Wi-Fi-Based Localization in Dynamic Indoor Environment Using Denoising Filter and Neural Network

Wlan–Based Indoor Localization Using Neural Networks

“…Indoor localization systems have received significant attention in the last few years because of many health and safety-related applications for locating patients, devices, or law-enforcement personnel in large buildings [2], [3]. The Global positioning system (GPS) signals are greatly attenuated while penetrating building materials [4], which necessitates the need for developing alternative indoor localization techniques.…”

“…However, in recent years the widespread availability of Wireless local area networks (WLANs) based on the IEEE802.11 standard, commonly termed as WiFi, has led to new techniques for efficient indoor localization [8]. The common availability of multiple Wi-Fi access points (APs) in indoors and widespread use of portable smart devices allows cost-effective solutions for location estimation, which require minimal additional hardware; see for example [3], [7], [9], and [10]. The WLAN-based localization systems determine user location by utilizing the received signal strength (RSS), which is a measure of signal power received by a device, from multiple APs.…”

“…The fingerprinting approach provides accurate positioning even in complex indoor environments without the necessity to model the wireless propagation channel. In the literature, indoor localization has been investigated for many variants of the fingerprinting approach such as k-nearest-neighbor (k -NN) [13], probabilistic methods [7] and artificial neural networks (ANNs) [3].…”

“…However, the authors in [14], [15], [16], and [17] have not used an ANN in combination with the fingerprinting approach. A performance comparison of three types of dynamic neural networks was presented in [3]. It was shown that the feed forward time-delay neural network gives better location estimation with smaller training time than the nonlinear autoregressive network and the layer recurrent network.…”