Bearing estimation in a Ricean channel - Part I: Inherent accuracy limitations

Citation
G. Fuks et al., Bearing estimation in a Ricean channel - Part I: Inherent accuracy limitations, IEEE SIGNAL, 49(5), 2001, pp. 925-937
Citations number
22
Categorie Soggetti
Eletrical & Eletronics Engineeing
Journal title
IEEE TRANSACTIONS ON SIGNAL PROCESSING
ISSN journal
1053587X → ACNP
Volume
49
Issue
5
Year of publication
2001
Pages
925 - 937
Database
ISI
SICI code
1053-587X(200105)49:5<925:BEIARC>2.0.ZU;2-2
Abstract
This paper considers the problem of estimating the bearing of a single, far -field source using passive sensor array measurements when the spatial prop agation channel formed between the source and the array may be described as "Ricean." Such a channel consists of a direct, line-of-sight (LOS) compone nt as well as an indirect, nonline-of-sight (NLOS) component due to scatter ing. A parametric description of the resulting spatial propagation channel is presented. A related source-bearing estimation problem is formulated, an d the associated Cramer-Rao lower bound (CRLB) is evaluated. The bound is u sed to study the relationship among the bearing estimation problems under t he Ricean (LOS/NLOS), point source (LOS), and scattered source (NLOS) model s. Exact and simplified approximate forms of the bound are derived explicitly in terms of the point source and scattered source CRLBs. A number of proper ties of the bound are presented. In particular, it is shown that the bound is a monotonically decreasing function of Rice factor (the ratio of the LOS component power to the NLOS component power). This implies that for a give n signal-to-noise ratio (SNR), the CRLB is bounded from below by the point source bound and bounded from above by the scattered source bound. It is al so shown that given a NLOS component, the addition of a LOS component neces sarily makes the bearing estimation problem easier. On the other hand, give n an LOS component, the addition of an NLOS component does not necessarily make the bearing estimation problem easier (and may even make it harder). Last, the CRLB for estimation of the Rice factor is considered, and some of its properties are studied.