension angle information, the traditional point signal model can not exactly describe observation data, and its estimation algorithm will make some center angle estimation performance, even can't obtain correct results. This paper work is based on distributed signal's model characteristic, and study on the methods to estimate center DOA.1. According to the characteristics of the distributed signal model, the paper researched some mathematical models which are about the coherent distributed source and incoherent distributed source. Because of accurate models are often complex, several the approximation models are introduced.2. Several algorithms for estimating the DOA of distributed signal are studied, such as DSPE algorithm, the generalized ESPRIT algorithm, beam-forming algorithm, etc. First described in detail of the theory of various algorithms, then, simulated several algorithms, finally to compare the performances of several algorithms.3. For non-ideal Gaussian white noise in the channel, the de-noising algorithm of the estimation the signals DOAs were studied. And different conditions performances were compared with the original algorithm.4. Because DSPE algorithm and GC algorithm in estimating DOA of distributed signal need exactly clear distribution form, the algorithms may be restricted the use range. In order to overcome the disadvantages, the paper proposes a kind of non-linear optimization algorithm, when distribution forms of the sources are unknown, utilizing the optimization results instead of unknown type function, it can obtain correct estimation results. Through simulating, the proposed algorithm is verified.