Infrared ranging in multipath environments for indoor localization of mobile targets

  1. Salido Monzú, David
unter der Leitung von:
  1. José Luis Lázaro Galilea Doktorvater
  2. Ernesto Martín Gorostiza Doktorvater

Universität der Verteidigung: Universidad de Alcalá

Fecha de defensa: 16 von Juli von 2015

Gericht:
  1. Jesús Ureña Ureña Präsident
  2. Daniel Pizarro Pérez Sekretär
  3. Esther Palomar González Vocal
  4. Eduardo Martos Naya Vocal
  5. Fernando Seco Granja Vocal
Fachbereiche:
  1. Electrónica

Art: Dissertation

Zusammenfassung

This dissertation focusses on the problem of the measurement of distance differences with optical signals affected by multipath, applied to indoor localization of mobile targets. Last advances in robotics, intelligent environments and autonomous vehicles have created a specific application field for indoor localization. Accuracy requirements in this context (in the cm level) are much stronger than the ones needed in human-oriented systems, which draw most of the technological development interest. Research on projective geometry methods (based on cameras) and multilateration (used with ultrawideband radiofrequency signals, ultrasounds and optical signals), have demonstrated potential performance to meet these requirements. However, all the alternatives, still under research, find some limitations when applied to real scenarios. Regarding optical systems, less studied in this context, previous works have been based on the measurement of differential phase of arrival using sinusoidally-modulated infrared signals. A centralized infrastructure computes the phase differences between fixed receivers from a signal emitted by the mobile target. The position is then obtained by hyperbolic trilateration. This research showed that precisions of few cm are achievable but, depending on the environment features, multipath interferences due to signal reflection in the surfaces may reduce accuracy up to tens of cm. Therefore, multipath effects are currently the main error source in this approach and, in consequence, the main handicap towards its practical implementation. In this thesis, a system to obtain accurate distance difference estimations reducing the critical effect of multipath is proposed. The approach is based on performing a pseudo random noise modulation on the sinusoidally-modulated signal typically used for continuous wave phase measurement, making use of the frequency spreading properties of these sequences to reduce multipath. The system, which sequentially performs a time and phase of arrival estimation, is composed of a synchronization stage, which enables partially coherent demodulation of the received signal, followed by a differential phase meter operating with the components dispread in the demodulation. Typical optical indoor multipath situations, with a clearly dominant lineof-sight component, allow the demodulation stage to recover higher amount of power from this contribution than from any other, consequently reducing multipath effects in the final estimation. All stages of the system have been designed and tested taking into account the relevant error sources: noise, multipath and dynamic effects due to target movement and emitter-receiver asynchronous operation. Tests have been carried out on a digital implementation of the proposed system and input signals have been generated with an emulation platform of the optical link. The results show that, considering the properties of up-to-date devices and with receivers separated 3.5 m, a global error below 10 cm is obtained in 90% of the cases, including noise and dynamic effects, for a wide room-height range and target velocities up to 1 m/s. Multipath tests demonstrate that the mitigation capabilities of the system are strongly conditioned by the optical link bandwidth and the environment height. Therefore, with the features of a currently available link, mitigation ratios bellow 20%, with respect to a standard phase measuring system operating in the same conditions, are obtained if the height is lower than 3 m. However, for environments with height greater than 4 m the mitigation ratio is between 70% and 95%, reducing multipath errors below 1 cm in most cases. The results obtained show that the application of the proposed method would contribute to enabling indoor localization using optical signals with an adequate performance for robotic and vehicle guidance applications. Furthermore, a significant improvement in the performance of the proposal is expectable in next years due to the progressive increase in the power and bandwidth of optoelectronic devices.