Leto: Crowdsourced Radio Map Construction With Learned Topology and a Few Landmarks

Existing crowdsourced indoor positioning systems (CIPSs) usually require prior knowledge about the site and a tedious calibration process. Moreover, they may large number of landmarks while ignoring topology information that be contained in data. In this paper, we present Leto, system uses le arned xmlns:xlink="http://www.w3.org/1999/xlink">to pology from combined user traces to construct radio map. Leto relies on WiFi accelerometer signals only without requiring any site. Our key idea is learned can reduce required landmarks, available transform into We propose novel framework efficiently learns map by hybrid multidimensional scaling (HMDS) algorithm accurately rectifies using few anchors an adaptive force-directed (AFD) algorithm. also provide theoretical convergence analysis HMDS Experimental results real-world datasets show capture useful achieve significant improvements construction compared existing systems.