Abstract:

In order to adapt to the actual direction finding system, aiming at the problem that the estimation of performance reveals a change for the worse when using the conventional method of estimating the number of signals in colored noise field, a new approach based on characteristic subspaces projection is proposed. First, grouping the eigenvectors, the eigen-subspaces which only contain noise subspace and the other contain both noise and signal subspace are received. The array manifold matrix is orthogonal to the noise subspaces of the covariance matrix, and it has the same expanded space with the signal subspaces. According to this principle, the covariance matrix is projected on the eigen-subspaces, so projection data are received. Finally, the variance of the projection data is calculated, then the projection variance is received and is regarded as the evidence for signals number estimation, and the number of signals is estimated. Computer simulation and measured data test show the correctness and effectiveness of the algorithm.