WLAN Based Indoor Localization System for Evacuation of Victims in a Building

Severe damage to a building can trap victims inside and cause human fatalities. One of the problems that emerge during the victims' evacuation process is detecting the number of victims and their location. An indoor localization system can be a solution to this problem. Indoor localization systems can utilize WLAN Access Point (AP) devices, which are generally available in most buildings that provide internet access. Therefore, this research aims to collect fingerprinting data, analyze the accuracy of victims' position using the Euclidean Distance (ED) algorithm, and build an Android mobile application to detect several victims trapped in a building WLAN network, analyze the functionalities and the usability of the application. The fingerprinting method is collected using an Android application designed explicitly for Received Signal Strength (RSS) collecting. This mapping process collects signal data from 1 and 3 sources of AP for every 2 meters in four directions. The training data is used to compare the user's current position with the actual point using the ED algorithm. The use of KNN classification gives a better result than no classification method. Also, the accuracy of point and room prediction using 3 APs is higher than using 1 AP. The accuracy of room prediction is 87% by using 3 APs. The Android application consists of regular user and rescuer applications. Both applications passed the functional testing with the Black Box method and tested for usability using the USE questionnaire method. Based on the result obtained, the application has met all of the usability requirements.