Research On Estimation Of Time Delay And Related Parameters

This dissertation focuses on the time delay estimation problem which has various and important applications in the fields of radar, sonar, communications and so on. Corresponding to different application conditions, such as noise/interference properties, relative motions between transmitters, receivers and reflectors, absence/presence of multipath propagation, and known/unknown of the transmitted signal, different models are developed and analyzed in detail. Several theoretically and practically valued algorithms are proposed and derived. Computer simulations and their results are given to verify the efficiency of these proposed algorithms. The main contributions of this dissertation can be concluded as follows:(1) Approaches using second order, higher order, fractional lower order and cyclonstationary statistics to estimate time delay(with/without Doppler shift) in the absence of multipath are summarized. Their applicabilities, indispensabilities, and similarities between each other are carefully analyzed.(2) Based on the invariability of the cyclic frequency which belongs to the Doppler-shifted cyclonsationary signal and corresponding to the cyclic correlation function, a new algorithm is proposed, which can estimate time delay and Doppler shift parameters separatedly. Compared with the cyclic ambiguity function which estimates time delay and Doppler shift simultaneously, the proposed algorithm has much less computational complexity and almost equivalent estimation accuracy.(3) EM algorithm is used to implement the ML estimator of multipath time delay for known signals to supersede the multi-dimensional optimization by iterations of the one-dimensional optimization. Improvements in the scheme of itration initialization, parameter updating, and search range designation are also proposed for the enhancement of the convergence of the iteration and further reduction of the computational complexity.(4) Two efficient multipath parameters estimation algorithms (one for multipath time delay estimation, and the other for multipath time delay and Doppler shift joint estimation), which can only use the noise contaminated reference signals, are proposed. Both of them are borrowed from the idea of WRELAX algorithm, which...