WiFi-based indoor localization is essential for asset tracking, healthcare monitoring, and smart buildings. However, existing systems face challenges such as data variability, environmental noise, and difficulty detecting floor and building levels, compounded by limited labeled data and high received signal strength (RSS) collection costs. This paper introduces quantum stochastic contrast learning (QSCL), a novel framework grounded in rigorous theoretical foundations. We present four theorems and one lemma that establish the probabilistic augmentation, diversity enhancement, relationship preservation, and resilience of QSCL under quantum noise, supported by formal proofs. Leveraging these foundations, QSCL utilizes quantum computing (QC) to generate strong data augmentations with stochastic perturbations, enhancing data diversity, while classical weak augmentations provide subtle variations for robust feature learning. We propose a spatial temporal adaptive attention (STAA) encoder that integrates convolutional layers with adaptive attention mechanisms to capture spatial and temporal dependencies in sequential data. Furthermore, a bidirectional contrastive loss function is introduced to capture forward and reverse relationships between augmented views, ensuring robust representations. Comprehensive evaluations on the UJIIndoorLoc and UTSIndoorLoc datasets validate QSCL, demonstrating superior performance with reduced labeled data and resilience to quantum noise such as bit-flip, dephasing, and measurement noise. The proposed framework significantly improves localization accuracy, floor and building detection, and generalizability in challenging indoor environments.

Accurate indoor localization and navigation enable real-time, ubiquitous, location-based services. Over the past decade, data-driven approaches for inertial odometry have shown the potential to enhance indoor positioning accuracy. However, low-cost inertial measurement units (IMUs), commonly used in smartphones and IoT devices, are prone to significant noise, leading to drift and degraded performance in navigation algorithms. This article presents a novel, lightweight, and real-time end-to-end framework, DeepILS Brossard et al., (2020), designed to process raw inertial data for precise pedestrian localization in indoor environments. DeepILS utilizes a residual network enhanced with channel-wise and spatial attention mechanisms, enabling accurate velocity and position estimation across diverse motion dynamics. The framework's effectiveness is validated using four benchmarks and two newly introduced datasets in real-time edge scenarios. These datasets were collected across diverse indoor environments at the KAIST campus and Incheon National Airport, using multiple hardware platforms, including the KAIST IoT positioning module and Android smartphones. Experimental results, including tests on unseen data and comprehensive ablation studies, demonstrate that DeepILS improves localization accuracy by 70% compared to state-of-the-art methods while effectively mitigating sensor noise and enhancing robustness in real-world environments. Specifically, DeepILS exhibits excellent edge performance on IoT devices, making it highly suitable for real-time applications.

Collaborative indoor positioning requires maps with explicitly defined spatial connectivity. This paper presents a framework for constructing indoor map layouts using spatial entities and reachability structures. We introduce the Floor-Group-Area (FGA) Matrix Problem to visualize and optimize layout connectivity through column reordering. The framework exports 1bpp OGMs and spatial metadata in JSON. In user evaluations, trained participants completed full layouts of a four-story museum in 20 minutes, with all computationally intensive algorithms running in few seconds on a standard PC.

The integration of various sensors in smartphones has enabled the use of sensor fusion for indoor positioning. Sensor fusion frameworks, which integrate data from multiple sensors in a smartphone, are being recognized for their potential to improve indoor positioning accuracy. However, real-time errors arising from the dynamic indoor environment can accumulate, posing challenges in detecting and correcting these errors to maintain precise indoor navigation services. In this article, we propose a novel sensor fusion framework that achieves high positioning accuracy by learning real-time errors accumulated during pedestrian navigation. The proposed system identifies sensor measurement errors, accumulates them, adjusts measurement values based on the accumulated error distribution, and employs these refined data for indoor position estimation. Additionally, we introduce a new technique to detect and exclude anomalous errors during the error adjustment process. The analyzed error information is subsequently used to update the radio map through long-term memory learning during the offline stage, ensuring rapid convergence for our proposed system. In experimental scenarios, the proposed framework achieved an average error distance (AED) of approximately 1.68 m, which is a significant improvement after error correction.

