Multi-parameter Estimation Technology Of Wide-band Array Signals Based On High-order Cumulant

Estimating parameters of wide-band array signals acts as an important role in modem array signal processing. However, there are a lot of difficulties in the real time application of wide-band array system, such as: I) Sub-Nyquist sampling is necessary in the wide-band array system to avoid sampling multi-channel array signals synchronously with Nyquist rate. However, M delayed channels are needed to resolve the frequency aliasing caused by undersampling in a traditional wide-band array system with M sensors. This means large additional hardware cost and can cause some other problems. 2) the performance of a practical wide-band array system can be seriously degraded by the errors in the model, mainly resulting from sensors positioning errors, mutual coupling among sensors, mismatching among the array sensors and mismatching among receiving channels. 3) The complex signal environment which includes colored noise and coherent signals can also degrade the p~formance of algorithm, etc. Aiming at melioration of the above problems~ the author studied most keypoints in a wide-band array system used to estimate the frequencies and 2D-DOA signals based on 4~-order cumulants in this dissertation. The main contributions and conclusions of the dissertation are in the several aspects as follows: 1. The capability of array extending, the geometry and the properties of the extended based on 4d1~order cumulants were studied in the dissertation. The results show that the array can be extended in 2 aspects simultaneously. One is that more virtual sensors are achieved that break through the limitation of the number of signals that can be resolved by 2I2d~order algorithm. Another is that larger array aperture is achieved that improves the performance of the algorithm effectively. 2. An algorithm to estimate the frequencies and 2D-DOAs of wide-band array signals based on 4~?order cumulants was presented. Sub-Nyquist sampling is used in the algorithm and only one additional delayed channel is needed to resolve the frequency aliasing caused by the undersampliflg. Compared to the traditional 2nd~order algorithm, the novel algorithm can save hardware cost, meanwhile it makes multi-channel sampling synchronously and keeping channels matching relatively easy. 3. Traditionally, the calculation of cumulant matrix is a large burden for any algorithm based ?? on high-order cumulants. A DFT preprocessing method which involves FFT and a simple peak-picking process is proposed in the dissertation to decrease the calculation cost effectively. 4. An improved non-uniform spaced array interpolating algorithm was proposed to achieve better performance, then an algorithm based on 4w-order cumulants was presented to estimating frequencies and 2D-DQAs of coherent signals with non-uniform spaced array, which also includes a new interpolating sector selecting method. 5. The model errors of wide-band array system were analyzed and a model-error estimation algorithm was proposed to estimate the various kinds of primary model errors that can degrade the performance of a wide-band array system. With the estimated errors in hand, array calibration can be easily done.