Hybrid Blended WiFi Fingerprint Indoor Localization Using Multi-Task Learning and Feature-Space WKNN

Indoor localization finds wide-ranging applications in user navigation and intelligent building systems. Nevertheless, signal interference within complex indoor environments and challenges regarding localization generalization in multi-building and multi-floor scenarios have restricted the performance of traditional localization methods based on Wi-Fi fingerprinting. To tackle these issues, this paper presents the SE-MTCAELoc model, a multi-task convolutional autoencoder approach that integrates a squeeze-excitation (SE) attention mechanism for indoor positioning. Firstly, the method preprocesses Wi-Fi Received Signal Strength (RSSI) data. In the UJIIndoorLoc dataset, the 520-dimensional RSSI features are extended to 576 dimensions and reshaped into a 24 × 24 matrix. Meanwhile, Gaussian noise is introduced to enhance the robustness of the data. Subsequently, an integrated SE module combined with a convolutional autoencoder (CAE) is constructed. This module aggregates channel spatial information through squeezing operations and learns channel weights via excitation operations. It dynamically enhances key positioning features and suppresses noise. Finally, a multi-task learning architecture based on the SE-CAE encoder is established to jointly optimize building classification, floor classification, and coordinate regression tasks. Priority balancing is achieved using weighted losses (0.1 for building classification, 0.2 for floor classification, and 0.7 for coordinate regression). Experimental results on the UJIIndoorLoc dataset indicate that the accuracy of building classification reaches 99.57%, the accuracy of floor classification is 98.57%, and the mean absolute error (MAE) for coordinate regression is 5.23 m. Furthermore, the model demonstrates exceptional time efficiency. The cumulative training duration (including SE-CAE pre-training) is merely 9.83 min, with single-sample inference taking only 0.347 milliseconds, fully meeting the requirements of real-time indoor localization applications. On the TUT2018 dataset, the floor classification accuracy attains 98.13%, with an MAE of 6.16 m. These results suggest that the SE-MTCAELoc model can effectively enhance the localization accuracy and generalization ability in complex indoor scenarios and meet the localization requirements of multiple scenarios.

Indoor positioning systems are crucial for public safety, healthcare, and IoT, but Wi-Fi fingerprinting faces challenges such as signal interference, multipath effects, and high computational costs. These issues reduce positioning accuracy and make real-time localization difficult.This paper introduces an Ant Colony Optimization (ACO)-based dual clustering method to enhance Wi-Fi fingerprinting accuracy and efficiency. ACO performs coarse clustering by optimizing initial data groupings, while K-means refines clusters for improved precision. The Weighted K-Nearest Neighbor (WKNN) algorithm is then applied for real-time positioning by selecting the most similar signal sub-bases.Experiments show that the proposed method achieves 100% accuracy in building classification and 91% accuracy in floor classification. For latitude and longitude prediction, Random Forest and SVC outperform XGBoost, achieving MSE values of 0.0048 (latitude) and 0.0055 (longitude). The approach also reduces computational overhead by 93.51%, improving efficiency.The study presents a robust, scalable solution for indoor positioning and introduces the Dual Clustering Wi-Fi Localization Dataset (DCWiLD) for future research. Future work will focus on dataset balancing, BLE/UWB integration, and energy optimization for IoT applications.

Feature selection on database optimization for Wi-Fi fingerprint indoor positioning

