Indoor Localization System Based on Bluetooth Low Energy Beacons and Spiking Neural Networks

Indoor localization is a technology that is used to quickly respond to first aid when a patient experiences sudden cardiac arrest. Users of this system are parties who would like to know the patient position and patients as detected objects. The way to do this research is to position the patient using Bluetooth Low Energy (BLE) as a Received Signal Strength Indicator (RSSI) transmitter that is a simple and affordable solution. Emotibit, a tool used by patients as an RSSI scanner from BLE, will send RSSI data to the cloud and be processed into a decision on the location of the patient. In this experiment, the data processing method employs a spiking Neural Network (SNN) algorithm, which has the ability to perform spatiotemporal feature extraction and low-power computation through binary spikes. This model produces a regression performance metric, namely Mean Square Error (MSE), which can reduce the error by 53.4% compared to other methods such as Support Vector Classification (SCV), K-Nearest Neighbor (KNN), and decision tree that are often used in indoor localization systems.