Indoor trajectory mapping is essential for applications such as health status monitoring, context-aware computing, and tracking systems. These applications often require accurate location information in environments where the Global Navigation Satellite System (GNSS) is unavailable or unreliable. Infrastructure-free approaches, such as inertial sensor-based Pedestrian Dead Reckoning (PDR), have gained popularity due to their ease of deployment and low cost. However, a key limitation of PDR is not only its susceptibility to cumulative drift over time but also its lack of absolute position references, which prevents reliable alignment of the trajectory with real-world coordinates. To address this challenge, we propose an activity-driven audio-IMU fusion solution that uses Factor Graph Optimization (FGO) to accurately reconstruct indoor walking trajectories on a room-scale floor plan. Specifically, we use two fixed microphone arrays to estimate absolute coordinates from activity-driven audio events. These events act as sparse, high-confidence anchors to map PDR trajectories. We validate our system through experiments in a 6 m × 6 m room. The proposed fusion solution achieves a mean positioning error of 0.31 m, with 68% and 95% of errors below 0.38 m and 0.53 m, respectively. The results demonstrate the potential of sparse, activity-driven audio anchoring to improve infrastructure-free indoor trajectory mapping.
This paper presents the implementation of a localization system in a senior living residence located in a town in the province of Guadalajara, Spain. The system integrates precise location tracking using Ultra-Wideband (UWB) technology in the common areas of the residence, and symbolic positioning via millimeter-wave (mmWave) radar modules installed in private rooms. In UWB-monitored areas, users carry a wearable UWB tag, while in mmWave-radar monitored rooms, a device-free approach is adopted. The localization system is designed to be scalable, easy to deploy, and to support remote maintenance and management, facilitating long-term operation with minimal on-site intervention. This paper provides a detailed description of the devices used, the deployment process, and the overall system setup. Furthermore, by leveraging open-source tools, efficient data management has been achieved. Finally, raw data collected from the implemented system are presented graphically. These data will support future development of tools to detect behavioral anomalies early, which may indicate physical or cognitive decline among the monitored individuals. The collected data are also valuable for identifying daily behavioral patterns and routines, and for detecting deviations that could signal the onset of conditions associated with physical and/or cognitive impairment.
This paper introduces a three-dimensional positioning platform designed to enhance the quality of Korea Telecom's emergency location service by providing floor-level accuracy. Since 2019, the Korea Communications Commission has publicly announced the test results for the positioning quality of mobile operators’ emergency location services. These results are classified by three positioning methods—GNSS, Wi-Fi, and cellular network—in terms of accuracy and time. The main guideline for positioning accuracy is as follows 50 metres horizontal positioning accuracy and 30 seconds total positioning time limit. More recently, the Federal Communications Commission's guideline for vertical positioning requirements, specifying 3 metres of Z-axis accuracy with 90% confidence, has been a subject of discussion among many stakeholders. In order to achieve these quality requirements for emergency location services, Korea Telecom is setting a new challenging accuracy target and working to provide enhanced emergency location services from the perspective of providing end-to-end location information using a floor-level accuracy technique of three-dimensional positioning platform.
This study presents a neural network-based method for static Bluetooth Low Energy (BLE) localization, using phase measurements and carrier frequency per data packet, captured from a single (or multiple) multi-antenna receiver(s). Deterministic phase-based techniques fail to accurately estimate the position of a tag using a single multi-antenna receiver with closely spaced antennas. This is due to the random and unknown carrier phase offset (CPO) at each antenna element, introduced during BLE’s inherent frequency hopping between consecutive packets. Our long short-term memory (LSTM) model learns to handle the unknown distribution of CPO and suppress phase noise, enabling robust localization with just one multi-antenna receiver. Among the tested architectures, the LSTM model showed notable resilience to phase noise and achieved higher accuracy during high-rate tag transmissions compared to the feedforward (FF) and convolutional neural network (CNN) models. All models were trained on simulated or real data and tested on real data from various environments, where they outperformed deterministic techniques—even in cases where those techniques either excelled or failed to estimate the tag’s position.
Indoor Positioning Systems (IPS) are increasingly employed across diverse applications, leveraging technologies such as fingerprinting, Bluetooth, and Wi-Fi. Despite their widespread use, current indoor positioning infrastructures often fall short due to their limited adaptability and inconsistent performance over time, primarily caused by environmental fluctuations. This lack of robustness in dynamic environments presents a significant barrier to effective indoor localization. Addressing these challenges calls for a new approach, one that emphasizes the development of adaptive and responsive systems. However, research and development have predominantly concentrated on fingerprinting techniques and artificial intelligence models, often neglecting the potential of flexible and reconfigurable hardware solutions. In this study, we use an innovative dynamic antenna system tailored to respond to environmental changes, offering an improved indoor positioning method based on Bluetooth Low Energy (BLE) and Angle-of-Arrival (AoA) technologies. Our objective is to evaluate a prior design by characterizing the main features of our receiver, identifying the optimal Field-of-View (FoV), and determining the best incident angles needed for optimal antenna re-orientation. With our approach, we do not only address immediate positioning and coverage needs but also propose a sustainable solution capable of adapting to evolving environments and integration with other systems over time.
