Lost in Space - Where the outer bound of localization space sets the lower bound on localization performance.
PhD thesis, University of Twente.
CTIT Ph.D.-thesis series No. 13-248
Full text available as:
Official URL: http://dx.doi.org/10.3990/1.9789036516914
This research reflects my theoretical and experimental journey into the lost space of wireless radio localization in the far field of 2.4GHz Commercial-Off- The-Shelf (COTS) radios. At the end of this journey, we arrive at the conclu- sion that existing phase- and time-based localization systems such as Radio Interferometric Positioning Systems (RIPS) and Time-Of-Flight (TOF) are not reliable in dynamic indoor environments. Our new localization system uses space-based rather than phase- or time-based measurements and shows ade- quate robustness for such environments.
In the far field, the measured signals are a function of the four wave param- eters time, position, temporal frequency and spatial frequency. These wave parameters are variables in propagation models that represent solutions to the Maxwell equations that govern the propagation of radio waves. Localization reduces to fitting the measured signals to the appropriate propagation model at the unknown locations. We identify three types of localization systems based on how the measurements deal with wave parameters: RSS-, phase- and TOF- based systems. The first part of this research explores these individual systems.
This journey starts by introducing a novel distributed connectivity-based localization system using a commonly employed flooding protocol. It exploits a certain part of the information in the protocol that other algorithms consider as redundant or false. This increases the localization performance in compar- ison with similar RSS-based systems, especially in harsh but static environ- ments.
In static environments, it is assumed that the optimal propagation model settings are known beforehand and are constant over space, time and hard- ware. In real indoor environments, these optimal propagation model settings depend on the locally and time varying permittivity and permeability of local- ization space. The challenge then becomes to determine the conditions under which RSS-based localization systems can calculate the optimal propagation model settings on-the-fly allowing for dynamic environments. These condi- tions turn out to be constraints on the localization surface acting as a spatial filter. Experiments verify that this approach can cope with dynamic environ- mental influences, like unknown and varying antenna orientations. However, the localization performance of such systems is of the order of meters, inade- quate for many applications. The located objects remain lost in space.
The research then turns to exploit the temporal coherence of our radio trans- mitters. Their narrow bandwidths allow two different transmitters to interfere and produce beat signals. Phase measurements of beat signals inherently pro- vide better localization performance, both in theory and in practice. Although the approach taken is unique and successful, earlier successful measurements in a different frequency regime had proven the feasibility of this rather complex but accurate localization technique. Our experiments in outdoor environments show accuracies of the order of decimeters. However, theory and experiments show that this approach cannot provide reliable indoor localization.
The final challenge then becomes to achieve robust outdoor as well as in- door localization. As space and time are interconnected through the constant speed of light, performing measurements in the space domain rather than in the time domain enable one to account for the high degree of spatial disper- sion in dynamic indoor environments. We call this approach space-based RSS. It is a simple and inexpensive localization technique that turns out to yield lo- calization performances approaching the theoretical limits as given by diffraction theory of electromagnetic radiation. Space-based RSS provides a simi- lar localization performance as phase- and TOF-based localization systems in outdoor environments. In Non-Line-Of-Sight (NLOS) indoor environments, space-based RSS outperforms existing phase- and TOF-based localization systems and provides our required robust localization performance.
In theory, resolving power in the far-field is determined by the ratio of wavelength and the outer dimension of localization space. This outer dimension in turn is limited by the spatial filter used as a constraint on our calibration-free localization system. In the end, it is not surprising that the outer bound of localization space sets the lower bound on localization performance in an inversely proportional relationship. Such relationships are commonly expressed by the well-known uncertainty principles for Fourier conjugates of wave parameters as well as by the equivalent Cramer-Rao-Lower-Bound principle. For the first time, this research compares these limits achieved by the relevant existing localization techniques, both in theory and in practice, and both in outdoor and indoor environments. As all measurements of comparable localization techniques such as RSS-, TOF- and phase-based localization were performed by us, this should leave little or no doubt about the validation of this theoretical and experimental comparison.
|Item Type:||PhD Thesis|
|Research Group:||EWI-PS: Pervasive Systems|
|Research Program:||CTIT-WiSe: Wireless and Sensor Systems|
|Research Project:||TSP: Tracking, tracing, Sensoring Platform|
|Deposited On:||06 February 2014|
Export this item as:
To correct this item please ask your editor
Repository Staff Only: edit this item