Smartphone Accelerometer Sensor Research Articles

Significance of handcrafted features in human activity recognition with attention-based RNN models

Sensors incorporated in devices are a source of temporal data that can be interpreted to learn the context of a user. The smartphone accelerometer sensor generates data streams that form distinct patterns in response to user activities. The human context can be predicted using deep learning models built from raw sensor data or features retrieved from raw data. This study analyzes data streams from the UCI-HAR public dataset for activity recognition to determine 31 handcrafted features in the temporal and frequency domain. Various stacked and combination RNN models, trained with attention mechanisms, are designed to work with computed features. Attention gave the models a good fit. When trained with all features, the two-stacked GRU model performed best with 99% accuracy. Selecting the most promising features helps reduce training time without compromising accuracy. The ranking supplied by the permutation feature importance measure and Shapley values are utilized to identify the best features from the highly correlated features. Models trained using optimal features, as determined by the importance measures, had a 96% accuracy rate. Misclassification in attention-based classifiers occurs in the prediction of dynamic activities, such as walking upstairs and walking downstairs, and in sedentary activities, such as sitting and standing, due to the similar range of each activity’s axis values. Our research emphasizes the design of streamlined neural network architectures, characterized by fewer layers and a reduced number of neurons when compared to existing models in the field, to design lightweight models to be implemented in resource-constraint gadgets.

Accelerometers in Our Pocket: Does Smartphone Accelerometer Technology Provide Accurate Data?

This study evaluates accelerometer performance of three new state of the art smartphones and focuses on accuracy. The motivating research question was whether accelerator accuracy obtained with these off-the-shelf modern smartphone accelerometers was or was not statistically different from that of a gold-standard reference system. We predicted that the accuracy of the three modern smartphone accelerometers in human movement data acquisition do not differ from that of the Vicon MX motion capture system. To test this prediction, we investigated the comparative performance of three different commercially available current generation smartphone accelerometers among themselves and to a gold-standard Vicon MX motion capture system. A single subject design was implemented for this study. Pearson's correlation coefficients® were calculated to verify the validity of the smartphones' accelerometer data against that of the Vicon MX motion capture system. The Intraclass Correlation Coefficient (ICC) was used to assess the smartphones' accelerometer performance reliability compared to that of the Vicon MX motion capture system. Results demonstrated that (a) the tested smartphone accelerometers are valid and reliable devices for estimating accelerations and (b) there were not significant differences among the three current generation smartphones and the Vicon MX motion capture system's mean acceleration data. This evidence indicates how well recent generation smartphone accelerometer sensors are capable of measuring human body motion. This study, which bridges a significant information gap between the accuracy of accelerometers measured close to production and their accuracy in actual smartphone research, should be interpreted within the confines of its scope, limitations and strengths. Further research is warranted to validate our arguments, suggestions, and results, since this is the first study on this topic.

Development of pavement roughness regression models based on smartphone measurements

PurposeThe purpose of this paper was to study the possibility of using smartphone roughness measurements for developing pavement roughness regression models as a function of pavement age, traffic loading and traffic volume variables. Also, the effects of patching and pavement distresses on pavement roughness were investigated. The work focused on establishing pavement roughness prediction models and applying these models to pavement management systems (PMS) to help decision-makers choose the best maintenance and rehabilitation (M&R) options by using cost-effective methods.Design/methodology/approachSignal processing techniques including filtering and processing techniques were used to obtain the International Roughness Index (IRI) from raw acceleration data collected from smartphone accelerometer sensors. The obtained IRI values were inputted as a dependent variable in analytical regression models as well as several independent variables with proper transformations.FindingsAccording to the study results, several regression models were developed with a big variation in the coefficients of determination (R2). However, the best models included pavement age, accumulated traffic volume (∑TV) and construction quality factor (CQF) with R2 equal to 0.63. It was also found that the effects of pavement distresses and patching was significant at a-level < 0.05. The patching effect on pavement roughness was found higher than the effect of other pavement distresses.Practical implicationsThe presented results and methods in this paper could be used in the future predictions of pavement roughness and help the decision-makers to estimate M&R needs. The work focused on establishing IRI prediction models and applying these models to the PMS to help decision-makers choose the best M & R options.Originality/valueTo develop sound pavement roughness models, it is essential to collect roughness data using automated procedures. However, applying these procedures in developing countries faces several difficulties such as the high price and operation costs of roughness equipment and lack of technical experience. The advantage of using IRI values taken from smartphones is that the roughness evaluation survey may be expanded to cover the full road network at a cheaper cost than with automated instruments. Therefore, if the roughness survey covers more roads, the prediction model’s accuracy will be improved.

