Machine Learning for Practical Localization System Using Multiview CSI

Due to its high-dimensional data characteristics, the channel state information (CSI) of Wi-Fi signals has become a strong candidate for use in indoor localization. In addition, machine learning techniques can improve the accuracy of indoor localization systems using multiview CSI data received at multiple access points (APs). However, in complex environments, most CSI views collected at APs in non-line-of-sight (NLoS) configurations relative to a transmitter may lose so much useful data information as to become nonsalient. In this paper, we propose a practical machine learning approach named unsupervised view-selective deep learning (UVSDL), in which only the most salient CSI view is selected in an unsupervised manner to be applied in regression for localization. In an experiment in a complex building, our variational deep learning (VDL)-based regression method with the most salient CSI view achieves a localization accuracy of 1.36 m, significantly outperforming the best-known system BiLoc by 25 %.

Radio channel state information (CSI) measured with many receivers is a good resource for localizing a transmitter device with machine learning with a discriminative model. However, CSI localization is nontrivial when the radio map is complicated, such as in building corridors. This article introduces a view-selective deep learning (VSDL) system for indoor localization using CSI of WiFi. The multiview training with CSI obtained from multiple groups of access points (APs) generates latent features on a supervised variational deep network. This information is then applied to an additional network for dominant view classification to minimize the regression loss of localization. As noninformative latent features from multiple views are rejected, we can achieve a localization accuracy of 1.28 m, which outperforms by 30% the best known accuracy in practical applications in a complex building environment. To the best of our knowledge, this is the first approach to apply variational inference and to construct a practical system for radio localization. Furthermore, our work investigates a methodology for supervised learning with multiview data where informative and noninformative views coexist.

The indoor positioning system (IPS) is becoming increasing important in accurately determining the locations of objects by the utilization of micro-electro-mechanical-systems (MEMS) involving smartphone sensors, embedded sources, mapping localizations, and wireless communication networks. Generally, a global positioning system (GPS) may not be effective in servicing the reality of a complex indoor environment, due to the limitations of the line-of-sight (LoS) path from the satellite. Different techniques have been used in indoor localization services (ILSs) in order to solve particular issues, such as multipath environments, the energy inefficiency of long-term battery usage, intensive labour and the resources of offline information collection and the estimation of accumulated positioning errors. Moreover, advanced algorithms, machine learning, and valuable algorithms have given rise to effective ways in determining indoor locations. This paper presents a comprehensive review on the positioning algorithms for indoors, based on advances reported in radio wave, infrared, visible light, sound, and magnetic field technologies. The traditional ranging parameters in addition to advanced parameters such as channel state information (CSI), reference signal received power (RSRP), and reference signal received quality (RSRQ) are also presented for distance estimation in localization systems. In summary, the recent advanced algorithms can offer precise positioning behaviour for an unknown environment in indoor locations.

With the popularization of machine learning (ML) techniques and the increased chipset’s performance, the application of ML to pedestrian localization systems has received significant attention in the last years. Several survey papers have attempted to provide a state-of-the-art overview, but they usually limit their scope to a particular type of positioning system or technology. In addition, they are written from the point of view of ML techniques and their practice, not from the point of view of the localization system and the specific problems that ML techniques can help to solve. This article is intended to offer a comprehensive state-of-the-art survey of the ML techniques that have been adopted over the last ten years to improve the performance of pedestrian localization systems, addressing the applicability of ML techniques in this domain, along with the main localization strategies. It concludes by indicating the underlying open issues and challenges associated with the existing systems, and possible future directions in which ML techniques could improve the performance of pedestrian localization systems. Among other open issues, most previous authors have focused their attention on position estimation accuracy, which wastes the potential of ML techniques to improve other performance parameters (e.g., response time, computational complexity, robustness, scalability or energy efficiency). This study shows that there is a strong trend towards the application of supervised learning. Consequently, there are many potential research opportunities in the use of other learning types, such as unsupervised and reinforcement learning, to improve the performance of pedestrian localization systems.

