Populating a Database to be used with an Indoor Positioning System

Sammanfattning: Indoor Positioning System (IPS) are becoming more common in many areas such as retail, warehouses, smart facilities, and manufacturing.In recent years, Bluetooth Low Energy (BLE)-based IPS has become increasingly popular due to its low cost and low energy consumption. One of the more recent updates, Bluetooth 5.1, provides the ability to compute the location using Angle of Arrival (AoA) or Angle of Departure (AoD). These new features have allowed for better positioning accuracies, where AoA-based positioning has shown sub-meter accuracy. An application area for BLE-based IPS is retail stores where the technology can benefit both the store and its customers. This thesis investigates how to populate a database to be used with an IPS in a real-life store. The assumption is that customers will have BLE equipped devices and run an application that will send the properly formatted BLE advertisements, such that an BLE IPS can locate the user in the store. Additionally, we assume that the application can use the device's e-compass or other means to determine in which direction the user's device is oriented. Based on the position and orientation of the user, the software is assumed to access a database to know what item(s) are near the customer. However, the question remains of how did this data get into the database? This degree project explores this in detail and assesses the amount of time and effort needed to populate this database and the amount of time and effort needed to keep this database up to date. This project followed an iterative Design Science Research (DSR) methodology where the artifact is the database. A relational database was used as they are widely used and joins can easily be performed and it is easy to modify existing tables. The application was developed in Spring Boot and React. Amazon Web Services (AWS) was used to host and provide the necessary services for the database and application. The result showed that the estimated time needed to populate the database in a supermarket with a sales area of 5300 m2, 36623 products, and 220 containers is 106.64 hours and 107.13 hours in the worst case assuming a walking speed of 1.4 m s-1. Updating a product would take 10.34 s and 10.37 s if the time it takes for a staff member to walk to the place where the product is located is excluded.