Research On Parameter Estimation For Digital Signals And FDOA Estimation

In this paper, the symbol rate estimation, carrier frequency offset(CFO) estimation, and frequency difference of arrival(FDOA) estimation are studied. Novel estimation are proposed to overcome the disadvantages of conventional estimation. The defect of conventional estimation were analysed and the novel estimation were researched to overcome the disadvantages of conventional estimation, with their performance simulated.There exist various methods to address the symbol rate estimation, such as cyclic-autocorrelationbased estimation, cyclic-spectrum-based estimation, and wavelet-transform-based estimation. However, these estimation require the priori knowledge of modulation schemes. And they are only adapted to limited types of modulation schemes. In this paper, a self-adapted symbol rate estimation based on optimal scale wavelet transform is proposed. Based on the conventional wavelet-transform-based estimation, the novel estimation employed the“hold-differentiation” process to eliminate the difference of the results of wavelet transform between different signals. The novel estimation is adaptable to various modulation schemes including amplitude shift keying(ASK), phase shift keying(PSK), quadrature amplitude modulation(QAM), and frequency shift modulation(FSK). Further, the optimal wavelet scale was chosen to avoid the “invalid wavelet scale” problem that exists in conventional estimation. Finally, the relationship between the sampling rate and the estimation performance was analyzed, and a choice of optimal sampling rate was obtained to resample the signal, which greatly improves the performance. The simulation results show that the novel estimation is effective to signals like ASK, PSK, QAM, and FSK. The performance of the novel estimation is superior to the optimal-scale-based estimation and the multi-scale-based estimation.In existing works, several methods have been proposed to estimate CFO, such as highcumulant-based scheme and non-linear-least-squire(NLLS). However, those estimation require the priori knowledge of modulation scheme. Thus, they are not blind. In this paper,we propose a blind CFO estimation for single carrier and OFDM signals using least cyclostationary order(LCO) cyclic moments. First, the concept of LCO was proposed, and the relationship between the LCO order cyclic moments of the medium frequency signal and of the baseband signal was researched. Second, according to the relationship, the estimation based on least-order cyclic moments was proposed. Finally, employing the detection of cyclostationary frequency, a blind CFO estimation for single carrier and OFDM signals was proposed based on least-order cyclic moments. The novel estimation requires no priori knowledge, such as the modulation scheme, symbol rate and so on. And it is adaptable to both single carrier signals and OFDM signals. The simulation results verify the effectiveness of the novel estimation, and show that for PSK and pulse amplitude modulation(PAM),the root mean square error(RMSE) is approaching to Cram?er-Rao bound(CRB), and that the performance of the proposed scheme is superior to the blind estimation based on fourthorder cumulant-based CFO estimation.For the estimation of FDOA, there are several schemes, such as maximum likelihood(ML) estimation, cumulant-based estimation, and ambiguity-function-based estimation. In this paper, we studied the estimation of FDOA under the circumstance of two star assisted localization based on satellite carrying and compared the performances of those schemes. The conclusion is that the maximum likelihood estimation and the ambiguity-function-based estimation have the best performance among these schemes. Moreover, to overcome the low estimation precision and the high complexity of existing schemes, the FDOA estimation based on ambiguity function with high precision and speed is proposed. The simulation results verify that the proposed estimation is superior to traditional ambiguity-function-based estimation in precision, and the RMSE of the novel estimation is approaching to CRB.