Accurate localization is critical for a wide range of applications, including navigation, augmented reality, safety management, and healthcare. However, achieving accurate positioning remains challenging, especially in GPS-degraded environments. Pedestrian Dead Reckoning (PDR), which estimates position based on inertial sensor data, offers a promising alternative but typically demands substantial effort in algorithm design and system integration. This complexity often leads to prolonged development timelines and elevated costs. To address these challenges, we present the PDR Sensor (PDRS), a plug-and-play, black-box PDR module optimized for seamless integration into wearable and IoT platforms. PDRS integrates a microprocessor and an inertial measurement unit (IMU) within a compact, low-power module. It executes real-time sensor fusion onboard, providing position updates via standard interfaces ((IC)-C-2, SPI, UART). Experimental evaluations demonstrate its high robustness, achieving a step detection error of just 0.16% over 1250 steps and a traveled distance error of 1.87% under diverse conditions. The system further exhibits low latency (2.5 ms) and modest power consumption (52.35 mA). Unlike conventional studies, the PDRS leverages a holistic co-design of hardware and software to enhance performance and integration aspects, simplifying adoption without requiring deep expertise. By lowering technical barriers, PDRS enables developers and researchers to prioritize innovation in IoT and wearable technologies, accelerating prototyping, reducing development time, and advancing industry progress. Future work aims to enhance accuracy, reduce power usage and module size, and integrate AI-driven techniques to better mitigate drift.

Deep learning-based inertial navigation remains a formidable challenge due to the intricate temporal dynamics of human motion and the stringent computational constraints of edge devices. This study introduces NanoMST, a highly efficient multiscale transformer architecture designed for precise pedestrian inertial motion tracking with minimal computational overhead. The proposed model integrates a hierarchical multiscale embedding strategy with a scale-adaptive attention mechanism, effectively capturing motion patterns across diverse temporal resolutions while optimizing efficiency through hardware-aware quantization. With 298K parameters and 7.59M floating-point operations, Nano Multiscale Transformer (NanoMST) achieves performance comparable to substantially larger models while maintaining an exceptionally low computational burden. Extensive evaluations on benchmark datasets, including OxIOD, robust neural inertial navigation (RoNIN), and RIDI, yield average trajectory errors of 2.68m on RoNIN, 1.64m on RIDI, and 1.80m on OxIOD. The quantized 8-bit implementation reduces the model size from 1.23MB to 0.41MB while retaining 94% of the original model's accuracy. Profiling on edge devices confirms real-time feasibility, with inference latencies ranging from 0.18 to 0.96 ms across various smartphone generations and an average throughput exceeding 6000 samples per second, surpassing contemporary architectures, such as IMUNet and CTIN. This study illustrates an efficient engineering approach for deep learning-based inertial tracking, demonstrating that high-precision sequential motion estimation can be achieved with a minimal computational footprint. The efficiency and real-time capability of NanoMST make it particularly suitable for deployment in resource-constrained environments, including mobile, wearable, and Internet of Things applications.

Indoor-outdoor detection (IOD) has gained prominence recently, particularly in positioning technology, leveraging smartphone-embedded sensors. It is pivotal in pedestrian localization, activity recognition, transportation mode classification, and power management of Internet of Things (IoT) devices. While several approaches have been explored for IOD, including threshold-based methods and machine learning-based models, challenges remain in addressing these models’ temporal variations and computational complexities. Supervised learning approaches heavily rely on labeled datasets, which are costly and time-consuming to synthesize. We propose TabCLR, the first self-supervised learning (SSL) framework for IOD, to overcome these challenges. TabCLR utilizes contrastive learning representation tailored for tabular data classification using smartphone inertial sensors. It comprises data augmentation, a novel encoder network with self-attention, and an optimized contrastive loss function. Evaluation of TabCLR on multiple indoor-outdoor datasets demonstrates its superiority in both supervised and semi-supervised classification compared to existing methods. Notably, TabCLR outperforms SCARF by 6%-7%, indicating its effectiveness in capturing temporal feature representation patterns. Visualization analysis further illustrates TabCLR’s distinctive clustering of feature embeddings compared to SCARF. TabCLR represents a significant advancement in SSL methodologies for indoor-outdoor detection classification. Its robust performance showcases its potential to enhance accuracy in indoor-outdoor integrated GPS systems, addressing critical challenges in IOD classification.

