<p>With the massive growth of wireless technology, it is becoming increasingly necessary to use system position to provide high resource efficiency. One of the primary solutions is the Global Positioning System (GPS); however, it fails to localize devices in an indoor environment. Previous work focused on implementing localization schemes utilizing additional hardware and sensors, which may suffer from overhead costs. Furthermore, manufacturers build a wide variety of wireless products (i.e, the internet cameras or any IoT product) with different structures and sizes, which may not have similar features, or hard to attach new devices. Thus, it becomes difficult to use one specific localization scheme for all kinds of devices. Eventually, we can conclude that all wireless devices must come with a network interface card (NIC). Therefore, we propose a novel hierarchical deep learning based approach that utilizes channel signal properties to localize devices efficiently. The hierarchical indoor localization system does not require any additional hardware. We verify our work by conducting a real-life experiment that employs commercial-off-the-shelf (COTS) WiFi devices to extract physical layer information (channel state information (CSI)). Our system learns features from CSI, which can discriminate and identify the location of each connected device.</p>