Article Preview
TopIntroduction
The proliferation of location-based services in today’s world, which employ a user’s geographic location to supply location-specific data, has led to a greater need for precise and up-to-date indoor positioning systems that can support indoor location-based services (Tan et al., 2021). Although the GPS is widely employed to enable outdoor wayfinding and positioning, the system is not ideal for use in indoor environments because it necessitates an unobstructed view of the link connecting the GPS satellites and users (Ezhumalai et al., 2021; J. Wang & Park, 2021). In indoor environments, meeting this requirement is challenging because signals are often blocked by the thick walls of buildings, resulting in a weakened signal that diminishes the accuracy of indoor positioning data.
Pertaining to the above-mentioned issue, various wireless technologies such as Bluetooth, RFID, geomagnetism, proximity sensor, ultra-wideband (UWB), visible light, and Wi-Fi have been extensively studied for their applications to facilitate indoor positioning systems (Ezhumalai et al., 2021; J. Wang & Park, 2021). Among the available approaches, the fingerprinting method based on received signal strength (RSS) is unique because it does not need any additional infrastructure other than the commonly placed Wi-Fi access points (APs) or Bluetooth low energy (BLE) beacons, along with mobile devices equipped with network interface cards to measure RSS (Ezhumalai et al., 2021; Khalajmehrabadi et al., 2017a). Fingerprint-based indoor positioning using RSS measurements faces limitations due to the complexities of Wi-Fi signal propagation indoors. Multipath interference caused by reflections from walls, furniture, and even people disrupt the direct signal path, leading to unreliable RSS values and hindering radio map accuracy (Ji et al., 2022). Additionally, environmental factors like temperature and humidity can subtly affect signal strength, while human movement during measurements and device orientation can introduce further inconsistencies. These limitations can create significant discrepancies between the user’s actual location and the estimated position based on RSS fingerprints.
The fingerprinting method based on RSS encompasses two main processes: the offline phase and the online phase. The offline phase is the process where the RSS measurements are taken from nearby APs at various reference points (RPs) throughout the indoor environment of interest to create a radio map containing the location-tagged RSS measurements (Shang & Wang, 2022). Specifically, Wi-Fi APs bridge wired networks (Ethernet) with Wi-Fi devices using radio frequencies for data transmission and reception. Meanwhile, a RP is a specific location within the indoor environment where RSS measurements are taken. These measurements capture the unique “fingerprint” of the Wi-Fi signal at that particular point. On the other hand, in the online phase, the user’s unknown location can be approximated by comparing the RSS values obtained from visible APs near the user’s current location with the labeled RSS vectors pre-collected and stored in the radio map, using a localization technique such as decision tree, random forest, or k-nearest neighbor (KNN; Ezhumalai et al., 2021; J. Wang & Park, 2021).
However, generating the radio map can be a time-consuming and labor-intensive process since it requires the RSS measurements to be performed at each RP defined over the entire indoor environment (Bi et al., 2018). Taking the real-world scenario, which usually involves a large-scale multi-floor indoor environment, would imply that a more significant number of RPs must be defined to cover the whole area of interest. Apart from that, to suppress the adverse effect introduced by outliers and noises, it is common to calculate and store the average RSS vectors as fingerprints in the radio map by collecting multiple measurements at each RP. Additionally, in some cases, multiple directional sampling is performed at each RP to account for the influence of human body shielding on RSS measurements.