Understanding the indoor whereabouts of individuals and objects is important, especially for those who fall within the 71% of visually impaired individuals with a school education, students in 450 special education units and many other areas and aspects in Sri Lanka. Researchers have declared that, there isn’t any particularly good localization system, and the performance should be evaluated considering the approach and application. The most well-known indoor positioning (IP) technologies that have been historically deployed are Bluetooth, Wi-Fi, RFID (radio frequency identification), IR (Infrared), and UV (ultraviolet) whereas received signal strength (RSSI), fingerprinting, and triangulation methods have been used as common IP techniques. The combination of both IP technologies and techniques creates an IP system, and the integration of machine learning and IoT with the structured system essentially delivers an accurate and more advanced system. This paper contains a detailed, analytical review of a developed indoor positioning system derived from the existing indoor localization techniques, localization technologies, localization systems, algorithms, and performance matrixes. This also provides a comprehensive comparison between numerous existing systems to justify the proposed solution. This project has been developed to achieve better accuracy through low-cost deployment as an effective system to fill the gap in the scarcity of positioning systems in the world. This paper presents a descriptive introduction and problem definition, a critical discussion of results, machine learning models, benefits of the project, and future works. As later justified, ESP32 microcontroller and BLE beacons are utilized with RSSI fingerprinting method to develop this IP system and, as a part of the project, two data visualization methods have been introduced here using NodeRED dashboard and LC display. Overall, this project was developed with an effective combination of RSSI fingerprinting, IoT protocols, machine learning, and data interpretation methods. KEYWORDS: Indoor positioning, RSSI, Fingerprinting, Machine Learning, MQTT, ESP32, BLE beacons, NodeRED

HandFi: WiFi Sensing based Hand Gesture Recognition using Channel State Information

In this paper, a novel indoor localization system with channel state information (CSI) fingerprints is proposed, which learns the spatial and frequency features of CSI in the fifth-generation (5G) cellular network by a Siamese convolution neural network. In particular, considering that the CSI continuously collected by a moving target possesses the implicit spatial association, we locate the target by the successive CSI data gathered within a time interval which can be regarded as an information subspace of the fingerprint database. Therefore, the fingerprint localization can be modeled as a subspace matching problem and solved by the Siamese network-based similarity learning. In the proposed system, we design a structure of CSI fingerprint which includes the information from multiple base stations in spatial and frequency domains. Then, the proposed Siamese architecture extracts the CSI feature and estimates the location of the target by feature similarity comparison. Compared with the existing algorithms, it can increase the positioning accuracy significantly by the feature relevance among the CSI data collected at different positions. The field tests indicate that compared to other CSI fingerprint-based positioning methods, our proposed algorithm can effectively reduce the localization error.

A comprehensive review on detection of plant disease using machine learning and deep learning approaches

In this paper, we investigate the machine learning approaches (sparse Bayesian linear regression (SBLR) and support vector machine (SVM)) for channel state information (CSI) prediction and dynamic radio frequency (RF) slicing for software defined virtual wireless networks in large-scale multi-input multi-output (MIMO) wireless networks. Specifically, a subset of the antennas of virtual wireless networks transmits pilot symbols for estimating the CSI and use the estimated CSI dataset to train and estimate the remaining channels and future CSI for virtual networks using machine learning algorithms. This helps not only to predict the CSI with least overhead and fulfills the service demands of users but also to reduce the power consumption and computation overhead in the network. Predicted CSI is leveraged for RF slicing for virtual wireless networks. Simulation results show that the proposed SBLR for predicting CSI results in lower BER and higher data rate for the wireless users. Furthermore, SBLR outperforms the other approaches when we have sparse CSI information and we need to generalize the prediction process.

Machine Learning (ML) techniques are pervasive tools in various emerging commercial applications, but have to be accommodated by powerful computer systems to process very large data. Although general-purpose CPUs and GPUs have provided straightforward solutions, their energy-efficiencies are limited due to their excessive supports for flexibility. Hardware accelerators may achieve better energy-efficiencies, but each accelerator often accommodates only a single ML technique (family). According to the famous No-Free-Lunch theorem in the ML domain, however, an ML technique performs well on a dataset may perform poorly on another dataset, which implies that such accelerator may sometimes lead to poor learning accuracy. Even if regardless of the learning accuracy, such accelerator can still become inapplicable simply because the concrete ML task is altered, or the user chooses another ML technique.In this study, we present an ML accelerator called PuDianNao, which accommodates seven representative ML techniques, including k-means, k-nearest neighbors, naive bayes, support vector machine, linear regression, classification tree, and deep neural network. Benefited from our thorough analysis on computational primitives and locality properties of different ML techniques, PuDianNao can perform up to 1056 GOP/s (e.g., additions and multiplications) in an area of 3.51 mm^2, and consumes 596 mW only. Compared with the NVIDIA K20M GPU (28nm process), PuDianNao (65nm process) is 1.20x faster, and can reduce the energy by 128.41x.