Smartphone positioning utilizing a single indoor speaker and its reflections offers the advantage of low deployment cost. However, unlike the direct wave, reflected waves are not always stably detectable, even under Line-of-Sight (LOS) conditions, leading to inconsistent positioning performance. Additionally, associating each detected reflection with its corresponding virtual speaker poses a challenge. This study proposes a method to improve positioning performance by tracking the smartphone based on hypotheses about the identity of each detected reflection. Evaluations were conducted in a real-world environment along both rectangular and circular paths, using only the smartphone’s bottom built-in microphone. The results demonstrated 90th percentile errors of 0.62 m and 0.77 m for the rectangular and circular paths, respectively. This accuracy, approximately equivalent to a step length, suggests potential for navigation applications. These results represent error reductions of 55% and 46%, respectively, compared to conventional approaches.
The global navigation satellite system (GNSS) is a widely used positioning technology that provides heading for moving objects with speed greater than 1 m/s; however, it cannot determine heading for a stationary receiver. The pedestrian dead reckoning (PDR) is a good substitute, especially in the absence of GNSS, due to its independence of external measurements. Pedestrian location and trajectory cannot be effectively estimated by a PDR system without an accurate and precise heading estimation. Heading is a vital part that has a direct effect on the PDR system’s overall accuracy and performance. The complex nature of human movement and the variety of phone orientations and placements make pedestrian heading estimation a persistent challenge. Techniques introduced in the literature often suffer from drift and inaccuracies, particularly when using gyroscope measurements, which might become biased over time. This study presents a novel heading estimation method that utilizes optimum axes mapping through the fusion of accelerometer and magnetometer data. Data is gathered from ten different realistic phone placements, providing a rich dataset that considers a wide range of input patterns. It is evident from the results of the experimental evaluation that the proposed heading estimation outperforms other well-known heading estimation solutions proposed in the literature, including a commercial application. The heading is estimated in a linear constant manner without fluctuations, thus avoiding future drift. Furthermore, the root mean square error (RMSE) is less than 14.67° for all experimental settings, with the lowest value of 1.75° achieved when placing the phone in the left front pants pocket with the screen facing outwards.
In this paper we revisit positioning from Doppler measurements, using techniques from algebraic geometry to produce new theoretical insights and efficient localization algorithms. First, we give a full characterization of the problem, identifying for each problem their minimal configurations, that is the minimum number of receivers and transmitters to have a finite number of solutions. We also study the degree of each configuration. Finally, we show how these algebraic techniques can be applied to obtain an accurate position estimation in challenging environments, such as in the presence of noise and outliers. We also describe how the derived optimized algebraic solver can be integrated in a positioning engine for Low Earth Orbit position, navigation and timing (LEO PNT) systems.
Ultra-Wideband (UWB) indoor positioning systems suffer significant accuracy degradation in Non-Line-of-Sight (NLoS) conditions, where multipath distortions in the Channel Impulse Response (CIR) lead to biased range estimates. While deep learning (DL) models have shown potential in predicting such errors from CIR data, most prior studies focus on isolated architectures and static pipelines, without evaluating their broader integration into full positioning systems. This work presents a comparative study of three DL models—Convolutional Neural Networks (CNN), Bidirectional Long Short-Term Memory networks (BiLSTM), and Transformer—for CIR-based range error prediction, integrated into both Weighted Least Squares (WLS) and Extended Kalman Filter (EKF) frameworks. In WLS, predicted errors are used as adaptive weights to suppress unreliable measurements; in EKF, they dynamically scale the measurement noise covariance for more accurate filtering. We evaluate all models across four realistic indoor environments using a public UWB dataset. Our results show that Transformer-EKF achieves the best performance, reducing mean positioning error to 0.59m in Environment 2 (severe NLoS/multipath) while maintaining real-time inference at 0.059 milliseconds per position. These findings establish a comprehensive benchmark for learning-based UWB positioning and demonstrate the value of fusing data-driven range correction with model-based tracking.
