Source Enumeration Based on a Uniform Circular Array in a Determined Case.

The performance of the source enumeration with the uniform circular array (UCA) is deteriorated considerably by the colored noise. This effect becomes more obvious especially when the total number of the sources is lower and closed to that of the array sensors. To address this problem, this paper proposes a source enumeration modeling based on a UCA having M antennas (M-UCA). The M-UCA is operated in a narrow band far field environment with the arbitrary azimuth and elevation angles. First, the Hilbert Huang transformation is employed to derive the phase sequence of each selected intrinsic mode function (IMF) of the signal acquired by each antenna of the M-UCA. Then, the phase sequences of the selected IMFs of the signal acquired by all the antennas of the M-UCA are employed to construct an instantaneous phase covariance matrix (IPCM) among the antennas in the UCA. An IPCM is constructed for each order of the selected IMFs. After performing the eigenvalue decomposition of the all IPCMs, a set of eigenvalues carrying useful information for the estimation of the total number of the sources is acquired. This set is served as the training set. Finally, the training set is used to derive the mathematical model via the support vector machine. The trained model can be used to perform the sources number detection. Both the computer numerical simulations and the radio frequency anechoic chamber experiments validate that the source estimation accuracy of the constructed M-UCA mathematical model based on our proposed method has a high robustness to the incident angles variations, the SNR variations, and the snapshot number variations in the colored noise environment in the determined case, especially in the case of that the total number of the signal sources is only one less than the number of the antennas in the UCA. [ABSTRACT FROM AUTHOR]