Wearable sensor technology emerges significantly in sensing physiological data for real-time health monitoring, showing promise for early disease detection by continuously monitoring vital signs like body temperature, heart rate, and electrodermal activity. However, current emotion recognition methods are limited by high measurement noise in wearable health sensors, negatively impacting performance. Additionally, most classification models are too complex to be deployed on IoT devices for real-time performance. This paper presents EmoHeal1, a lightweight deep learning-based fusion architecture to process wrist sensor-based physiological sensor data for IoT devices. EmoHeal integrates data augmentation, attention mechanisms, Temporal Convolutional Networks (TCNs), and Gated Recurrent Units (GRUs) for robust emotion classification. Each sensor's data is processed through a separate TCN block that extracts temporal features using dilated convolutions. These blocks are enhanced by channel and spatial attention mechanisms to emphasize significant features. The extracted features are then fused and fed into a bidirectional GRU, which captures long-term dependencies in the wrist sensor data. EmoHeal demonstrated excellent performance on the K-EmoCon dataset, achieving a mean accuracy of similar to 72% arousal for an edge-deployable emotion recognition model, surpassing state-of-theart models. This research significantly benefits real-time mobile health sensing in Internet of Things (IoT) applications.

Modern inertial measurement units (IMUs) in smartphones, while offering a pathway to innovative indoor positioning solutions, often suffer from significant drift errors, with typical error margins that preclude localization within a three-meter threshold. This limitation severely hinders the practical deployment of indoor positioning systems in critical areas such as shopping malls, hospitals, and office buildings. Traditional methods to enhance localization accuracy typically involve integrating complex sensor fusion technologies, such as UWB, Wi-Fi, or Bluetooth, further complicating such solutions’ adoption and scalability. Additionally, traditional deep learning techniques for enhancing IMU performance face challenges related to privacy protection, data collection, and training efficiency. To address these issues, we propose FedNav, a novel federated learning (FL)-based inertial localization approach that leverages IoT client devices’ computational power to enhance IMU performance while adhering to stringent privacy requirements. FedNav significantly improves pedestrian localization accuracy by reducing the average positioning error, ranging from 5% to over 25% on benchmark datasets and state-of-the-art models. It ensures robust privacy protection and efficient data utilization without additional hardware dependencies. Experimental evaluations across multiple inertial odometry datasets (OxIOD, RONIN, and RIDI) and real-world inertial data collected on the KAIST campus illustrate FedNav’s superior performance, achieving localization accuracies that outperform existing methods. FedNav thus offers a promising solution for secure, accurate, and robust indoor positioning in AIoT systems, demonstrating the practical benefits of federated learning in overcoming the prevailing challenges in the field.

Particle filtering is a reliable Monte Carlo algorithm for estimating the state of a system in modeling non-linear, non-gaussian elements for estimation and tracking applications in various fields, including robotics, navigation, and computer vision. However, particle filtering can be computationally expensive, particularly in high-dimensional state spaces, and can be a bottleneck for real-time applications due to high memory consumption. This paper proposes a particle filter accelerator that employs a cellular automata-based pseudo-random number generator and an improved systematic resampler based on the Vose Alias method. The particles are distributed across several sub-filters, performing concurrent resampling and importance weights computations. The proposed accelerator leveraged the inherent parallelism and pipelining stages of FPGAs to perform the resampling stage in a parallel fashion, significantly enhancing the particle convergence time. The proposed accelerator deployed on the Zedboard (ZC7020) system-on-chip achieves a low execution time of approximately 4.63 $\mu \text{s}$ , 21.3 % speedup, and 3.1 % area reduction compared to the recent particle filter accelerator. The proposed design also demonstrates modularity, achieved through multiple parallel hardware subfilters that provide high throughput for real-time sensor data processing. Furthermore, the proposed accelerator performs a high sampling frequency of 216kHz, making it suitable for high throughput and real-time applications.

Quantum computing, combined with deep learning, leverages principles like superposition and entanglement to enhance complex data-driven tasks. The Noisy Intermediate-Scale Quantum (NISQ) era presents opportunities for hybrid quantum-classical architectures to address this challenge. Despite significant progress, practical applications of these hybrid models are limited. This letter proposes a novel hybrid quantum-classical deep learning architecture, integrating Quantum Convolutional Neural Networks (QCNNs) and Long-Short-Term Memory (LSTM) networks, enhanced by Cluster State Signal Processing. Furthermore, this letter addresses indoor-outdoor detection using high-dimensional signal data, utilizing the Cirq platform-a Python framework for developing and simulating Noisy Intermediate Scale Quantum (NISQ) circuits on quantum computers and simulators. The approach addresses noise and decoherence issues. Preliminary results show that the QCNN-LSTM model outperforms pure quantum and hybrid models in accuracy and efficiency. This validates the practical benefits of hybrid architectures, paving the way for advancements in complex data classification like indoor-outdoor detection.