Infrared Local Positioning Systems (IRLPS) offer a cost-effective and accurate alternative for indoor localization, where GNSS signals are typically not available. Despite their potential, IRLPS face significant challenges such as noise, multipath effects, multiple access interference, and calibration requirements, which limit their performance. In response, this work explores the integration of machine learning by proposing a Feedforward Neural Network (FNN) trained on energy measurements collected from a quadrant photodiode. We conducted a thorough analysis of hyperparameters and an ablation study across five neural network topologies and identified configurations that balance accuracy and model complexity. Experimental evaluations in a controlled indoor environment (2.4×2.4×3.4 m3) demonstrate that even simple FNN architectures can generalize well and achieve a high accuracy, with 90% of the positioning errors being below 0.05 m.
This paper presents an algorithm that exploits multipath propagation for the position estimation of mobile receivers. The proposed method utilizes two pieces of information of this multipath signal component: First the delay and thus the path length restricting possible reflection points to an ellipse. And second the Doppler shift of this multipath component to infer angular information along this ellipse. By exploiting relative Doppler information, obtained from the phase difference between the line-of-sight path and multipath components, the approach eliminates the need for strict synchronization requirements. In contrast to state-of-the-art methods that rely on the concept of static virtual transmitters and assume idealized straight wall geometries, the proposed algorithm directly estimates the positions of reflection points. The direct estimation of reflection points allows simultaneous localization and mapping of environments with arbitrary wall shapes. The feasibility of the approach is demonstrated through simulations incorporating delay and Doppler measurements of multipath components. Results confirm that the method enables accurate estimation of both receiver position and reflection points for a variety of wall geometries, including convex and concave surfaces.
Wi-Fi-based indoor positioning applications usually employ methods based on the Received Signal Strength Indicator (RSSI), an indicator of the signal attenuation. However, relying on this indicator has accuracy and stability limitations, hampering the performance of indoor positioning systems. This work studied the indoor positioning based on Wi-Fi Round Trip Time, which employs the signal’s time-of-flight to accurately measure the distance between compatible devices. We deployed four RTT-compatible access points in a research laboratory and collected RTT measurements from two compatible Android smartphones on 20 reference locations. Then, we compared the centroid, Euclidean-based k-Nearest Neighbours (KNN) and a KNN with an adapted Euclidean distance for RTT data for position estimation using the collected dataset. We also explored the effect of using individual and time-aggregated RTT measurements. The results of the study show that a 1s time aggregation improves the mean errors of the KNN-based methods and that the adapted Euclidean distance provides the best mean positioning errors (0.4−0.6m).
The rapid growth of 5G and future 6G networks requires efficient and scalable radio access network (RAN) densification, especially in dense urban and industrial areas. Traditional planning uses manual surveys and simple propagation models, but these lack spatial accuracy and adaptability. Stochastic RF simulation tools often fail to model real-world conditions, such as material properties and geometry. This leads to poor site selection, higher costs, and rollout delays.This paper proposes an AI-based framework that combines high-resolution 3D modeling, Digital Twin technology, and deterministic ray tracing. It uses aerial and ground imagery to build detailed 3D models, enhanced with object detection and material classification through segmentation models. These models enable automatic feature extraction for RF simulation and planning. The system uses open-source 3D tools, vision transformers for segmentation, and a simulation engine with antenna radiation patterns and material-aware propagation. Tests in urban and campus settings show better prediction accuracy, less manual work, and lower costs than traditional methods. Results show that AI and Digital Twins improve and automate network deployment.
The magnetic field odometry, as a research hotspot in indoor non-matching magnetic positioning, relies heavily on the accuracy of real-time indoor magnetic field modeling. However, existing mainstream magnetic field modeling methods face limitations in local magnetic anomaly scenarios, making it challenging to maintain modeling precision. To address this issue, this paper proposes an innovative indoor magnetic field modeling approach based on equivalent magnetic dipoles. This method constructs equivalent magnetic sources using parameters such as magnetic moments and locations, accurately reflecting the physical characteristics of the magnetic sources. It features a simple model structure, efficient real-time computation, and rapid magnetic field inversion. By superposing multiple magnetic dipoles, the model can approximate complex magnetic field distributions, thus enabling high precision magnetic field prediction. Experimental results demonstrate that the proposed equivalent magnetic dipole model outperforms traditional polynomial models in indoor environments, reducing the root mean square error (RMSE) of magnetic field prediction by approximately an order of magnitude within the modeled region, thereby significantly enhancing the spatial characterization accuracy of the magnetic field.
