TensorBeat

Enhancing System Performance through Objective Feature Scoring of Multiple Persons' Breathing Using Non-Contact RF Approach

Breathing monitoring is an efficient way of human health sensing and predicting numerous diseases. Various contact and non-contact-based methods are discussed in the literature for breathing monitoring. Radio frequency (RF)-based breathing monitoring has recently gained enormous popularity among non-contact methods. This method eliminates privacy concerns and the need for users to carry a device. In addition, such methods can reduce stress on healthcare facilities by providing intelligent digital health technologies. These intelligent digital technologies utilize a machine learning (ML)-based system for classifying breathing abnormalities. Despite advances in ML-based systems, the increasing dimensionality of data poses a significant challenge, as unrelated features can significantly impact the developed system's performance. Optimal feature scoring may appear to be a viable solution to this problem, as it has the potential to improve system performance significantly. Initially, in this study, software-defined radio (SDR) and RF sensing techniques were used to develop a breathing monitoring system. Minute variations in wireless channel state information (CSI) due to breathing movement were used to detect breathing abnormalities in breathing patterns. Furthermore, ML algorithms intelligently classified breathing abnormalities in single and multiple-person scenarios. The results were validated by referencing a wearable sensor. Finally, optimal feature scoring was used to improve the developed system's performance in terms of accuracy, training time, and prediction speed. The results showed that optimal feature scoring can help achieve maximum accuracy of up to 93.8% and 91.7% for single-person and multi-person scenarios, respectively.

Monitoring Respiratory Motion With Wi-Fi CSI: Characterizing Performance and the BreatheSmart Algorithm

Respiratory motion (i.e., motion pattern and rate) can provide valuable information for many medical situations. This information may help in the diagnosis of different health disorders and diseases. Wi-Fi-based respiratory monitoring schemes utilizing commercial off-the-shelf (COTS) devices can provide contactless, low-cost, simple, and scalable respiratory monitoring without requiring specialized hardware. Despite intense research efforts, an in-depth investigation on how to evaluate this type of technology is missing. We demonstrated and assessed the feasibility of monitoring and extracting human respiratory motion from Wi-Fi channel state information (CSI) data. This demonstration involves implementing an end-to-end system for a COTS-based hardware platform, control software, data acquisition, and a proposed processing algorithm. The processing algorithm is a novel deep-learning-based approach that exploits small changes in both CSI amplitude and phase information to learn high-level abstractions of breathing-induced chest movements and to reveal the unique characteristics of their difference. We also conducted extensive laboratory experiments demonstrating an assessment technique that can be replicated when quantifying the performance of similar systems. The results indicate that the proposed scheme can classify respiratory patterns and rates with an accuracy of 99.54% and 98.69%, respectively, in moderately degraded RF channels. Comprehensive data acquisition revealed the capability of the proposed system in detecting and classifying respiratory motions. Understanding the feasible limits and potential failure factors of Wi-Fi CSI-based respiratory monitoring scheme - and how to evaluate them - is an essential step toward the practical deployment of this technology. This study discusses ideas for further expansion of this technology.

A survey on vital signs monitoring based on Wi-Fi CSI data

The COVID-19 pandemic further highlighted the need to use low-cost remote monitoring procedures for medical patients. Since the results reported in the literature have shown that the use of Channel State Information (CSI) from Wi-Fi networks to remotely monitor patients can provide means to obtain a powerful medical information package in a non-invasive way and at low cost, a consistent review and analysis of the state of the art on this applied technique is developed in the present work. Initially, a mathematical overview of the CSI technology and its functional model is done. Subsequently, details about the technical approach necessary to use CSI in medical applications and a summary of the studies reported in the literature with such applications are presented. Based on the analyses and discussions carried out throughout this work, a better understanding of the current state of the art is achieved. Challenges and perspectives for future research are also highlighted.

Machine Learning for Identifying Group Trajectory Outliers

Prior works on the trajectory outlier detection problem solely consider individual outliers. However, in real-world scenarios, trajectory outliers can often appear in groups, e.g., a group of bikes that deviates to the usual trajectory due to the maintenance of streets in the context of intelligent transportation. The current paper considers the Group Trajectory Outlier (GTO) problem and proposes three algorithms. The first and the second algorithms are extensions of the well-known DBSCAN and k NN algorithms, while the third one models the GTO problem as a feature selection problem. Furthermore, two different enhancements for the proposed algorithms are proposed. The first one is based on ensemble learning and computational intelligence, which allows for merging algorithms’ outputs to possibly improve the final result. The second is a general high-performance computing framework that deals with big trajectory databases, which we used for a GPU-based implementation. Experimental results on different real trajectory databases show the scalability of the proposed approaches.

