Deployment of Localization System in Complex Environment Using Machine Learning Methods

Conference lecture rooms are considered as an indoor environment for research with indoor localization that faces different challenges. In this paper, we design an indoor localization system in such a complex environment focus on machine learning algorithms with a large scale data. In this system, users can get information where a specified people locate in. Our system has two main sections: system design and algorithm approach. We use Bluetooth Low Energy devices to achieve communication and collect the information consist of people's name and corresponding RSSI data. To illustrate the advantages of machine learning algorithms using in this situation, a traditional algorithm will be proposed as a comparison. We also design and implement a Nine-Rectangle-Grid localization system in a practical laboratory environment. Experiments on real-world environment and simulations show our system can live up to locate people with low cost, high accuracy and short computing time.