A highly accurate method for parameter estimation

The problems of signal detection and parameter estimation are of fundamental importance in detection and recognition of weak target. The typical problem is the measurement of the harmonic signal parameter in the noise. In traditional theory, the processing method is based on uniformly sampling and all the data have the same effect. This paper presents a novel theory to deal with the problem, which treats the data in a different way. The novel method is based on the sub-sampling technique. The position of the new sampling data is not regular in source sequence, but they have the same feature that they are affected by noise less seriously than those that are deserted. Since the new data is not uniformly sampled, the traditional estimation algorithm will be modified and a novel algorithm is discussed to deal with the sub-sampling data. The method presented in the paper offer the higher accuracy of the estimation, because the sub-sampling data have higher SNR. Computer simulations are used to obtain the parameter estimation performance results. Results obtained by this new method are compared with traditional FFT method, and the parameters obtained by new method are more accurate.