Mobile indoor localization using Kalman filtering
Author
Bilgrau, Anders Ellern
Term
2. term
Education
Publication year
2011
Submitted on
2011-05-30
Pages
56
Abstract
Afhandlingen undersøger, hvordan man kan bestemme en mobil enheds position indendørs ved at kombinere usikre målinger med matematiske modeller. Vi bruger et Kalman-filter – en metode, der estimerer systemets skiftende men ikke direkte observerbare tilstand (en skjult Markov-kæde) – til at spore position. Afhandlingen forklarer Kalman-filteret og placerer det i den bredere familie af Bayesianske filtre, som opdaterer sandsynligheder, når nye data kommer til. For at vurdere ydeevnen simulerer vi bevægelse i et fiktivt kontormiljø med en variation af random waypoint-mobilitetsmodellen, hvor rutevalg styres af varmeligningen. Vi afprøver også en enkel heuristik, der genbruger tidligere estimerede positioner for at forbedre resultaterne. Resultaterne er blandede: Kalman-filteret kan forbedres en smule, men gevinsten er for lille til at retfærdiggøre den ekstra beregningsindsats.
This thesis explores indoor localization of mobile devices by combining noisy measurements with mathematical models. We use a Kalman filter—a method that estimates a system’s changing but unobserved state (a hidden Markov chain)—to track position. The thesis introduces the Kalman filter and places it within the broader family of Bayesian filters, which update probabilities as new data arrive. To evaluate performance, we simulate movement in a mock office using a variation of the random waypoint mobility model, with path finding guided by the heat equation. We also test a simple heuristic that reuses previously inferred positions to refine estimates. The results are mixed: the Kalman filter can be improved slightly, but the gains are too small to justify the extra computational effort.
[This abstract was generated with the help of AI]
Keywords
Documents