Utilization of Different Wireless Technologies’ RSSI for Indoor Environment Classification Using Support Vector Machine

This paper presents the development of three indoor environment classifier model using support vector machine algorithm with Gaussian Kernel for three different wireless technology. The dataset that was used in this paper contained RSSI from three nodes for each wireless technology. Three different type of indoor environment was considered in this paper namely, small room with low interference, small room with high interference and large room with medium interference. Based on the results of the validation and testing for the three models, an overall accuracy of 57.1% was obtained for the classifier model using the RSSI of Zigbee technology while 84.8% was obtained for the model using the RSSI of BLE and 86.7% for the model using the RSSI of Wi-Fi technology. This corresponds to a conclusion that the model based on the RSSI of Wi-Fi technology is the best classifier model among the three for indoor environment classification. This paper can be used to further increase the accuracy of indoor localization using RSSI.