This paper presents a hybrid indoor positioning system based on the combination of IEEE 802.11 (Wi-Fi) fingerprinting and magnetic matching (MM) which uses the built-in sensors on a smartphone. These sensors have made smartphones become ubiquitous portable devices providing not only communication services in everyday life but also personal positioning uses. The main reason of using Wi-Fi signals as fingerprints is that Wi-Fi access points are commonly distributed in indoor environments and are the basic devices in smartphones. On the other hand, the concrete building frame causes perturbations in indoor magnetic field and thus formed in each building unique distribution. We can then make use of this unique magnetic distribution through a specific algorithm to acquire a more accurate indoor positioning result. In general, MM results might have small errors on some occasions, and it might suffer from significant mismatches. Hence, it is important to utilize other technologies to detect these mismatches. If two positioning techniques could be integrated, then the combined positioning performance could be improved than that of Wi-Fi fingerprinting or MM alone. The concept of the Wi-Fi fingerprinting is to use the received signal strength (RSS) value as a tag of the user position, and it requires two stages to carry out the fingerprinting method including the calibration (i.e., training) stage and the verification (i.e., positioning) stage. In this paper the positioning algorithm for the Wi-Fi fingerprinting algorithm is the k-weighted nearest neighbors (KWNN) algorithm. On the other hand, the concept of MM is similar to that of Wi-Fi fingerprinting, which is divided into two steps: first, the offline training step, a set of reference points with known coordinates and the corresponding magnetic intensities are stored into the. Second, the positioning step is implemented to find the optimal match between the features of the measured magnetic profile and the database. Since MM is a profile-matching method, we utilize the dynamic time warping algorithm to address the real-time step-length estimation. All the work is implemented on an off-the-shelf portable smart device, and the 1st floor of Department of Aeronautics and Astronautics building at National Cheng Kung University as well as a warehouse test field in Tainan City, Taiwan are used as examples to demonstrate this proposed hybrid indoor positioning system. The experimental results including the positioning error statistics of the Wi-Fi fingerprinting, MM, and Wi-Fi fingerprinting/MM integration are evaluated in the paper. As shown in the experimental results, the positioning performance of this proposed hybrid indoor positioning system is improved than that of Wi-Fi fingerprinting or MM alone. Keywords: Wi-Fi fingerprinting, magnetic matching, indoor positioning system, dynamic time warping.

Large buildings grow rapidly nowadays, indoor positioning technology is becoming increasingly important. Satellite navigation system, though very mature in the outdoors, loses the ability to locate indoors due to the block by the buildings. The combination of widely covered Wi-Fi signals and independent inertial navigation system is proposed to solve the indoor positioning. In order to improve the accuracy of indoor positioning, an indoor positioning method based on Wi-Fi fingerprint and inertial navigation technique is proposed in this paper. The motion state and the position information are obtained by an inertial sensor including a gyroscope, an acceleration sensor and a magnetometer, the dynamic adjustment threshold algorithm is proposed to determine the walking attitude of pedestrians, the walking direction is obtained by using the direction detection algorithm, the particle filter algorithm with improved map matching is used to calculate the pedestrian position, and the inertial sensor navigation information combined with Wi-Fi fingerprint location information is used to correct the position of pedestrians and reduce the position error. Simulation experiments were carried out to verify the effectiveness and practicability of the algorithm on the 4th floor of the new main building of Beihang University.

With the development of urbanization in recent years, the conditions for building the smart city gradually mature, the increase in large airports, shopping centers and office buildings makes indoor positioning and navigation services more important. The satellite signal in indoor environment is too weak to locate, while notebook computers, mobile phones and other intelligent terminals greatly increase, and Wi-Fi signals become more dense and extensive, which provide the material foundation for the development of indoor positioning technology. In order to solve the indoor positioning problem, we propose an indoor positioning and navigation technique based on Wi-Fi fingerprint and environment information. The user’s intelligent terminal can detect the signal strength of each wireless access point and obtain the received signal strength indication (RSSI), then the specific location of the device can be calculated according to the indoor radio pass loss model and Wi-Fi signal strength fingerprint database composed by each node. The combination of k nearest neighbor (k-NN) algorithm and particle filter (PF) algorithm helps to improve the positioning accuracy and robust stability in the paper, and the combination of the Dijkstra algorithm and Wi-Fi fingerprint database based on reference nodes provides the optimal navigation route by calculating the shortest path from the start position to destination. Simulation experiment was carried out on the 4th floor of new main building of Beihang University, from which we obtain the location map and the navigation route map from the starting point to destination. The experiment result shows the efficiency and reliability of the algorithm.

