Resumen
The project developed here solves the problem of how to relate maps in different coordinate systems. We are in a scenario where, at the same time that a user is walking, we estimate his position and the position of the transmitters necessary to triangulate the user. Thanks to multipath we can position the user with a single physical transmitter, using Multipath Assisted Positioning (MAP), and with the help of Channel-SLAM (Simultaneous Localization and
Mapping) the map of the virtual transmitters is estimated.
MAP uses multipath propagation produced by both reflection and scattering to estimate a point on the map where we can locate a virtual transmitter (VT), and treat each multipath
component as if they were direct signals between the user and the estimated VT. In order to position both the user and the VT’s, a particle filter (PF) is used, which exploits information such as the angle of arrival (AoA) or the time of arrival (ToA) of multipath components. To
improve the positioning performance, an Inertial Measurement Unit (IMU) is used.
Since multipaths are kept more or less static for a few tens of meters, a map of the positions of the previously estimated VTs can be used, comparing at each time step with the map that the user is plotting on his current route. If both maps are considered to belong to the same environment,
the user will no longer compute the estimates of the positions of the virtual transmitters and will switch to using the information contained in the map that the user previously had.
This reduces the power consumption of the terminal and considerably accelerates the process of stamping the user’s position.
The thesis ends up presenting the results obtained in the simulations of the developed algorithm. The simulation set contains both code in Matlab and C++. This is because of how simple it is to work in Matlab with vectors and arrays, and at the high speed of execution of C++ code.