Full Respiration Rate Monitoring Exploiting Doppler Information with Commodity Wi-Fi Devices

Respiration rate is an essential indicator of vital signs, which can demonstrate the physiological condition of the human body and provide clues to some diseases. Commercial Wi-Fi devices can provide a non-invasive, cost-effective and long-term respiration rate-monitoring scheme for home scenarios. However, previous studies show that the breathing depth and location may affect the detectability of respiratory signals. In this study, we leverage the variation of the Doppler spectral energy extracted from the channel state information (CSI) collected by Wi-Fi devices to track the chest displacement induced by respiration. First, the random phase is eliminated by phase-fitting method to obtain the complex CSI containing the Doppler shift. Then, the multipath decomposition of CSI is carried out to obtain the channel impulse response, which eliminates the interference phase of the time delay and retains the Doppler shift. The dynamic path units are also separate from the multipath, which overcomes the indoor multipath effect. Finally, we conduct a time–frequency analysis to dynamic units to accumulate Doppler spectral energy. Based on these ideas, we design a complete respiration rate-monitoring system to obtain the respiration rate by using the consistency between the Doppler energy change period and the respiratory cycle. We evaluate our system through extensive experiments in several typical home environments filled with multipath. Experimental results show that the errors of the three scenarios are approximate, the maximum error is less than 0.7 bpm, and the average errors are approximately 0.15 bpm. This result indicates that our scheme can achieve high precision respiration monitoring and has good anti-multipath ability compared with existing methods.

IoT Platform for COVID-19 Prevention and Control: A Survey

As a result of the worldwide transmission of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), coronavirus disease 2019 (COVID-19) has evolved into an unprecedented pandemic. Currently, with unavailable pharmaceutical treatments and low vaccination rates, this novel coronavirus results in a great impact on public health, human society, and global economy, which is likely to last for many years. One of the lessons learned from the COVID-19 pandemic is that a long-term system with non-pharmaceutical interventions for preventing and controlling new infectious diseases is desirable to be implemented. Internet of things (IoT) platform is preferred to be utilized to achieve this goal, due to its ubiquitous sensing ability and seamless connectivity. IoT technology is changing our lives through smart healthcare, smart home, and smart city, which aims to build a more convenient and intelligent community. This paper presents how the IoT could be incorporated into the epidemic prevention and control system. Specifically, we demonstrate a potential fog-cloud combined IoT platform that can be used in the systematic and intelligent COVID-19 prevention and control, which involves five interventions including COVID-19 Symptom Diagnosis, Quarantine Monitoring, Contact Tracing & Social Distancing, COVID-19 Outbreak Forecasting, and SARS-CoV-2 Mutation Tracking. We investigate and review the state-of-the-art literatures of these five interventions to present the capabilities of IoT in countering against the current COVID-19 pandemic or future infectious disease epidemics.

Classification of Hemodynamics Scenarios from a Public Radar Dataset Using a Deep Learning Approach

Contact-free sensors offer important advantages compared to traditional wearables. Radio-frequency sensors (e.g., radars) offer the means to monitor cardiorespiratory activity of people without compromising their privacy, however, only limited information can be obtained via movement, traditionally related to heart or breathing rate. We investigated whether five complex hemodynamics scenarios (resting, apnea simulation, Valsalva maneuver, tilt up and tilt down on a tilt table) can be classified directly from publicly available contact and radar input signals in an end-to-end deep learning approach. A series of robust k-fold cross-validation evaluation experiments were conducted in which neural network architectures and hyperparameters were optimized, and different data input modalities (contact, radar and fusion) and data types (time and frequency domain) were investigated. We achieved reasonably high accuracies of 88% for contact, 83% for radar and 88% for fusion of modalities. These results are valuable in showing large potential of radar sensing even for more complex scenarios going beyond just heart and breathing rate. Such contact-free sensing can be valuable for fast privacy-preserving hospital screenings and for cases where traditional werables are impossible to use.

Signal Separation and Tracking Algorithm for Multi-Person Vital Signs by Using Doppler Radar

Noninvasive monitoring is an important Internet-of-Things application, which is made possible with the advances in radio-frequency based detection technologies. Existing techniques however rely on the use of antenna array and/or frequency modulated continuous wave radar to detect vital signs of multiple adjacent objects. Antenna size and limited bandwidth greatly limit the applicability. In this paper, we propose our system termed 'DeepMining' which is a single-antenna, narrowband Doppler radar system that can simultaneously track the respiration and heartbeat rates of multiple persons with high accuracy. DeepMining uses a number of signal observations over a period of time as input and returns the trajectory of the respiration and heartbeat rates of each person. The extraction is based on frequency separation algorithms using successive signal cancellation. The proposed system is implemented using the self-injection locking radar architecture and tested in a series of experiments, showing accuracies of 90% and 85% for two and three objects, respectively, even for closely located persons.

