On the identification of harmonic signal fields convergence method in an arbitrary noise field using the 1-D MV spectrum

• Published in: IEEE transactions on signal processing Vol. 44; no. 9; pp. 2311 - 2318
• Main Authors: Lyon, D.E., Sherman, P.J.
• Format: Journal Article
• Language: English
• Published: New York, NY IEEE 01.09.1996; Institute of Electrical and Electronics Engineers
• Subjects: Applied sciences; Colored noise; Computational complexity; Convergence; Detection, estimation, filtering, equalization, prediction; Exact sciences and technology; Information, signal and communications theory; Multidimensional systems; Narrowband; Polynomials; Random processes; Signal and communications theory; Signal processing; Signal to noise ratio; Signal, noise; Telecommunications and information theory; Harmonic; Signal processing; Identification; Noise; Spectral analysis; Signal estimation
• ISSN: 1053-587X; 1941-0476
• DOI: 10.1109/78.536686

This paper considers the recovery of a multichannel harmonic signal field corrupted by a possibly unknown homogeneous noise field. An approach is presented using the convergence-based spectra developed by Foias et al. (1990) in the random process setting. This technique has the advantage of discerning between the point and narrowband noise spectrum based on the monotonically decreasing convergence properties of a sequence of minimum variance (MV) spectra. For the proposed technique, the random field is reduced to a sequence of random processes using a set of condensing functions. An additional advantage of the proposed technique is that these condensing functions can be used to reflect a priori information and, hence, improve the effective signal-to-noise ratio (SNR). This technique uses information from all dimensions. Traditional techniques would separately apply a spectral algorithm to each dimension of the random field and thereby lose joint information from other dimensions.