A robust room-level localization method based on transition probability for indoor environments

Integration of Directional Antennas in an RSS Fingerprinting-Based Indoor Localization System

“…In [ 24 ], a kind of artificial neural network, called multilayer perceptron (MLP), was proposed to map for the first time the relation between coordinates ( X,Y ) of the tag position and the RSS gathered by an ad hoc WLAN. In [ 23 , 24 , 25 , 26 , 27 ], other types of learning machines, such as support vector machines (SVM), learning vector quantization (LVQ) and radial basis functions (RBF), were respectively studied for the same purpose, obtaining similar results. Up to date, most of these systems offer a coarse precision of 3–6 m for 90% of estimations [ 4 ], if no additional information/sensors are integrated [ 16 ].…”

“…Up to date, most of these systems offer a coarse precision of 3–6 m for 90% of estimations [ 4 ], if no additional information/sensors are integrated [ 16 ]. This precision can be acceptable for many location-based services (LBS), which require room-/area-level localization [ 13 , 21 , 22 , 23 ], but not for exact point location. In fact, room-level accuracy has high interest given the number of highly practical applications that can be integrated with low cost techniques and quite extended commercial systems (smartphones, tablets, etc.…”

“…This issue is improved in [ 21 ], where a path-restricted room-level localization system is proposed with additional intelligence, achieving an average success probability of 94%. Furthermore, high success is achieved in [ 22 , 23 ], where RSS-based probabilistic algorithms are applied to refine the room-level tag location based on detecting tag distance to passable boundary points (e.g., at entrances) [ 22 ] or calculating a transition probability between points [ 23 ], so they detect when the tag is likely moving between predefined areas with additional post-processing techniques. According to the experiments, around 95% of success is the average in [ 22 ] and 97% in [ 23 ], respectively, in controlled conditions.…”

“…Several approaches based on different architectures and positioning algorithms [ 4 ] have been proposed to estimate the tag position from the radio map. Among others, the most outstanding algorithms are maximum likelihood (ML) [ 20 , 21 , 22 ], K-nearest neighbors (KNN), weighted KNN (WKNN) [ 11 ] and diverse learning machines, such as support vector machines (SVMs) [ 23 , 24 ], radial basis functions (RBF) or artificial neural networks (ANN) [ 25 , 26 , 27 , 28 , 29 ]. ML, KNN or WKNN consist of comparing gathered RSS vectors with the RSS fingerprints stored in the radio map.…”

User-installable Indoor Positioning System Using a Wi-Fi Beacon and PDR Module

“…Here, the authors describe a basic Wi-Fi fingerprint technique. [4] high discontinuous Visual Marker [5] high discontinuous Light Beacon [6] high discontinuous Ultrasonic Beacon [7] high discontinuous Bluetooth Beacon [8] medium discontinuous IMES [9], [10] low discontinuous Wi-Fi Fingerprint [11], [12] low continuous UWB TOA [13] low continuous Wi-Fi Area Distinction [14] low continuous The Wi-Fi fingerprint technique is an RSSI-based positioning technique. This technique estimates the current position by using an RSSI vector map which is measured by a survey work.…”