A Unified Fourth-Order Tensor-Based Smart Community System

Empowered by the ubiquitous sensing capabilities of Internet of Things (IoT) technologies, smart communities could benefit our daily life in many aspects. Various smart community studies and practices have been conducted, especially in China thanks to the government’s support. However, most intelligent systems are designed and built individually by different manufacturers in diverging platforms with different functionalities. Therefore, multiple individual systems must be deployed in a smart community to have a set of functions, which could lead to hardware waste, high energy consumption and high deployment cost. More importantly, current smart community systems mainly focus on the technologies involved, while the effects of human activity are neglected. In this paper, a fourth-order tensor model representing object, time, location and human activity is proposed for human-centered smart communities, based on which a unified smart community system is designed. Thanks to the powerful data management abilities of a high-order tensor, multiple functions can be integrated into our system. In addition, since the tensor model embeds human activity information, complex functions could be implemented by exploring the effects of human activity. Two exemplary applications are presented to demonstrate the flexibility of the proposed unified fourth-order tensor-based smart community system.

Non-Contact Sensing Testbed for Post-Surgery Monitoring by Exploiting Artificial-Intelligence

Non-contact health care monitoring is a unique feature in the emerging 5G networks that is achieved by exploiting artificial intelligence (AI). The ratio of the number of health care problems and patients is increasing exponentially and creating burgeoning data. The integration of AI and Internet of things (IoT) systems enables us to increase the huge volume of data to be generated. The approach by which AI is applied to the IoT systems enhances the intelligence of the health care system. In post-surgery monitoring of the patient, timely consultation is essential before further loss. Unfortunately, even after the advice of the doctor to the patient, he/she may forget to perform the activity in the correct way, which may lead to complications in recovery. In this research, the idea is to design a non-contact sensing testbed using AI for the classification of post-surgery activities. Universal software-defined radio peripheral (USRP) is utilized to collect the data of spinal cord operated patients during weight lifting activity. The wireless channel state information (WCSI) is extracted by using orthogonal frequency division multiplexing (OFDM) technique. AI applies machine learning to classify the correct and wrong way of weight lifting activity that was considered for experimental analysis. The accuracy achieved by the proposed testbed by using a fine K-nearest neighbor (FKNN) algorithm is 99.6%.

Towards Breathing as a Sensing Modality in Depth-Based Activity Recognition

Depth imaging has, through recent technological advances, become ubiquitous as products become smaller, more affordable, and more precise. Depth cameras have also emerged as a promising modality for activity recognition as they allow detection of users' body joints and postures. Increased resolutions have now enabled a novel use of depth cameras that facilitate more fine-grained activity descriptors: The remote detection of a person's breathing by picking up the small distance changes from the user's chest over time. We propose in this work a novel method to model chest elevation to robustly monitor a user's respiration, whenever users are sitting or standing, and facing the camera. The method is robust to users occasionally blocking their torso region and is able to provide meaningful breathing features to allow classification in activity recognition tasks. We illustrate that with this method, with specific activities such as paced-breathing meditating, performing breathing exercises, or post-exercise recovery, our model delivers a breathing accuracy that matches that of a commercial respiration chest monitor belt. Results show that the breathing rate can be detected with our method at an accuracy of 92 to 97% from a distance of two metres, outperforming state-of-the-art depth imagining methods especially for non-sedentary persons, and allowing separation of activities in respiration-derived features space.

Improved LDA Dimension Reduction Based Behavior Learning with Commodity WiFi for Cyber-Physical Systems

In recent years, rapid development of sensing and computing has led to very large datasets. There is an urgent demand for innovative data analysis and processing techniques that are secure, privacy-protected and sustainable. In this article, taking human activities and interactions with Cyber-Physical Systems (CPS) into consideration, we propose a human behavior learning system based on Channel State Information (CSI) utilizing a series of algorithms for data analysis and processing. Aiming to recognize a set of gestures, our system is designed based on the observation that different gestures have different effects on signals and specific gesture signals have a unique energy spectrum. Specifically, an improved Linear Discriminant Analysis Algorithm (I-LDA) is devised to reduce the dimension of human behavior signals. Additionally, behaviors are learned by Logistic Regression Algorithm (LRA). Bandwidth ratios in an energy spectrum are selected as features to eliminate the impact of speed differences on results. The system is based on commercial off-the-shelf WiFi devices and we conduct a large number of experiments in a typical indoor environment to evaluate its performance. Experimental results show that our system is robust with average recognition accuracy of up to 96%.