Statistical and Machine Learning-Based Recognition of Coughing Events Using Triaxial Accelerometer Sensor Data From Multiple Wearable Points

Recent studies have investigated the use of the accelerometer sensors in smartphones and wearable devices for human activity recognition such as sitting, standing, walking, and laying down with reasonably high accuracy. In this letter, we use a triaxial microelectromechanical accelerometer that is commonly used in smartphones to detect and discern coughing events. Our letter focuses on detecting and differentiating coughing from other human activities such as sitting, standing, and walking using accelerometer's x, y, and z data from various body positions where electronics such as smartphones, watches, headphones, and earphones are commonly worn. Our research compares acceleration measured at five different positions on the body: chest, stomach, shirt-pocket, upper arm, and ear. The measurements are analyzed in the x, y, and z directions using the statistical and machine learning (ML) approaches to study how well coughing activity can be differentiated from acceleration due to other human motions. Analysis of the measured data using both methods show accelerometers mounted on the ear/headphones to be the ideal spot to detect and differentiate coughing with the highest accuracy. ML analysis of accelerometer measurements on ear and chest shows 96 and 93% accuracy, respectively, for cough detection. Measurements show the standard deviation and convolution neural network detection accuracy for cough detection in the ear to be 15 and 3% more sensitive compared to the next best position, which is chest.

Exfiltrating data from air-gapped computers via ViBrAtIoNs

Air-gap covert channels are special types of covert communication channels that enable attackers to exfiltrate data from isolated, network-less computers. Various types of air-gap covert channels have been demonstrated over the years, including electromagnetic, magnetic, acoustic, optical, and thermal.In this paper, we introduce a new type of vibrational (seismic) covert channel. We observe that computers vibrate at a frequency correlated to the rotation speed of their internal fans. These inaudible vibrations affect the entire structure on which the computer is placed. Our method is based on malware’s capability of controlling the vibrations generated by a computer, by regulating its internal fan speeds. We show that the malware-generated covert vibrations can be sensed by nearby smartphones via the integrated, sensitive accelerometers. Notably, the accelerometer sensors in smartphones can be accessed by any app without requiring the user permissions, which make this attack highly evasive. We implemented AiR-ViBeR, malware that encodes binary information, and modulate it over a low frequency vibrational carrier. The data is then decoded by malicious application on a smartphone placed on the same surface of the air-gapped computer (e.g., on a desk). We discuss the attack model, provide technical background, and present the implementation details and evaluation results. Our results show that using AiR-ViBeR, data can be exfiltrated from air-gapped computer to a nearby smartphone on the same table, or even an adjacent table, via vibrations. Finally, we propose a set of countermeasures for this new type of attack.

RETRACTED ARTICLE: Application of back propagation artificial neural network in detection and analysis of diabetes mellitus

Diabetes Mellitus affects adults and children, causing changes in lifestyle. The diabetic affected person count has increased drastically worldwide over the last few years; about 425 million people have diabetes. By 2030, it is predicted that diabetic disorder will be the seventh leading cause of human death. Diabetes mellitus is measured invasively. This method has limitations such as patient’s preparation, piercing of the skin, which causes infection and needs for skilled technicians. In order to avoid the limitations of invasive methods, vibrations from the pancreas are acquired using a smartphone accelerometer sensor and detecting the value of diabetes. The human body has a unique energy signature for every organ, which leads to vibrations with different frequencies. The frequency of the vibration signal from the pancreas is proportional to insulin secretion and dynamics. The signals obtained from the accelerometer sensor are trained and analyzed with the Levenberg–Marquardt algorithm for obtaining the relation between the excess insulin secretion and clinical value of the diabetic level of the person. The accelerometer signals and clinical values are modeled with Regression analysis for the diabetic and non-diabetic persons. The results show the correlation between the fluid dynamics of insulin and clinical value at about 95% in prediction.