A scalable indoor localization technique is a vital technology for future large-scale location-aware services covering a complex of multi-story buildings. Our research on the usage of ResNet for scalable building/floor categorization and floor-level position estimation based on Wi-Fi fingerprinting is presented in this publication. Building and floor-level coordinates are estimated using our new ResNet architecture, which utilizes a stacked autoencoder to reduce feature space and a feed-forward classifier to classify multiple labels of building/floor/location. This architecture is the foundation for our multi-building and multi-floor indoor localization system based on Wi-Fi fingerprinting. On the Jaume I University (UJI) campus in Spain, we test the accuracy of building/floor estimation and floor-level coordinates estimation for three buildings with four or five stories each. ResNet-based indoor localization using a single ResNet has been proven to be feasible, with results that are close to the state of the art. One ResNet is all that is needed in order for the proposed indoor localization system based on Wi-Fi fingerprinting to function at levels close to the current state of the art, allowing it to be implemented with less complexity and less energy consumption on mobile devices.

Indoor localization has gained significant attention recently due to the growing demand for location-aware applications within indoor environments. Among various techniques, Wi-Fi fingerprinting combined with the Internet of Things (IoT) has emerged as a promising solution for accurate and cost-effective indoor localization. This paper surveys the existing research related to indoor localization using Wi-Fi fingerprinting with the IoT, aiming to provide a comprehensive understanding of the advancements, challenges, and potential applications in this field. The paper begins by introducing the fundamental concepts of indoor localization and the role of Wi-Fi fingerprinting in achieving accurate position estimation. In addition, this paper focuses on the contributions of Artificial Intelligence (AI), Machine Learning (ML), and Deep Learning (DL) in improving localization accuracy, robustness, and scalability. The reviewed papers have been examined from various aspects, including system architecture, deployment strategies, fingerprint creation techniques, and localization algorithms, with a discussion about the advantages and limitations of each approach. In addition to discussing the state-of-the-art techniques, this paper identifies research gaps and open challenges in indoor localization using Wi-Fi fingerprinting with the IoT. The findings presented in this paper can guide future research efforts, leading to the development of intelligent and context-aware IoT applications within indoor environments.

With the unprecedented demand of location-based services in indoor scenarios, wireless indoor localization is emerging as an essential application for mobile users. While the line-of-sight GPS signal is not available at indoor spaces, WiFi fingerprinting using received signal strength (RSS) has become popular with its ubiquitous accessibility. Although the fingerprinting data can be easily collected by portable mobile devices, to achieve robust and efficient indoor localization remains challenging with two constraints. First, the localization accuracy will be degraded by the random fluctuation of signals that caused by multipath effects from RSS signals. Second, indoor localization algorithms are time-consuming due to the handcrafting features and complex filtering on raw dataset. To achieve high localization accuracy with WiFi fingerprinting, in this paper, we propose CapsLoc, a robust indoor localization system by using capsule networks. Specifically, the capsule network model can efficiently extract hierarchical structures from WiFi fingerprint with three main components, including a convolutional layer, a primary capsule layer and a feature capsule layer. We conduct a real-world experimental field test with over 33600 data points. The experimental results show that CapsLoc can achieve accurate indoor localization with an averaged error of 0.68 m, which outperforms conventional machine learning methods (KNN and SVM) and existing deep learning methods (CNN and SAE-CNN).

Wireless Fidelity (Wi-Fi) fingerprinting is a remarkable approach developed by modern science to detect the user’s location efficiently. Today, the Global Positioning System (GPS) is used to keep track of our current location for outdoor positioning. In GPS technology, satellite signals cannot reach indoor environments as they are shielded from obstructions so that indoor environments with a lack of Line of Sight (LoS) do not provide enough satellite signal accuracy. Since indoor environments are very difficult to track, thus, a wide variety of techniques for dealing with them have been suggested. The best way to offer an indoor positioning service with the current technology is Wi-Fi since most commercial infrastructure is well equipped with Wi-Fi routers. The use of Wi-Fi fingerprinting techniques is becoming more common in the field of indoor positioning systems (IPS). For the purpose of indoor location localization, each method for obtaining a Wi-Fi fingerprint has been analyzed and discussed in this work. In this study, the majority of the algorithms that are associated with Wi-Fi fingerprinting have been interpreted, and the earlier works of other researchershave been critically evaluated, in order to get a more in-depth understanding of how the Wi-Fi fingerprinting process works.

