Implementation, Performance Assessment, and Refinement of Selected Distributed Localization Algorithms for Cooperative Wireless Networks
Author
Koldsgaard, Michael Bak
Term
4. term
Education
Publication year
2013
Abstract
This thesis addresses localization algorithms for mobile devices in cooperative wireless networks, focusing on indoor environments where GPS is unavailable or unreliable. After a brief introduction to localization techniques and indoor challenges such as obstacles and ambiguous distance measurements, the work builds on a distributed localization algorithm based on Variational Message Passing (VMP). The original unimodal VMP approach is extended to a bimodal variant (B-VMP) to better represent multi-peaked position distributions. Both algorithms are implemented and evaluated using Monte Carlo simulations and tests on physical measurement data from the EU-funded WHERE2 project. The results indicate that the proposed bimodal extension can yield improved position estimates in cooperative settings at the cost of increased computational complexity. The thesis documents the implementation, benchmarks performance, and discusses the trade-off between accuracy and complexity.
Specialet behandler algoritmer til lokalisering af mobile enheder i kooperative trådløse netværk med fokus på indendørs brug, hvor GPS ofte er utilgængelig eller upålidelig. Efter en kort introduktion til lokaliseringsteknikker og indendørs udfordringer som forhindringer og tvetydige afstandsmålinger, tager arbejdet udgangspunkt i en Variational Message Passing (VMP)-baseret distribueret lokaliseringsalgoritme. Den oprindelige, unimodale VMP-tilgang udvides til en bimodal variant (B-VMP) for bedre at kunne repræsentere flertydige positionsfordelinger. Begge algoritmer implementeres og evalueres gennem Monte Carlo-simuleringer og tests på fysiske måledata fra det EU-finansierede WHERE2-projekt. Resultaterne viser, at den foreslåede bimodale udvidelse kan give forbedrede positionsestimater i kooperative scenarier, dog med øget beregningskompleksitet. Specialet dokumenterer implementeringen, sammenligner ydeevnen og diskuterer afvejningen mellem nøjagtighed og kompleksitet.
[This apstract has been generated with the help of AI directly from the project full text]