Reliable location estimates are attainable despite measurement errorsDetermining a mobile phone user’s location with global navigation systems is often problematic due to a great energy requirement and poor indoors signals, for example. MSc Philipp Müller has developed advanced methods for modelling measurement errors and attaining more precise position estimates.
In his dissertation, MSc Philipp Müller has developed methods for solving positioning problems in which measurements used for position estimate computation contain errors. The new methods model these measurement errors with enhanced precision, thus resulting in more accurate position estimates. They also require only little more computation power than the currently used, simple methods, which facilitates their usage on mobile devices, such as smartphones.
For indoor positioning, Müller uses ultra wideband (UWB) measurements instead of GPS to improve the quality of the signals. Measurements from the mobile phone network (2G, 3G or 4G) are not designed for positioning purposes, but they contain position dependent information and using them reduces the energy requirement and saves battery life.
The received signal strength measured from a user’s mobile phone depends on the distance between the phone and the base station used. The measurement estimates for the distances between the base station and the phone are used to infer the user’s position. However, the equation that describes the dependence between the distance estimates and the user’s position generally includes a non-linear function, which is somewhat problematic. Furthermore, the distance estimates are not exact but rather under- or overestimate the distance between base station and user; in other words, they contain errors.
Müller has developed two methods that take into account the non-linearity of the function and the errors in the distance estimates. The first method is a generalization of the widely used concept of Gaussian mixtures. The novel method tremendously reduces the number of computations required for determining the user’s position, enabling the real-time execution of the method on mobile phones. The second method focuses on the errors. Many existing methods assume the errors to be normally distributed, but this is an oversimplified hypothesis. Müller applies a more flexible model that provides a more accurate fit to empirical error data than the normal distribution, resulting in more precise position estimates. The computation time required by the novel model is somewhat greater than that of the normal model, but still enables real-time positioning on mobile phones.
Public defence of a doctoral dissertation on Friday, 26 August
The doctoral dissertation of MSc Philipp Müller in the field of automation science and engineering entitled ‘Algorithms for Positioning with Nonlinear Measurement Models and Heavy-tailed and Asymmetric Distributed Additive Noise’ will be publicly examined at the Faculty of Engineering Sciences of Tampere University of Technology (TUT) in Auditorium S2 in the Sähkötalo building (address: Korkeakoulunkatu 3, 33720 Tampere, Finland) at 12 noon on Friday, 26 August 2016.
The opponent will be Professor Guiseppe Abreu (Jacobs University Bremen, Germany). Professor Robert Piché from the Positioning Algorithms group at TUT will act as Chairman.
Philipp Müller (32) comes from Chemnitz, Germany, and works as a researcher in the Positioning Algorithms group at the Department of Automation Science and Engineering of TUT.
Philipp Müller, tel. +358 45 610 4290, email@example.com
The dissertation is available online at: http://urn.fi/URN:ISBN:978-952-15-3784-4