This paper introduces a real-time pedestrian dead reckoning (PDR) algorithm for Internet of Things (IoT) devices. It is motivated by a practical challenge encountered during the development and operation of a positioning system named KAILOS, which provides positioning services via a smartphone application. Its dependence on smartphones causes data collection constraints imposed by operating systems. To mitigate this issue, we designed and incorporated a dedicated IoT device into the system, enabling full access to sensing data. This invention demands developing and porting a PDR algorithm into the IoT device to address the communication bottleneck with a remote positioning server. Our goal was to process a large amount of data from an Inertial Measurement Unit (IMU) instantly on the device to produce precise relative location information without forwarding all of it to the server. To this end, we propose a new approach for detecting steps using acceleration differential in our PDR. Additionally, a dynamic gyroscope bias update strategy is also included to enhance the capability of heading estimation. These advancements not only enhance the accuracy of the PDR algorithm but also facilitate its implementation on IoT devices. We practically deployed the PDR algorithm into our IoT hardware platform called Kailos Tag (K-Tag). Via the extensive experiments conducted both indoors and outdoors, we found that our real-time PDR outperformed the conventional methods. It reduces step detection errors (SDE) to approximately 1.6%, travel distance errors (TDE) to below 1.8%, and end/start errors(E/SE) to about 3.2m regardless of environment. Moreover, it enables an average positioning latency of 2.49 ms, while consuming only 20% of CPU usage and 8.6% of total power consumption.

Accurate indoor–outdoor detection (IOD) is essential for location-based services, context-aware computing, and mobile applications, as it enhances service relevance and precision. However, traditional IOD methods, which rely only on GPS data, often fail in indoor environments due to signal obstructions, while IMU data are unreliable on unseen data in real-time applications due to reduced generalizability. This study addresses this research gap by introducing the DeepIOD framework, which leverages IMU sensor data, GPS, and light information to accurately classify environments as indoor or outdoor. The framework preprocesses input data and employs multiple deep neural network models, combining outputs using an adaptive majority voting mechanism to ensure robust and reliable predictions. Experimental results evaluated on six unseen environments using a smartphone demonstrate that DeepIOD achieves significantly higher accuracy than methods using only IMU sensors. Our DeepIOD system achieves a remarkable accuracy rate of 98–99% with a transition time of less than 10 ms. This research concludes that DeepIOD offers a robust and reliable solution for indoor–outdoor classification with high generalizability, highlighting the importance of integrating diverse data sources to improve location-based services and other applications requiring precise environmental context awareness.

This paper presents an improved model based on the Swin Transformer for image super-resolution and enhancement, specifically focusing on applications relevant to the metaverse. Image compression is crucial for efficient image and video transmission and storage. Yet, it often results in compression artifacts and loss of original details, degrading compressed images' perceptual quality and fidelity. While modern methods for image restoration have primarily relied on convolutional neural networks, there have been limited attempts to leverage Transformers, which excel in complex vision tasks. This study investigates the lightweight Meta-Swin transformer-based model to enhance SwinIR specifically for image super-resolution in compressed image scenarios. We address key challenges in training the vision transformer and original Swin transformers, such as unstable training, inconsistencies in resolution between pre-training and fine-tuning stages, and limitations imposed by the scarcity of labeled data. We perform experiments on three typical tasks: eliminating JPEG compression artifacts, traditional image enhancement, and super-resolution for compressed input images. The experimental findings indicate that our newly proposed approach, Meta-Swin, significantly enhances the training convergence and performance of the original model SwinIR. Moreover, Meta-Swin outperforms SwinIR and IPT regarding results across different representative tasks. The findings highlight the potential of the modified transformer network as a powerful tool for super-resolution of low-resolution and compressed images and reconstruction, particularly in scenarios involving compressed inputs. The proposed lightweight transformer architecture is a good candidate for metaverse technologies by boosting visual quality and creating a more realistic digital experience.

With the exponential growth of the Internet of Things (IoT), ensuring robust end-to-end encryption is paramount. Current cryptographic accelerators often struggle with balancing security, area efficiency, and power consumption, which are critical for compact IoT devices and system-on-chips (SoCs). This work presents a novel approach to designing substitution boxes (S-boxes) for Advanced Encryption Standard (AES) encryption, leveraging dual quad-bit structures to enhance cryptographic security and hardware efficiency. By utilizing Algebraic Normal Forms (ANFs) and Walsh–Hadamard Transforms, the proposed Register Transfer Level (RTL) circuitry ensures optimal non-linearity, low differential uniformity, and bijectiveness, making it a robust and efficient solution for ASIC implementations. Implemented on 65 nm CMOS technology, our design undergoes rigorous statistical analysis to validate its security strength, followed by hardware implementation and functional verification on a ZedBoard. Leveraging Cadence EDA tools, the ASIC implementation achieves a central circuit area of approximately 199 μm2. The design incurs a hardware cost of roughly 80 gate equivalents and exhibits a maximum path delay of 0.38 ns. Power dissipation is measured at approximately 28.622 μW with a supply voltage of 0.72 V. According to the ASIC implementation on the TSMC 65 nm process, the proposed design achieves the best area efficiency, approximately 66.46% better than state-of-the-art designs.