A Preliminary Study Using Smartphone Accelerometers to Sense Gait Impairments Due to Alcohol Intoxication

Sensing the effects of alcohol consumption in real time could offer numerous opportunities to reduce related harms. This study sought to explore accuracy of gait-related features measured by smartphone accelerometer sensors on detecting alcohol intoxication (breath alcohol concentration [BrAC] > .08%). In a controlled laboratory study, participants (N = 17; 12 male) were asked to walk 10 steps in a straight line, turn, and walk 10 steps back before drinking and each hour, for up to 7 hours after drinking a weight-based dose of alcohol to reach a BrAC of .20%. Smartphones were placed on the lumbar region and 3-axis accelerometer data was recorded at a rate of 100 Hz. Accelerometer data were segmented into task segments (i.e., walk forward, walk backward). Features were generated for each overlapping 1-second windows, and the data set was split into training and testing data sets. Logistic regression models were used to estimate accuracy for classifying BrAC ≤ .08% from BrAC > .08% for each subject. Across participants, BrAC > .08% was predicted with a mean accuracy of 92.5% using logistic regression, an improvement from a naive model accuracy of 88.2% (mean sensitivity = .89; specificity = .92; positive predictive value = .77; and negative predictive value = .97). The two most informative accelerometer features were mean signal amplitude and variance of the signal in the x-axis (i.e., gait sway). We found preliminary evidence supporting use of gait-related features measured by smartphone accelerometer sensors to detect alcohol intoxication. Future research should determine whether these findings replicate in situ.

Analyzing the Effectiveness and Contribution of Each Axis of Tri-Axial Accelerometer Sensor for Accurate Activity Recognition.

Recognizing human physical activities from streaming smartphone sensor readings is essential for the successful realization of a smart environment. Physical activity recognition is one of the active research topics to provide users the adaptive services using smart devices. Existing physical activity recognition methods lack in providing fast and accurate recognition of activities. This paper proposes an approach to recognize physical activities using only2-axes of the smartphone accelerometer sensor. It also investigates the effectiveness and contribution of each axis of the accelerometer in the recognition of physical activities. To implement our approach, data of daily life activities are collected labeled using the accelerometer from 12 participants. Furthermore, three machine learning classifiers are implemented to train the model on the collected dataset and in predicting the activities. Our proposed approach provides more promising results compared to the existing techniques and presents a strong rationale behind the effectiveness and contribution of each axis of an accelerometer for activity recognition. To ensure the reliability of the model, we evaluate the proposed approach and observations on standard publicly available dataset WISDM also and provide a comparative analysis with state-of-the-art studies. The proposed approach achieved 93% weighted accuracy with Multilayer Perceptron (MLP) classifier, which is almost 13% higher than the existing methods.

A Deep Learning Model for Transportation Mode Detection Based on Smartphone Sensing Data

Understanding people’s transportation modes is beneficial for empowering many intelligent transportation systems, such as supporting urban transportation planning. Yet, current methodologies in collecting travelers’ transportation modes are costly and inaccurate. Fortunately, the increasing sensing and computing capabilities of smartphones and their high penetration rate offer a promising approach to automatic transportation mode detection via mobile computation. This paper introduces a light-weighted and energy-efficient transportation mode detection system using only accelerometer sensors in smartphones. The system collects accelerometer data in an efficient way and leverages a deep learning model to determine transportation modes. Different architectures and classification methods are tested with the proposed deep learning model to optimize the system design. Performance evaluation shows that the proposed new approach achieves a better accuracy than existing work in detecting people’s transportation modes.

