Study On Beamforming Algorithm And Its Implementation In The Case Of Contaminated Sample Data

The weights of adaptive beamforming are calculated by means of definite algorithms using sample data, and sample data are required to obey certain distributions. The performance of adaptive beamforming degrades when desired signal is present in sample data, i.e. the sample data are contaminated by desired signal. The kind of sample data is called as contaminated sample data for short in this thesis. The beamforming algorithms are studied under the certain conditions of contaminated sample data. The details of this thesis are organized as follows:The performance will degrade if the traditional beamformer via orthogonal projection is used directly in the case of contaminated sample data. A novel algorithm is studied which blocks desired signal by preprocessing sample data, constructs a new covariance matrix and then implement the orthogonal projection algorithm. The performance of the proposed algorithm is the same as that of the orthogonal projection which is implemented employing the desired signal-free sample data. The rapider convergence of the out SINR is attained.ESB algorithm (Eigenspace-Based Algorithm) is also valid in presence of desired signal, but the key of ESB algorithm is eigendecomposition of covariance matrix and the computational complexity of eigendecomposition is enormous. We discuss a modified ESB algorithm via a unitary transformation. The algorithm has the overall computational load reduced extremely by means of a unitary transformation. At the same time, due to the existence of the forward-backward averaging effect, the method can converge rapider, and get excellent performance in scenarios with partially correlated or fully coherent source pairs. The simulations and performance analysises demonstrate the feasibility and validity of the two algorithms that are mentioned hereinbefore.This thesis also discusses the methods to design universal beamformer. In the hardware design, multi-processor parallel processing technology is come up. Then the architecture of a universal beamformer is introduced. The software programming is also modularized in this dissertation. The function frame of the system is established in the main program, and modules are called. It is of great universality and independence. The key problems about how to implement a modified ESB algorithm via a unitary transformation are discussed. In the end , the flow chart of the program is shown.