Simultaneous Localization And Mapping for Wireless Networks: SLAMWiN

Roost, Lars ; Østergaard, Michael

10. term

Elektronik og IT, Kandidatuddannelsen (Spec. Distribuerede Systemer)

2007

252

Dette speciale undersøger, hvordan Simultaneous Localization And Mapping (SLAM) kan forbedre indendørs lokalisering i eksisterende trådløse netværk. Udgangspunktet er, at modtagne signalstyrker (RSSI) fra Access Points (AP’er) kan udnyttes sammen med et Switching Extended Kalman Filter (SEKF) med fem bevægelsesmodeller til at spore Mobile Users (MU’er). Metoden overvåger ændringer i radiokanalen, når en MU bevæger sig mellem Line-of-Sight (LOS) og Non-Line-of-Sight (NLOS), og bygger iterativt en videnbase, der kompenserer for vægges negative indflydelse på signalerne. Herved estimeres både AP-positioner og vægplaceringer uden forudgående kendskab til miljøet, hvilket markant reducerer behovet for manuel opsætning og gør løsningen anvendelig i allerede etablerede Bluetooth- eller WLAN-net. Som biprodukt genereres et kort over miljøet, der kan bruges til yderligere nøjagtighedsforbedringer og visuel inspektion. Rapporten beskriver baggrund, systemarkitektur og komponenter (herunder filtrering, viden­deling, AP-estimering og kortbygning) samt en systemevaluering. Projektet konkluderer, at det er muligt at opbygge et næsten plug-and-play-lokaliseringssystem, der udnytter estimerede vægge, AP-positioner og initiale MU-placeringer med minimal menneskelig interaktion.

This thesis explores how Simultaneous Localization And Mapping (SLAM) can improve indoor localization in existing wireless networks. The approach uses received signal strength (RSSI) from Access Points (APs) together with a Switching Extended Kalman Filter (SEKF) featuring five motion models to track Mobile Users (MUs). It monitors radio channel changes as users move between Line-of-Sight (LOS) and Non-Line-of-Sight (NLOS) and iteratively builds a knowledge base that compensates for the impact of walls. In this way, both AP positions and wall locations are estimated without prior maps, substantially reducing manual setup and enabling deployment in established Bluetooth or WLAN networks. A by-product is an environment map that can further enhance accuracy and support visual inspection. The report covers background, system architecture and components (including filtering, knowledge sharing, AP estimation, and map construction), and a system-level evaluation. The project concludes that an almost plug-and-play localization system is feasible, leveraging estimates of walls, AP positions, and initial MU locations with minimal human intervention.

[This summary has been generated with the help of AI directly from the project (PDF)]

Download
View record in AAU Student Projects