BackgroundOne of the key technologies for future large-scale location-aware services covering a complex of multi-story buildings is a scalable indoor localization technique. In this paper, we report the current status of our investigation on the use of deep neural networks (DNNs) for the scalable building/floor classification and floor-level position estimation based on Wi-Fi fingerprinting. Exploiting the hierarchical nature of the building/floor estimation and floor-level coordinates estimation of a location, we propose a new DNN architecture consisting of a stacked autoencoder for the reduction of feature space dimension and a feed-forward classifier for multi-label classification of building/floor/location, on which the multi-building and multi-floor indoor localization system based on Wi-Fi fingerprinting is built.ResultsWe evaluate the performance of building/floor estimation and floor-level coordinates estimation of a given location using the UJIIndoorLoc dataset covering three buildings with four or five floors in the Jaume I University (UJI) campus, Spain. Experimental results demonstrate the feasibility of the proposed DNN-based indoor localization system, which can provide near state-of-the-art performance using a single DNN.ConclusionsThe proposed scalable DNN architecture for multi-building and multi-floor indoor localization based on Wi-Fi fingerprinting can achieve near state-of-the-art performance with just a single DNN and enables the implementation with lower complexity and energy consumption at mobile devices.

The location-based services for Internet of Things (IoTs) have attracted extensive research effort during the last decades. Wi-Fi fingerprinting with received signal strength indicator (RSSI) has been widely adopted in vast indoor localization systems due to its relatively low cost and the potency for high accuracy. However, the fluctuation of wireless signal resulting from environment uncertainties leads to considerable variations on RSSIs, which poses grand challenges to the fingerprint-based indoor localization regarding positioning accuracy. In this paper, we propose a top-down searching method using a deep reinforcement learning agent to tackle environment dynamics in indoor positioning with Wi-Fi fingerprints. Our model learns an action policy that is capable to localize 75% of the targets in an area of 25000m2 within 0.55m.

Indoor Positioning System (IPS) is a technology that provides precise and accurate location information in indoor environments. Since Global Positioning system (GPS) relies on line of sight communications with a satellite transceiver, it does not work within buildings. To overcome this challenge, IPS technology could use RF based communication and locate humans and objects indoors. The key advantage of RF based communication is its high coverage range and non-line of sight communications. Wi-Fi finger printing is an efficient RF based technique for IPS which utilizes signal strength of wireless LAN to identify different indoor locations. In this paper, we develop an experimental testbed to implement Wi-Fi fingerprinting based IPS. We analyse the performance of Wi-Fi fingerprinting based IPS for three different indoor scenarios. Results show that Wi-Fi finger printing based IPS accurately predicts user location when the multi-path fading is low and access points are closer to the user.

