REPOSITORY > RESULTS

Abstract

As the number of location-aware applications grows, the demand for indoor localization
systems is increasing. Motivated by the need for localization systems that enable accurate
location estimation, robustness to multipath, simple deployment, and scalability, supported
by low-cost devices, this dissertation focuses on two phase-based approaches within the
measurement stage of localization: direction of arrival (DoA) estimation using single-radiofrequency
(RF)-chain architectures, and distance estimation with the multi-carrier phase
difference (MCPD) method. In both cases, the estimation parameter is derived from
carrier-phase measurements, making both techniques sensitive to hardware non-idealities
that are more pronounced in low-cost devices. The dissertation examines hardware-related
mechanisms that distort phase measurements and introduce systematic bias, and develops
mitigation approaches that enable accurate localization for resource-constrained devices.
The first part of the dissertation addresses DoA estimation using single-RF-chain architectures.
The work initially examines phase errors introduced by the switched antenna
array (SAA), which is required to obtain spatial samples for direction estimation. A complete
SAA design workflow is presented, with an emphasis on minimizing phase distortions
across switched antenna elements and RF paths to enable accurate DoA estimation without
post-fabrication calibration, supporting simpler and lower-cost deployments. The dissertation
then examines how asynchronization between devices, particularly sampling frequency
offset (SFO) and carrier frequency offset (CFO), caused by non-ideal crystal oscillators,
leads to phase-measurement errors in DoA estimation. It is shown that a conventional
sequential spatial sweep across antenna elements induces a phase-error pattern that can
be coupled with the DoA-dependent phase pattern, resulting in an estimation bias. An
efficient mitigation approach based on antenna activation order is proposed to reshape the
measured phase-error pattern across antenna elements and reduce bias.
The dissertation then transitions to MCPD ranging and analyzes how asynchronization
between devices affects distance estimation. Specifically, the work first examines how SFO
between devices alters sampling times, which, under the conventional linear frequency
sweep, induces a linear phase-error trend across carriers that biases the distance estimate.
An efficient bias-reduction approach based on carrier order is proposed to suppress this
trend. In addition, the dissertation analyzes the impact of frequency-dependent CFO on
distance estimation arising from phase-locked loop-based RF carrier generation, which can
introduce a linear phase-error trend that biases distance estimates. To suppress this bias,
timing diagram design strategies are developed using temporal differential correction and
role-switching approaches.
The proposed asynchronization-mitigation approaches diverge from traditional estimation-
and-correction and joint parameter-estimation approaches, which require additional
measurements, longer observation windows, and greater computational load for effective
phase-error mitigation. In contrast, the proposed mitigation approaches leverage switchingsequence
and timing design to reduce systematic bias with negligible additional computational
and measurement overhead.