State-of-the-Art Internet of Things in Protected Agriculture

The Internet of Things (IoT) has tremendous success in health care, smart city, industrial production and so on. Protected agriculture is one of the fields which has broad application prospects of IoT. Protected agriculture is a mode of highly efficient development of modern agriculture that uses artificial techniques to change climatic factors such as temperature, to create environmental conditions suitable for the growth of animals and plants. This review aims to gain insight into the state-of-the-art of IoT applications in protected agriculture and to identify the system structure and key technologies. Therefore, we completed a systematic literature review of IoT research and deployments in protected agriculture over the past 10 years and evaluated the contributions made by different academicians and organizations. Selected references were clustered into three application domains corresponding to plant management, animal farming and food/agricultural product supply traceability. Furthermore, we discussed the challenges along with future research prospects, to help new researchers of this domain understand the current research progress of IoT in protected agriculture and to propose more novel and innovative ideas in the future.

Danger-Pose Detection System Using Commodity Wi-Fi for Bathroom Monitoring

A bathroom has higher probability of accidents than other rooms due to a slippery floor and temperature change. Because of high privacy and humidity, we face difficulties in monitoring inside a bathroom using traditional healthcare methods based on cameras and wearable sensors. In this paper, we present a danger-pose detection system using commodity Wi-Fi devices, which can be applied to bathroom monitoring, preserving privacy. A machine learning-based detection method usually requires data collected in target situations, which is difficult in detection-of-danger situations. We therefore employ a machine learning-based anomaly-detection method that requires a small amount of data in anomaly conditions, minimizing the required training data collected in dangerous conditions. We first derive the amplitude and phase shift from Wi-Fi channel state information (CSI) to extract low-frequency components that are related to human activities. We then separately extract static and dynamic features from the CSI changes in time. Finally, the static and dynamic features are fed into a one-class support vector machine (SVM), which is used as an anomaly-detection method, to classify whether a user is not in bathtub, bathing safely, or in dangerous conditions. We conducted experimental evaluations and demonstrated that our danger-pose detection system achieved a high detection performance in a non-line-of-sight (NLOS) scenario.

Mitigation of CSI Temporal Phase Rotation with B2B Calibration Method for Fine-Grained Motion Detection Analysis on Commodity Wi-Fi Devices

Limitations of optical devices for motion sensing such as small coverage, sensitivity to obstacles, and privacy exposure result in the need for improvement. As motion sensing based on radio frequency signals is not constrained by the limitation above, channel state information (CSI) from Wi-Fi devices could be used to improve sensing performance under the above circumstances. Unfortunately, CSI phase cannot be practically obtained due to the temporal phase rotation generated from Wi-Fi chips. Therefore, it would be rather complicated to realize motion analysis, especially the direction of motion. To mitigate the issue, this paper proposes a CSI calibration method that employs a back-to-back channel between Wi-Fi transceivers for phase rotation removal while preserving the original CSI phase. Through experiment, calibrated CSI showed a high similarity to the channel without phase rotation measured using a Vector Network Analyzer (VNA). Another experiment was conducted to observe Doppler frequency due to simple hand gestures using the Wavelet transform. A visual analysis revealed that the Doppler frequency of calibrated CSI could correctly capture the motion pattern. To the best of the authors' knowledge, this is the first calibration method that maintains the original CSI and is applicable for in-depth motion analysis.

From Signal to Image: Enabling Fine-Grained Gesture Recognition with Commercial Wi-Fi Devices

Gesture recognition acts as a key enabler for user-friendly human-computer interfaces (HCI). To bridge the human-computer barrier, numerous efforts have been devoted to designing accurate fine-grained gesture recognition systems. Recent advances in wireless sensing hold promise for a ubiquitous, non-invasive and low-cost system with existing Wi-Fi infrastructures. In this paper, we propose DeepNum, which enables fine-grained finger gesture recognition with only a pair of commercial Wi-Fi devices. The key insight of DeepNum is to incorporate the quintessence of deep learning-based image processing so as to better depict the influence induced by subtle finger movements. In particular, we make multiple efforts to transfer sensitive Channel State Information (CSI) into depth radio images, including antenna selection, gesture segmentation and image construction, followed by noisy image purification using high-dimensional relations. To fulfill the restrictive size requirements of deep learning model, we propose a novel region-selection method to constrain the image size and select qualified regions with dominant color and texture features. Finally, a 7-layer Convolutional Neural Network (CNN) and SoftMax function are adopted to achieve automatic feature extraction and accurate gesture classification. Experimental results demonstrate the excellent performance of DeepNum, which recognizes 10 finger gestures with overall accuracy of 98% in three typical indoor scenarios.