AI’s applications in indoor localization

INLAN is pushing the boundaries of today’s asset-tracking solutions by innovating a highly scalable real-time asset-tracking technology. Implementation cost, tag cost, and system accuracy are the three primary parameters that need to be considered when evaluating an asset-tracking system. Obtaining high level of accuracy however, comes with several technological difficulties, among which the multipath effect is the most challenging one. In indoor environments, Radio Frequency (RF) signals bounce off different objects before arriving at a receiver. As a result, the received signals, are heavily distorted, making accurate localization in rich scattering environments an exceedingly more difficult task.
In this project, we want to investigate the possibility of using different AI algorithms to suppress the multipath effect and improve localization accuracy. Some of the possible approaches are:
1. Train a neural network to learn some of the key characteristics of an environment. This knowledge can then contribute to a modified signal processing approach on the receiver side and, subsequently, an improved localization accuracy.
2. A selected number of reference tags with known locations can be used to train the model. Reference tags can also be helpful to prevent the system from overfitting and to track the changes in time-variant environments.

Faculty Supervisor:

Mohammadhadi Shateri;Éric Granger

Student:

Partner:

INLAN

Discipline:

Engineering

Sector:

Manufacturing

University:

École de technologie supérieure

Program: