Traditional power-based localization methods suffer from low accuracy in the practical application environment. The main challenges are the antenna directivity and fading effect. Conventional methods assume omnidirectional antenna directivity such that the solution is the intersections of multiple circle-shape contours. This strong assumption results in significant localization error in practical non-isotropic antenna applications. In this article, a novel antenna radiation-pattern-aware power-based positioning (ARPAP) scheme is proposed. It reduces the antenna directivity effect by including the antenna pattern into the localization system model. It reduces the bias error that introduced by power measurement through estimating the line-of-sight (LoS) component in received signal strength (RSS). Moreover, the error mode for the proposed ARPAP system, along with the theoretical limit, Cramer-Rao Bound (CRB), and bias of the proposed positioning system are derived. The Pearson correlation coefficient between the proposed error model and simulation result shows a high similarity score. The proposed positioning scheme and analytic error model are instantiated for the cellular network. Both analytical model and simulation results demonstrate the superiority of the proposed method over traditional methods.

In this paper, bias and CRB analysis for both proposed line-of-sight (LoS) and conventional radio-signal-strength (RSS) based positioning methods are derived. In contrast to the conventional lognormal based analysis, the derived error models take into account the environmental multipath fading effect. The proposed LoS-based scheme improves positioning accuracy by eliminating the effect of non-range dependent portion of powers from the received signal. It suppresses non-line-of-sight (NLoS) disturbance to the positioning result through estimating the LoS component from fading signal. Hence, it is robust to environmental fading disturbance. The derived error models are validated by simulation. Both analytical model and simulation results show that the accuracy and robustness of the proposed LoS-based positioning are superior to that of the conventional RSS-based methods.

In this paper, a cost efficient fusion scheme, Ubiquitous Tracking with Motion and Location Sensor (UTMLS), is proposed for the accurate localization and tracking in mixed GPS-friendly, GPS-challenging, and GPS-denied scenario. The proposed drift-reduction method in UTMLS addresses the cumulating error issue in the indoor tracking with the consumer grade motion sensor. The proposed hypothesis test method in UTMLS improves the tracking sensor fusion precision by detecting distorted GPS reports and intelligently switching between GPS and inertial sensor based schemes. The proposed scheme is instantiated and implemented on an Android smartphone platform. Experiments have been conducted to evaluate and validate the accuracy. Experimental results show that (1) the proposed drift-reduction method effectively suppresses the non-GPS tracking error accumulation due to the integration of acceleration noise with time (2) UTMLS realizes robust indoor/outdoor seamless tracking, preventing GPS fault estimates introduced tracking error in the conventional Kalman filtering process.