Indoor localization knows many applications, such as industry 4.0, warehouses, healthcare, drones, etc., where high accuracy becomes more critical than ever. Recent advances in ultra-wideband localization systems allow high accuracies for multiple active users in line-of-sight environments, while they still introduce errors above 300 mm in non-line-of-sight environments due to multi-path effects. Current work tries to improve the localization accuracy of ultra-wideband through offline error correction approaches using popular machine learning techniques. However, these techniques are still limited to simple environments with few multi-path effects and focus on offline correction. With the upcoming demand for high accuracy and low latency indoor localization systems, there is a need to deploy (online) efficient error correction techniques with fast response times in dynamic and complex environments. To address this, we propose (i) a novel semi-supervised autoencoder-based machine learning approach for improving ranging accuracy of ultra-wideband localization beyond the limitations of current improvements while aiming for performance improvements and a small memory footprint and (ii) an edge inference architecture for online UWB ranging error correction. As such, this paper allows the design of accurate localization systems by using machine learning for low-cost edge devices. Compared to a deep neural network (as state-of-the-art, with a baseline error of 75 mm) the proposed autoencoder achieves a 29% higher accuracy. The proposed approach leverages robust and accurate ultra-wideband localization, which reduces the errors from 214 mm without correction to 58 mm with correction. Validation of edge inference using the proposed autoencoder on a NVIDIA Jetson Nano demonstrates significant uplink bandwidth savings and allows up to 20 rapidly ranging anchors per edge GPU.

Machine learning (ML) techniques have numerous applications in many fields, including healthcare, medicine, finance, marketing, and cyber security. For example, ML techniques are being applied to determine whether to give a loan to a customer or whether the computing system has been attacked. However, the ML techniques themselves may be subject to attacks and may discriminate when determining who should get the loan. Therefore, the ML techniques have to be secure, ensure privacy of the individuals, incorporate fairness and be accurate. Such collection of ML techniques has come to be known as trustworthy machine learning (trustworthy ML). This article describes an architecture to support scalable trustworthy ML and describes the features that have to be incorporated into the ML techniques to ensure that they are trustworthy.

Source apportionment is a cornerstone in environmental sciences, providing essential insights into the contributions of diverse pollution sources to ambient air quality. Understanding these contributions is vital for designing effective regulatory and mitigation strategies. Among the most commonly employed techniques for source apportionment, Positive Matrix Factorization (PMF) has proven to be a powerful receptor model. PMF decomposes complex environmental datasets into source profiles and their corresponding contributions while accommodating measurement uncertainties Hopke [2016], Paatero and Tapper [1994]. Despite its advantages, PMF is not without limitations, such as rotational ambiguity, reliance on accurate input uncertainties, and potential biases in source attribution Reff et al. [2007].In recent years, the emergence of artificial intelligence (AI)-based methodologies has opened new horizons for source apportionment. These approaches often build upon classical methods like PMF, aiming to enhance the interpretability and performance of source apportionment models Geng et al. [2020]. Machine learning techniques, including deep learning, leverage large datasets to identify patterns and relationships that may elude traditional approaches. Additionally, hybrid methods integrating classical models with AI frameworks have demonstrated potential for improved accuracy and robustness Wang et al. [2021].A critical question remains unanswered: do the intrinsic limitations of PMF, such as biases and errors, propagate into these AI-driven alternatives? For instance,  AI methods are theoretically capable of overcoming such challenges, their reliance on data-driven training may introduce new sources of bias or amplify existing uncertainties if the underlying data or assumptions are flawed.Moreover, how do these biases compare to those in other receptor models, such as the Chemical Mass Balance (CMB) model ?This work aims to look on the extent to which the errors and biases inherent to PMF are inherited by alternative methods, including AI-based approaches and other receptor models. By critically assessing these methods, we seek to provide a comprehensive understanding of the strengths and limitations of emerging tools for source apportionment and their potential to overcome traditional challenges. Xiaoyu Geng, Lin Zhang, and Yu Wang. Ai-based approaches for source apportionment in atmospheric science: A review and perspective. Atmospheric Environment, 223: 117276, 2020. doi: 10.1016/j.atmosenv.2020.117276. Philip K. Hopke. Review of receptor modeling methods for source apportionment. Journal of the Air Waste Management Association, 66(3):237–259, 2016. doi: 10.1080/10962247.2016.1140693.Pentti Paatero and Unto Tapper. Positive matrix factorization: A non-negative factor model with optimal utilization of error estimates of data values. Environmetrics, 5(2): 111–126, 1994. doi: 10.1002/env.3170050203. Adam Reff, Sierra I. Eberly, and Prakash V. Bhave. Receptor modeling of ambient particulate matter data using positive matrix factorization: Review of existing meth- ods. Journal of the Air Waste Management Association, 57(2):146–154, 2007. doi: 10.1080/10473289.2007.10465319.Feifei Wang, Hui Zhang, and Xiaodong Li. Hybrid receptor models and machine learning for air pollution source apportionment: Advances and future directions. Environmental Research, 195:110827, 2021. doi: 10.1016/j.envres.2021.110827. John G. Watson. Chemical mass balance receptor model methodology for assessing the sources of fine and coarse particulate matter. Environmental Software, 5(2):38–49, 1990. doi: 10.1016/0266-9838(90)90008- 