Many methods using various sensors of the latest smartphones are being studied for accurate indoor navigation. In particular, sensor fusion frameworks integrate all information collectible from smartphones, such as Wi-Fi signals, and measurements obtained from gyroscopes, accelerometers, or magnetometers. However, sensor measurements contain unpredictable real-time errors made in dynamic indoor environments. In this paper, we propose a new sensor fusion framework that attains high positioning accuracy by learning errors. The proposed system discriminates errors in the sensor measurements and accumulates the errors to adjust the measurement values based on the accumulated error distributions. High positioning accuracy was achieved in experiments conducted in two typical environments, a corridor-type space, and an open space.

The adaption of the NFT to the mainstream web3 marketplace has led to the rise of challenges like authenticity and verification. On the other hand, privacy-preserving is one of the important research topics in the data-driven metaverse due to its anonymity while maintaining transparency and security. As the traditional method does not preserve privacy, there is a dire need for a novel privacy-preserving method for NFT authentication while preserving anonymity, transparency, and security. This paper proposes a smart card-based approach for privacy preservation authentication of Non-Fungible Tokens using Non-Interactive Zero Knowledge Proof (NIZKP). The paper presents a Novel NIZKP response with the integration of challenge and proof with the encapsulation of a time-based one-time password (TOTP) solution to address the most critical problem of the proof of the source of the NFT and its rightful owner. The proposed solution is a lightweight mechanism for preventing unauthorized and hostile agents from the accessibility of the system while maintaining privacy and anonymity along with the verification and authenticity of the NFT. We have implemented our approach and compared the communication overhead and functional security features with related work and its precedence over them. The results show that the proposed approach has less communication overhead than the existing work and fulfills all the functional security and privacy-preserving features. It indicates that the proposed method can be an ideal solution for NFT authentication and transaction among users.

Passive fingerprinting collects the received signal strength (RSS) of wireless local area network (WLAN) signals emitted from a smartphone to create fingerprints and then to construct a radiomap for an indoor positioning system (IPS). However, human-impassable paths and irregular signal collection periods result in missing values in fingerprints. Moreover, missing fingerprints in shaded areas result in the degradation of the positioning accuracy. In this paper, we propose a novel passive fingerprinting method, introducing the concept of a WiFi monitor adjacency matrix and a method of transforming legacy active fingerprints to passive fingerprints by a linear regression model. Our system estimates walking paths with RSS data, removing outliers of human-impassable paths. Moreover, we can construct a high-performance radiomap by supplementing the missing signals as well as missing fingerprints in shaded areas. In an experiment conducted in a complex building environment, the proposed method achieved a positioning accuracy of 2.18 m, which is 32% better than that of a state-of-the-art system. The radiomap construction time was also greatly reduced.

Wi-Fi fingerprinting is the most popular indoor positioning method today, representing received signal strength (RSS) values as vector-type fingerprints. Passive fingerprinting, unlike the active fingerprinting method, has the advantage of being able to track location without user participation by utilizing the signals that are naturally emitted from the user's smartphone. However, since signals are generated depending on the user's network usage patterns, there is a problem in that data are irregularly collected according to the patterns. Therefore, this paper proposes an adaptive algorithm that shows stable tracking performances for fingerprints generated at irregular time intervals. The accuracy and stability of the proposed tracking method were verified by experiments conducted in three scenarios. Through the proposed method, it is expected that the stability of indoor positioning and the quality of location-based services will improve.

Utilizing implicitly crowdsourced data is a popular approach for a Wi-Fi radio map construction for indoor positioning. The main advantage of implicit crowdsourcing is demanding less effort. A Wi-Fi radio map is constructed in an automated way by analyzing crowdsourced data. However, some of the studies working on the crowdsourcing approach do not consider a multi-floor environment, making their methods less practical. In this paper, we propose a method separating implicitly crowdsourced data by floor. The proposed method assumes that the crowd-sourced data include sequences of barometer data and that the information of the building where the data were collected is given. The proposed method can transform the crowdsourcing-based method for single-floor environments into a method for multifloor environments.