Mismatch of Near-Field Bearing-Range Spatial Geometry in Source-Localization by a Uniform Linear Array

Many near-field source-localization algorithms intentionally simplifies the exact spatial geometry among the emitter and the sensors, in order to speed up the signal-processing involved. For example, the Fresnel approximation is a second order Taylor-series approximation. Such intentional approximation introduces a systemic error in the algorithm's modeling of the actual objective reality from which the measured data arise. A mismatch thus exists between the algorithm's presumptions versus the data it processes. This modeling-mismatch will introduce a systematic bias in the bearing-range estimates of the near-field source-localization algorithm. This bias is non-random, and adds towards the random estimation-errors due to the additive and/or multiplicative noises. The open literature currently offers no rigorous mathematical analysis on this issue. This proposed project aims to fill this literature gap, by deriving explicit formulas of the degrading effects in three-dimensional source-localization, due to approximating the source/sensor geometry by any order of the Taylor's series expansion.