Code smells are considered symptoms of poor implementation choices, which may hamper the software maintainability. Hence, code smells should be detected as early as possible to avoid software quality degradation. Unfortunately, detecting code smells is not a trivial task. Some preliminary studies investigated and concluded that machine learning (ML) techniques are a promising way to better support smell detection. However, these techniques are hard to be customized to promote an early and accurate detection of specific smell types. Yet, ML techniques usually require numerous code examples to be trained (composing a relevant dataset) in order to achieve satisfactory accuracy. Unfortunately, such a dependency on a large validated dataset is impractical and leads to late detection of code smells. Thus, a prevailing challenge is the early customized detection of code smells taking into account the typical limited training data. In this direction, this paper reports a study in which we collected code smells, from ten active projects, that were actually refactored by developers, differently from studies that rely on code smells inferred by researchers. These smells were used for evaluating the accuracy regarding early detection of code smells by using seven ML techniques. Once we take into account such smells that were considered as important by developers, the ML techniques are able to customize the detection in order to focus on smells observed as relevant in the investigated systems. The results showed that all the analyzed techniques are sensitive to the type of smell and obtained good results for the majority of them, especially JRip and Random Forest. We also observe that the ML techniques did not need a high number of examples to reach their best accuracy results. This finding implies that ML techniques can be successfully used for early detection of smells without depending on the curation of a large dataset.

Landslide disaster risk reduction necessitates the investigation of different geotechnical causal factors for slope failures. Machine learning (ML) techniques have been proposed to study causal factors across many application areas. However, the development of ensemble ML techniques for identifying the geotechnical causal factors for slope failures and their subsequent prediction has lacked in literature. The primary goal of this research is to develop and evaluate novel feature selection methods for identifying causal factors for slope failures and assess the potential of ensemble and individual ML techniques for slope failure prediction. Twenty-one geotechnical causal factors were obtained from 60 sites (both landslide and non-landslide) spread across a landslide-prone area in Mandi, India. Relevant causal factors were evaluated by developing a novel ensemble feature selection method that involved an average of different individual feature selection methods like correlation, information-gain, gain-ratio, OneR, and F-ratio. Furthermore, different ensemble ML techniques (Random Forest (RF), AdaBoost (AB), Bagging, Stacking, and Voting) and individual ML techniques (Bayesian network (BN), decision tree (DT), multilayer perceptron (MLP), and support vector machine (SVM)) were calibrated to 70% of the locations and tested on 30% of the sites. The ensemble feature selection method yielded six major contributing parameters to slope failures: relative compaction, porosity, saturated permeability, slope angle, angle of the internal friction, and in-situ moisture content. Furthermore, the ensemble RF and AB techniques performed the best compared to other ensemble and individual ML techniques on test data. The present study discusses the implications of different causal factors for slope failure prediction.