An Efficient Algorithm for Roadways Obstacles Detection using Smartphone in Infrastructure-less Environment

Apart from having smart systems in modern vehicles, roadways traffic positions in infrastructure-free environment are in various ways not suitable for driving which causes roadways accidents. Smart systems are costly, training and maintenance is not practiced by all, hence they are not useful for vehicles running in infrastructure-less environment. Modern days Smartphone are multifunctional, easily operated and consists of numerous sensors, can be utilized for roadways obstacles detection in infrastructure-less environment. In this study, we proposed an efficient algorithm depth learning approach for roadways obstacles detection using Smartphone which is useful for planned as well as unplanned roads. In this, road information is collected using Smartphone accelerometer sensors, data normalization take place using Euler angle method and location determination of obstacles take place using space interpolation method. In comparison to other same type of approach, our algorithm is efficient and cost-effective.

Smartphone Based Distance Estimation Algorithm Using Dynamic Step Length

In this paper, a new Smartphone sensor-based algorithm is proposed to detect accurate distance estimation. The algorithm consists of two phases, the first phase is for detecting the peaks from the Smartphone accelerometer sensor. The other one is for detecting the step length which varies from step to step. The proposed algorithm is tested and implemented in real environment and it showed promising results. Unlike the conventional approaches, the error of the proposed algorithm is fixed and is not affected by the long distance.

AN IOT BASED APPROACH TO VEHICLE ACCIDENT DETECTION, REPORTING AND NAVIGATION

Every day around the world, a large percentage of people die from traffic accident injuries. An effective approach for reducing traffic fatalities is: first building automatic traffic accident detection system, second, reducing the time between when an accident occurs and when first emergency responders are dispatched to the scene of the accident. Recent approaches are using built-in vehicle automatic accident detection and notification system. While these approaches work fine, they are expensive, maintenance complex task, and are not available in all cars. On the other hand, the ability to detect traffic accidents using smartphones has only recently become possible because of the advances in the processing power and sensors deployed on smartphones. Most of the smartphone based accident detection systems rely on the high speed of the vehicle (extracted from the smartphone GPS receiver) and the G-Force value (extracted from smartphone accelerometer sensor) to detect an accident. As many references assure that 90% of road-traffic accidents occur at low speed of the vehicle. Hence, in addition to the high speed accident detection, this paper concentrated on low speed car accident detection. The main obstacle that encounters the low speed accident is how to differentiate whether the user is inside the vehicle or outside the vehicle, walking or slowly running. The effect of this obstacle is minimized, in this work, by a proposed mechanism that distinguishes between the speed variation of low speed vehicle and walking or slowly running person. The proposed system consists of two phases; the detection phase which is used to detect car accident in low and high speeds. The notification phase, and immediately after an accident is indicated, is used to send detailed information such as images, video, accident location, etc. to the emergency responder for fast recovery. The system was practically tested in real simulated environment and achieved quite very good performance results.

A classification system for assessment and home monitoring of tremor in patients with Parkinson's disease

Tremor is one of the most common symptoms of Parkinson's disease (PD), which is widely being used in the diagnosis procedure. Accurate estimation of PD tremor based on Unified PD Rating Scale (UPDRS) provides aid for physicians in prescription and home monitoring. This article presents a robust design of a classification system to estimate PD patient's hand tremors and the results of the proposed system as compared to the UPDRS. A smartphone accelerometer sensor is used for accurate and noninvasive data acquisition. We applied short-time Fourier transform to time series data of 52 PD patients. Features were extracted based on the severity of PD patients' hand tremor. The wrapper method was employed to determine the most discriminative subset of the extracted features. Four different classifiers were implemented for achieving best possible accuracy in the estimation of PD hand tremor based on UPDRS. Of the four tested classifiers, the Naive Bayesian approach proved to be the most accurate one. The classification result for the assessment of PD tremor achieved close to 100% accuracy by selecting an optimum combination of extracted features of the acceleration signal acquired. For home health-care monitoring, the proposed algorithm was also implemented on a cost-effective embedded system equipped with a microcontroller, and the implemented classification algorithm achieved 93.8% average accuracy. The accuracy result of both implemented systems on MATLAB and microcontroller is acceptable in comparison with the previous works.

