# Automated Room-Level Localisation Using Building Plan Information

## TL;DR

This paper introduces a new automated method for placing wireless sensors in buildings using building plans, reducing the need for manual work during installation.

## Contribution

The novel contribution is an automated localization method using building plans to improve sensor placement accuracy in BMSs.

## Key findings

- The proposed method achieves room-level localization accuracy with minimal manual effort.
- Using building plan data improves localization accuracy, reducing errors to 1 in 20 or 1 in 200 sensors.
- Simulation tests confirmed the effectiveness of the automated approach compared to traditional methods.

## Abstract

Building Management Systems (BMSs) are transitioning from utilising wired installations to wireless Internet of Things (IoT) sensors and actuators. This shift introduces the requirement of robust localisation methods which can link the installed sensors to the correct Control Units (CTUs) which will facilitate continued communication. In order to lessen the installation burden on the technicians, the installation process should be made more complicated by the localisation method. We propose an automated version of the fingerprinting-based localisation method which estimates the location of sensors with room-level accuracy. This approach can be used for initialisation and maintenance of BMSs without introducing additional manual labour from the technician installing the sensors. The method is extended to two proposed localisation methods which take advantage of knowledge present in the building plan regarding the distribution of sensors in each room to estimate the location of groups of sensors at the same time. Through tests using a simulation environment based on a Bluetooth-based measurement campaign, the proposed methods showed an improved accuracy from the baseline automated fingerprinting method. The results showed an error rate of 1 in 20 sensors (if the number of sensors per room is known) or as few as 1 per 200 sensors (if a group of sensors are deployed and detected together for one room at a time).