Research On Sparse Adaptive Beamforming Based On Nonuniform Norm

In many practical applications,the array is required to have high resolution(the main lobe of the array pattern is narrow),and the gain of the scanning beam is not high,such as,the environmental interference satellite receiving antenna.Have a great relationship with the high resolution array aperture array,so you can use sparse array way to construct a sacrifice high directivity array gain.In addition,the sparse array can make the distance between the neighboring elements do not have to meet the constraints of half wavelength,which would allow the mutual coupling between array elements becomes very weak,namely: without considering the impact of adjacent array mutual coupling between elements of the direction map.On the other hand,in the field of mobile communication,in order to save power supply which can only use a small number of elements,but the strong interference environment,fewer sensors and can not get a satisfactory array diagram,this paper proposes an algorithm which can get a better tradeoff between the use of array element number and array pattern,satisfy the power supply demand and direction of the desired map,and improve the resolution of the array and reduce the mutual coupling between array elements.In this paper,the NU-CNLMS(non-uniform norm Linearly constrained least-mean-square)algorithm is mainly aimed at the existing method of sparse regularization:L1 norm sparse factor is the same punishment degree of all filter coefficients,larger estimation error convergence is slow,basically use the L0 norm approximation function to approximate solution,the estimation error is large,the basic the idea of NU-CNLMS algorithm is embedded in the non uniform norm to LMS punishment function,can be made by adjusting the related parameters to achieve an estimation error minimization and achieve a high degree of sparseness added weights to each of the array output data.Secondly,based on the above algorithm,propose an improved version of the original algorithm NU-WCNLMS(non-uniform weighted Variable Step Size Linearly constrained least-mean-square),the algorithm for the original algorithm for small coefficients are equally weighted,did not distinguish between the degree of sparsity problem,the faster a reweighted method makes the smaller coefficient of attenuation,high the coefficient of attenuation is slow or attenuation,to further accelerate the convergence of NU-CNLMS and sparse.Although the algorithm is applied to adaptive beamforming,this algorithm is not only limited to beamforming,but also applied to other sparse systems,such as echo cancellation,signal reconstruction,Alzheimer'sdisease diagnosis..Finally,the formation of the validity problem show that the proposed algorithm by adaptive beamforming,and the sparse recognition system simulation results show that the proposed algorithm has high performance compared with the algorithm in convergence.Finally simulation results show that the proposed algorithm in the above scenario,can achieve greater sparsity and faster convergence speed without loss of direction map to the case,achieves the desired performance.