스마트폰 가속도 센서를 이용한 사용자 인증 방법 연구

With the growth of financial industry with smartphone, interest on user authentication using smartphone has been arisen in these days. There are various type of biometric user authentication techniques, but gait recognition using accelerometer sensor in smartphone does not seem to develop remarkably. This paper suggests the method of user authentication using accelerometer sensor embedded in smartphone. Specifically, calibrate the sensor data from smartphone with 3D-transformation, extract features from transformed data and do principle component analysis, and learn model with using gaussian mixture model. Next, authenticate user data with confidence interval of GMM model. As result, proposed method is capable of user authentication with accelerometer sensor on smartphone as a high degree of accuracy(about 96%) even in the situation that environment control and limitation are minimum on the research.

Toward Developing a Computational Model for Bipedal Push Recovery–A Brief

The human being can negotiate with external push up to certain extent reactively. Grown up persons have better push recovery capability than kids and also the professional wrestlers acquire better push recovery capability than normal human being. The acquired push recovery capability, therefore, is based on learning. However, the mechanism of learning is not known to us. Researchers around the world are trying to explore this mystery through developing various models and implementing them on various humanoid robots. All the models based on conventional mechanics and controls have inherent limitations. We believe appropriate computational model based on learning will be able to effectively address this issue. Accordingly, we have collected extensively humanoid push recovery data using our innovative idea of exploiting the accelerometer sensor of smart phone. Through our experiments, we have studied the human push recovery by fusing data at feature level using physics toolbar accelerometer of android interface kit. The subjects for the experiments were selected both as right handed and left handed. Pushes were induced from the behind with close eyes to observe the motor action as well as with open eyes to observe learning-based reactive behaviors. A learning vector quantization-based classifier has been developed to identify the coordination between various push and hip and knee joints.

Analyzing Driving and Road Events via Smartphone

ABSTRACT Travelling in India by road is considered dangerous as traffic conditions are chaotic, the drivers drive recklessly, and the roads are poorly maintained. Therefore, there is need to monitor driver behavior and road condition regularly. In this work, we developed a method that collects data using accelerometer sensor present in Smartphone and analyze driving events and road anomalies. This paper presents the patterns of driving events, such as left turn, right turn, sudden braking and sudden forward acceleration, and road conditions, such as pothole, bump and rough patch, detected in the data set. This work will help in analyzing driving behavior and road anomalies to ensure driver safety and maintenance of roads. General Terms Driving Events, Road Anomalies Keywords Left Turn, Right Turn, Sudden Braking, Sudden Forward Acceleration, Pothole, Bump and Rough patch. 1. INTRODUCTION Although there are various transport modes but road transportation mode still plays a major role in people lives. In this busy world all people in hurry to reach their destination as quickly as possible which further leads them to drive fast, recklessly, without obeying the traffic rules which further lead them to accidents sometimes also to their loss of lives. In India potholes, shards of glass, speed breakers, cow dung, garbage, mud pits and many other things that may be found on the road and also less attention is given to the maintenance of roads. So there arises the need to monitor the driving style and road anomalies to ensure driver safety and for the maintenance of roads. Previously a lot of work is done in this field but researchers mainly focused on monitoring either driver behavior or road conditions using specialized hardware deployed inside the car [1] [2] [3] [4] [5] [6] or roadside which is expensive and also requires maintenance. Our method, contrary from previous work, use accelerometer sensor of Smartphone to observe both the pattern of driver’s driving style and road anomalies. Smartphone based approach is considered good as it is inexpensive, portable, requires less maintenance and also relying to the fact that many people already own it. This paper is organized as follows, section 2 discusses already present techniques for detecting driving events detection and road anomalies, section 3 contains experimental setup for data collection followed by data preprocessing , section 4 discusses the different patterns analyzed of various driving events and road anomalies, and section 5 provides the conclusion.

Unsupervised posture detection by smartphone accelerometer

Proposed is a light-weight unsupervised decision tree based classification method to detect the user's postural actions, such as sitting, standing, walking and running as user states by analysing the data from a smartphone accelerometer sensor. The proposed method differs from other approaches by applying a sufficient number of signal processing features to exploit the sensory data without knowing any a priori information. Experiments show that the proposed method still makes a solid differentiation in user states (e.g. an above 90% overall accuracy) even when the sensor is operated under slower sampling frequencies.