Research On Joint Blind Parameters Estimation In Signal Combining For Randomly Distributed Antennas

Signal combining technique for randomly distributed antenna system (RDAS) originates from deep space exploration field. To solve the problem of that it is dissatisfactory by utilizing existing large antenna system to effectively receive these long distant signals sometimes, signal combining technique for RDAS is proposed by utilizing multiple existing antennas to receive the same signal simultaneously. Signal combining techniques estimate the parameters of every received signals', compensate the differences and finish the signal combining by using optimal combining weights, to increase the received signal's SNR. Nowadays, signal combining technique for RDAS has gotten lots of attention in the deep space exploration field.How to correctly demodulate signals at low SNRs is a difficulty that communication signal processing always faced. Signal combining technique provides a new way for this kind of problem. However, most of relevant research work about signal combining focused on the special application environment of deep space exploration. To make the signal combining technique more applicable, the research of parameter estimation algorithm suited for the demand of multiple-antenna signal combining under low SNR condition is one of the key technologies. Signal combining technique for RDAS utilizes multiple independent antennas to receive signals from the same source. The received signals bear the same data information, which would be a beneficial for possible usage to improve the parameter estimation performance under low SNR condition. How to fully make use of this characteristic in the parameter estimation of signal combining is a new problem we faced.Focused on the improvement of parameters estimation performance at low SNRs, this thesis mainly researches joint blind estimation of M-ary phase-shift keying (MPSK) signal's frequencies, phases, SNRs, and combining weights. The primary work of this dissertation is summarized as follows:1. For the signal model in signal combining for RDAS, the Cramer-Rao lower bounds (CRLBs) of MPSK signal's frequencies, phases, combining weights and SNRs are derived respectively. The signal model in signal combining for RDAS can be modeled as a generalized kind of Single-Input-Multiple-Output (SIMO). Under this model, the Cramer-Rao lower bounds (CRLBs) of MPSK signal's frequencies, phases, combining weights and SNRs over SIMO channels are derived respectively. By comparing these results, it is discovered that, by utilizing multiple antennas cooperatively to receive the same signal, their parameter estimation's CRLBs are lower than the CRLBs with single receive antenna under NDA condition. And the NDA CRLBs approach the DA CRLBs as the number of antennas increased. For one hand, the above results verify the possibility of enhancement of parameter estimation performance under this condition. For the other hand, it offers a performance criterion for blind parameter estimation algorithms of the signal combining for RDAS.2. To the problem of optimal signal combining weights estimation in multiple-antenna signal combining, two joint blind combining weights and SNRs estimation algorithms for both the case of frequencies known and known is proposed. First, according the case of frequencies is known, a joint blind combining weights and SNRs estimation algorithms based on signal covariance matrix is proposed. By utilizing multiple signals'covariance matrix, this proposed algorithm formulates over-determined equations with respect to the signal amplitude and noise power. Then the combining weights as well as SNRs estimation are obtained through least squares (LS) method. Based on the above algorithm, an improved joint blind estimation algorithm base on signal differentials'covariance matrix is proposed for the case of unknown frequency. The improved algorithm utilizes the covariance matrix of constructed signals'differential form in order to eliminate signal frequency's influences. Simulation results indicate that, compared to existing moment-based SNR estimators, both of the proposed algorithms can achieve better estimation resolution performance. The design process is irrelevant to the signal modulation manner, thus they are applicable for many linearly modulated signals.3. To the problem of MPSK carrier phase estimation at low SNRs, two joint blind carrier phase estimation algorithms for MPSK signals over SIMO channels are developed. First, a LS based joint blind carrier phase estimation algorithm is proposed by utilizing the characteristic that the phases of constructed variant received signals'multiply forms are linear with the phases of the received signals. In order to further improve estimation performance, an approximate maximum likelihood estimation algorithm is proposed, where the joint maximum likelihood (ML) estimation problem of carrier phases is converted into weighted LS estimation of each signal'carrier phases. Compared with the above algorithm, this improved algorithm uses more forms of nonlinear transformation, and sets different weights for different multiply signals. Simulation outcomes demonstrate that, both algorithms can effectively improve MPSK signal's carrier phase estimation performance, and a significantly improvement is achieved for phase estimation especially at low SNRs. And the approximate ML algorithm has better estimation accuracy as a consequence.4. Based on the above proposed joint blind carrier phase estimation algorithms, a joint blind frequency and carrier phase estimator is proposed for MPSK signals over SIMO channels. The proposed algorithm utilizes different combinations of the received signals'multiplied forms to construct single tone signals. Meanwhile, in order to further enhance the estimation performance, a studentized residual based outlier detection method is designed to remove the outlier data probably existed in single tone signals'frequency estimation. Simulation results show that, the proposed algorithm estimator offers a superior estimation resolution performance with respect to its counterpart in single receive antenna at low SNRs, and have a much lower SNR threshold where an outlier phenomena is observed.5. Taking the multi-dimensional, multimodal, and non-linear characteristics of the MPSK signals'ML function in SIMO channels into consideration, particle swarm optimization (PSO) based joint blind ML estimation algorithms of MPSK signal's frequencies, phases, combining weights and SNRs are investigated. First, by analyzing the joint ML functions of two parameter kinds, and taking the performance characteristic of PSO and its improved forms into consideration, a PSO based combining weights and SNRs estimation algorithm, and a comprehensive learning particle swarm optimization (CLPSO) based frequencies and carrier phase estimation algorithm are researched respectively. On this basis, through analyzing ML function of the four parameter kinds, it is found that when frequency is limited to a specific range, the complexity of the ML function can be greatly reduced. Considering this characteristic, a two-stage PSO based joint blind ML estimation algorithm of MPSK signal's frequencies, phases, SNRs and combining weights is proposed. This algorithm uses the excellent global optima search ability of CLPSO and the fast convergence speed of basic PSO. Simulations demonstrate that, the two-stage PSO algorithm can achieve satisfying estimation of these parameters, and outperforms CLPSO or PSO for joint estimation of these four kinds of parameters.