Indoor Positioning System Using WiFi Packet Sniffing as Attendance System

• Main Author: Tanadi - NIM: 18214041, Theo
• Format: Final Project
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/31284
• Institution: Institut Teknologi Bandung

Attendance systems that exist right now have gone through a lot of development along with the development of information technology. If older attendance systems still need humans to do manual recordings on paper, newer attendance systems use many kinds of sensors to identify its user, record the data on database, and many other information technologies to do its task. Even so, there is still a common weakness that exists in older and newer attendance system, that it still need the human user to do the recording proses manually, even the process is already assisted with information technologies. <br /> <br /> <br /> <br /> In this final project report, a system that can solve the weakness above is designed and prototyped. The proposed solution is in the form of attendance system that is based on indoor positioning system using WiFi packet sniffing. If a user is located at a specified area, then the user is considered attending. RSSI feature on WiFi is used as a parameter to determine the user location. A prediction model is created using extremely randomized tree to predict the location based on RSSI features from sniffers. MAC address is used as an unique identifier for every user device. There are three sniffers used that use the ESP8266-01 module and a centralized server. <br /> <br /> <br /> <br /> System location prediction and attendance status determiner accuracy have been tested. Testing is done in two hours, one for testing inside the area, one for testing outside the area. System location prediction accuracy is 0.773, with its true positive rate of 0.773 and its true negative rate of 0.708 System can determined four out of four devices correctly when the testing is done inside the area and three out of four when the testing is done outside the area. There are some factors that can affect the accuracy of the prediction, such as existence of a wall, device type, and denseness of the broadcast frequency.