Indoor localization using Wi-Fi

• Main Author: Siswanto, Hadi
• Other Authors: Lee Bu Sung, Francis
• Format: Final Year Project
• Language: English
• Published: Nanyang Technological University 2020
• Subjects: Engineering::Computer science and engineering
• Online Access: https://hdl.handle.net/10356/138070
• Institution: Nanyang Technological University

Nowadays, the need for indoor localization is increasing as it has many possible implementations in many sectors, e.g. navigation, health care, etc. In order to obtain an accurate indoor location, fingerprinting is the most commonly used method. In this method, data is collected to build a fingerprint database during offline phase, and during online phase, unknown data is compared to the data in the database to estimate the location. The problem with fingerprinting approach is the high variation of RSSI values which could result in erroneous location estimation. Machine learning approach is a new alternative to fingerprinting approach that aims to solve this problem. This report provides a comparison of potential machine learning models for a classification problem of the grid-based indoor localization. All models are evaluated using data collected from several reference points on block N4, North Spine, Nanyang Technological University. A simple Android application is developed to assist data collection. Understanding the RSSI data, data preprocessing and feature selection techniques were applied to clean outliers and unimportant features to improve the accuracy of the prediction models. The results show that random forest classifier is the overall best choice for classification-based indoor localization, with up to 93% accuracy. An Android application is also developed to illustrate the usage of indoor localization. A map and the user’s estimated location will be shown in the application, and the user is able to get a navigation from their location to any point of interest inside the area of interest.