Indoor sensor localization using RSSI and building map
Authors
Jensen, Marcus Bisgaard ; Jensen, Frederik Huse
Term
4. term
Education
Publication year
2022
Submitted on
2022-06-02
Pages
60
Abstract
In this project, we developed an algorithm to locate sensors inside buildings using transmit power (TxP), the received signal strength (RSSI), and the building floor plan. We first reviewed the problem and existing indoor localization methods. We chose a path-loss model, which estimates distance from how much the signal weakens, and we used the building map to account for walls and rooms that reduce signal strength. Based on this, we designed an algorithm that estimates distance, position, and the room where a sensor is located. To evaluate performance, we built a simulation tool and compared our approach with other localization methods. Our method, 'Multilateration with filtration' (combining distances from multiple points and filtering measurement noise), performed better than the alternatives in our tests, though further improvements are possible.
I dette projekt udviklede vi en algoritme, der kan finde sensorer indendørs ved at bruge sendestyrke (TxP), den modtagne signalstyrke (RSSI) og bygningens plantegning. Først gennemgik vi problemet og eksisterende metoder til indendørs lokalisering. Vi valgte at bruge en model for signalets dæmpning (path-loss), hvor afstanden estimeres ud fra, hvor meget signalet bliver svagere, og vi brugte plantegningen til at tage højde for vægge og rum, der dæmper signalet. På den baggrund designede vi en algoritme, der estimerer både afstand, placering og hvilket rum sensoren befinder sig i. For at vurdere ydelsen udviklede vi et simuleringsværktøj og sammenlignede den foreslåede løsning med andre lokaliseringsmetoder. Vores metode, 'Multilateration with filtration' (at kombinere afstande til flere punkter og filtrere målestøj), klarede sig bedre end alternativerne i vores test, men der er stadig mulighed for forbedringer.
[This apstract has been rewritten with the help of AI based on the project's original abstract]