Dual-channel Narrow-band Signal Center Frequency And Bandwidth Estimation

Center frequency and bandwidth are two significant parameters in signal processing, which catch our attention in engineering application and took a long period for researching. There are many kinds of algorithms for different situations and various signals, which have been applied widely in military, communication, industrial engineering and other fields. Narrow band signal is an important and familiar signal in practice. We have taken a lot of research for its spectrum characteristic, and got a great deal of pertinent methods. As the research of location developed deeply, the application of two channel and multi channel signal becomes popular. Because of their special form, we can utilize more information to estimate the parameters. Therefore, this dissertation will focus on the estimation methods for narrow band signal's center frequency and bandwidth. Then we will develop a new domain for two channel signal.At first, on the basis of searching and reading lots of related papers, this dissertation introduces the research background, internal and external actuality of frequency and bandwidth estimation, evaluates the performance of various estimation algorithms.Secondly, this dissertation introduces classical frequency and bandwidth estimation technology. We make a great deal of simulation practice, compare the computation and accuracy with each other, and analyse their performance based on the deeply understanding of algorithms'theory and applied situation. As the simulations show, these methods are not the optimal estimation algorithm for different situation.Thirdly, this dissertation proposes a modified ESPRIT frequency estimation algorithm for narrow band real signal, which estimates the center frequency on the basis of subspace-based parameter estimation technology and a novel data covariance model of narrow band real-valued signal. Then it is developed into two channel signal. Due to introducing the cross spectrum, the new algorithm could reject the influence of color noise. From a simulation study it is shown that the modified ESPRIT algorithm yields accurate estimates which gives computational advantage.Finally, this dissertation introduces phase information of two channel signal into bandwidth estimation, and proposes a novel method for combining the two signals. This method is not only suitable for the signals with white noise but also suitable for strong interference color noise and low signal-noise-ratio signals. Moreover, we improve the method. In the interest of reducing computation, FFT pruning is combined in the new method based on the characteristic of narrow band signal. Aimed at enhancing the accuracy, we introduce wavelet transform for searching the break points. Simulations show that the performance is pretty good, especially for signals with color noise and low signal-noise-ratio.