Knowing self location matters a lot in people's daily life. While Global Positioning System (GPS) provides almost perfect solution for outdoor area, it would not work in indoor areas because of no line-of-sight to satellites. However, since human tend to spend more and more time in complexly constructed buildings, helping people localize themselves in indoor space become a critical problem. Raising localization accuracy and reducing deployment cost are two main objects in indoor localization problem. High localization accuracy ensures usability of service, while low deployment cost lessens the effort people must take to use localization service. Numerous technologies have been proposed to tackle the problem. However, practical indoor localization that can provide high localization accuracy with minimum cost is still a vacancy. In the first part of this thesis, we devote ourself to explore the possibility of fingerprint based localization. Although a large number of fingerprinting based indoor localization systems have been proposed, our field experience with Google Maps Indoor (GMI), the only system available for public testing, shows that it is far from mature for indoor navigation. Motivated by the obtained insights from field studies with GMI, we propose GROPING as a self-contained indoor navigation system independent of any infrastructural support. GROPING relies on Ambient Magnetic Field fingerprints, which is formed by ``twisted'' geomagnetic field by building structures, that are far more stable than WiFi fingerprints, and it exploits crowdsensing to construct floor maps rather than expecting individual venues to supply digitized maps. Based on our experiments with 20 participants in various floors of a big shopping mall, GROPING is able to deliver a sufficient accuracy for localization and thus provides smooth navigation experience. In our experiments with ambient magnetic field fingerprint, we see that scalability of ambient magnetic field based approach is not satisfactory comparing to WiFi based approaches. By further exploration, we find that dual properties naturally existed in ambient magnetic field fingerprint and WiFi fingerprint. Therefore based on GROPING, we present MaWi - a dual-sensor enabled indoor localization system in the second part of this thesis. Central to MaWi is a novel framework combining two self-contained but complementary localization techniques: Wi-Fi and Ambient Magnetic Field. Combining the two techniques, MaWi not only achieves a high localization accuracy, but also effectively reduces human labor in building fingerprint databases: to avoid war-driving, MaWi is designed to work with low quality fingerprint databases that can be efficiently built by only one person. Our experiments demonstrate that MaWi, with a fingerprint database as scarce as one data sample at each spot, outperforms the state-of-the-art proposals working on a richer fingerprint database. Although MaWi is designed to use minimum human effort to collect fingerprints, the initial spot surveying is still an inevitable burden for all fingerprint-based localization systems. To ultimately reduce human effort in initial phase, we try to find a localization solution in a model-based methodology. In the last part of this thesis, we focus on exploiting ``multipath" phenomenon in wireless signal propagation and utilize it to fully or partially reconstruct the geometry of the indoor space, as well as locate signal source. Whereas a few physical layer techniques have been proposed to locate a signal source indoors, they all deem multipath a ``curse'' and hence take great efforts to cope with it. We, on the contrary, deem multipath a ``blessing'' and thus innovatively exploit the power of it. Essentially, with minor assumption (or knowledge) of the geometry of an indoor space, each signal path may potentially contribute a new piece of information to the location of its source. As a result, it is possible to locate the source with very few sensors (most probably just one hand-held device). At the same time, the extra information provided by multipath effect can help to fully or partially reconstruct the geometry of the indoor space, which enables a floor plan generation process missing in most of the indoor localization systems. To demonstrate these ideas, we instrument a USRP-based radio sensor prototype named iLocScan; it can simultaneously scan an indoor space (hence generate a plan for it) and position the signal source in it. Through iLocScan, we mainly aim to showcase the feasibility of harnessing multipath in assisting indoor localization, rather than to rival existing proposals in terms of localization accuracy. Nevertheless, our experiments show that iLocScan can offer satisfactory results on both source localization and space scanning.

Introduction:An important and complicated problem related to the multilateration of Wi-Fi or Bluetooth Low Energy signals as well as Wi-Fi fingerprinting is the procedure of infrastructure adjustment which includes Wi-Fi radio map construction and Wi-Fi or Bluetooth Low Energy radio signal path loss model calibration.Purpose:Developing a method for navigation and Wi-Fi radio map construction, which would provide user’s indoor navigation, Bluetooth Low Energy path loss model calibration and Wi-Fi radio map collaborative semi-automatic construction.Results:The paper presents a collaborative semi-automatic Wi-Fi radio map construction method based on the combination of Bluetooth Low Energy multilateration, Wi-Fi fingerprinting, Wi-Fi radio map collaborative semiautomatic construction procedure and semi-automatic Bluetooth Low Energy path loss model calibration. For the semi-automatic calibration procedure of the Bluetooth Low Energy signal propagation model and for the method of collaborative semi-automatic construction of Wi-Fi radio map and indoor navigation, a calibration algorithm and an algorithm of collaborative semi-automatic construction of Wi-Fi radio map and indoor navigation were proposed, respectively. A mobile application has been developed which implements the proposed algorithms in order to test them and analyze the results.Practical relevance:The proposed method allows you to avoid time-consuming procedures of constructing a map of Wi-Fi radio signals and semi-automatic calibration of Bluetooth Low Energy signal propagation in the offline phase.