Beacons are key components in many location systems based on Bluetooth Low Energy (BLE), as they allow estimating the position of objects or people from the received signal strength (RSSI). Although there are many commercial solutions, they often have limitations in terms of the configuration of key parameters, such as transmission power or the advertising channel, which makes it difficult to adapt them to different environments or specific requirements. This work presents the characterisation of a beacon developed with a flexible and customisable approach, designed to facilitate its use in both experimental environments and practical applications. This beacon allows for easy modification of essential parameters via software, such as the channel(s) used for broadcasting the BLE advertisements, and has a strategically placed antenna to improve signal quality and stability. The characterisation includes a series of tests and measurements to analyse the signal behaviour in different scenarios. The results obtained show a higher RSSI stability and a better use of the emitted power, which can be translated into significant improvements in localisation algorithms based on this metric.
Logistics warehouses have struggled with labor shortages, but the inbound processes remain particularly human-powered. Worker location data is a key to higher productivity in such cases. Fixed cameras are a promising tool for localization, as they also offer valuable environmental information such as package status. However, identifying individuals from visual data alone is often impractical. To enable identity-aware localization, prior studies have attempted to identify people in videos by associating their trajectories with wearable sensor measurements. Although this appearance-independent approach has several advantages, existing methods may fail under real-world conditions. Therefore, we propose CorVS+, a novel data-driven person identification framework based on the correspondence between visual tracking trajectories and sensor measurements. Firstly, our deep learning model predicts the correspondence probabilities and reliabilities for every pair of a trajectory and sensor measurements. Secondly, our algorithm matches the pairs over time based on the model predictions. We developed a dataset comprising 27 hours of sensor measurements and 38 km of trajectories in a warehouse. This dataset covers actual activities and challenging situations, such as multiple stationary workers inspecting items. The evaluation indicated the superiority of CorVS+ over existing methods and the effectiveness of its unique designs for industrial-scale settings. The model and dataset will be available at https://doi.org/10.5281/zenodo.17745683.
Conventional Ultra-Wideband (UWB) localization systems typically require multiple anchors for effective tag localization. However, recent studies have demonstrated that multi-path components (MPCs) can provide additional information for indoor localization by introducing the concept of virtual anchors (VAs), even with a single-anchor setup. This work proposes a novel single-anchor UWB simultaneous localization and mapping (SLAM) system that requires no prior information on the anchor position or floor plan. The system uses measurements of the phase difference of arrival (PDOA) by the tag, and by fusing distance and PDOA estimations with inaccurate tag movement measurements the system refines the tag trajectory and estimates the anchor and VA positions. Simulations and experiments demonstrate that the proposed system can accurately estimate wall positions as well as significantly improve the tag localization.
Developing an Indoor Positioning System (IPS) for an industrial environment is a challenging task in its attempt to balance accuracy, cost, maintainability, and scalability. In many cases, sub-meter accuracy is not essential, allowing the use of less accurate yet more cost-effective technologies, such as Received Signal Strength Indicator (RSSI)-based positioning using Bluetooth Low Energy (BLE). While the common approach involves assigning one BLE beacon per target, utilizing multiple beacons can enhance both accuracy and system robustness, with no significant extra cost. We investigate the performance differences between single-beacon and multi-beacon setups using fingerprinting and multilateration algorithms in the context of a real-world. Our results show that fingerprinting with multiple beacons yields the best performance, achieving an average accuracy of 2.18m. Although multilateration offers slightly lower accuracy, it avoids the need to build and maintain a Radio Map, providing a balance between performance and deployment simplicity.
Indoor localization, despite its high demand for reliable, accurate, and low-cost positioning in location-based applications, often faces limitations due to data acquisition challenges. This study focuses on improving indoor localization accuracy by mitigating errors inherent in Received Signal Strength Indicator (RSSI) data. A novel indoor localization model is introduced, leveraging Convolutional Neural Network (CNN), VGG16, AlexNet -based Fingerprinting. This approach aims to extract pertinent features directly from RSSI data and map these features to specific location coordinates. By employing this technique, the study seeks to achieve more efficient and accurate indoor localization. A dedicated Fingerprinting dataset was constructed, comprising RSSI measurements from multiple access points collected at known locations. This raw RSSI data was then transformed into greyscale images to ensure compatibility with the CNN, VGG16 and AlexNet model inputs. Recognizing that standard CNNs are primarily designed for image classification tasks, the study emphasizes the need for careful optimization and evaluation of these Wi-Fi sample-based images to maximize model performance in indoor localization. To validate the proposed approach and the comparative models, we conducted real-time experiments in a controlled indoor environment. The results showed that the Modified VGG16, Modified AlexNet, CNN model achieved a loss of 0.13, 0.30, 0.66 respectively and an accuracy of 96%, 88%, 83% respectively outperforming the other models. This highlights the effectiveness of the specifically designed CNN, VGG16 and AlexNet architecture in extracting and utilizing relevant features from RSSI data for accurate indoor positioning. The performance of this proposed model is rigorously compared to state of art, adapted to address the same indoor localization problem
