Search

Tire Pressure Monitoring Systems (TPMS) are electronic wireless systems that monitor and report air pressure inside pneumatic tires in real time. An example of a TPMS module integrated with the valve stem and showing the typical tire mounting location is shown in Figure 1. For their safety and fuel economy benefits, starting with the mid-2000's, active TPMS were mandated on many vehicles worldwide. The NHTS estimates that there are approximately 23,000 accidents and 535 fatalities annually involving tire underinflation and blowouts [1]. The use of TPMS has been shown to result in improved fuel economy and therefore reduced carbon emissions [2]. TPMS in passenger vehicles was mandated in the US as of Sept. 1, 2007 under the TREAD Act, in the European Union as of Nov. 1, 2012 and in South Korea as of Jan.1, 2013. Countries like Russia, Indonesia, the Philippines, Israel, Malaysia, Turkey and many others soon followed [3]. The first TPMS systems were large and bulky with a significant electronics content [4]. Since that time TPMS electronics have gotten more energy efficient and form factors have come down dramatically. This paper will outline an effort to miniaturize an existing 1.0 mm pitch, 7x7x2.2 mm body size 24 lead QFN (Quad Flat No Leads) TPMS down to a 4x4x1.98 mm body size QFN with 0.5 mm pitch that would still meet automotive AEC Grade 1 reliability requirements. The original 7x7 mm three die QFN package consisted of an ASIC, a pressure sensor and an accelerometer. This miniaturization led to many technical challenges at both the package and board level. This paper will primarily address the board level reliability (BLR) challenges encountered due to the large silicon to package ratio along with the 50% reduction in pitch. Through a series of test vehicles with variables such as QFN leadframe surface finish, lead shape and size, wettable flank (WF) technology and anchors pads, the BLR was successfully improved to the point where it met application requirements.

PurposeAs drones are rapidly transforming tasks such as mapping and surveying, safety inspection and progress monitoring, human operators continue to play a critical role in ensuring safe drone missions in compliance with safety regulations and standard operating procedures. Research shows that operator's stress and fatigue are leading causes of drone accidents. Building upon the authors’ past work, this study presents a systematic approach to predicting impending drone accidents using data that capture the drone operator's physiological state preceding the accident.Design/methodology/approachThe authors collect physiological data from 25 participants in real-world and virtual reality flight experiments to design a feedforward neural network (FNN) with back propagation. Four time series signals, namely electrodermal activity (EDA), skin temperature (ST), electrocardiogram (ECG) and heart rate (HR), are selected, filtered for noise and used to extract 92 time- and frequency-domain features. The FNN is trained with data from a window of length t = 3…8 s to predict accidents in the next p = 3…8 s.FindingsAnalysis of model performance in all 36 combinations of analysis window (t) and prediction horizon (p) combinations reveals that the FNN trained with 8 s of physiological signal (i.e. t = 8) to predict drone accidents in the next 6 s (i.e. p = 6) achieved the highest F1-score of 0.81 and AP of 0.71 after feature selection and data balancing.Originality/valueThe safety and integrity of collaborative human–machine systems (e.g. remotely operated drones) rely on not only the attributes of the human operator or the machinery but also how one perceives the other and adopts to the evolving nature of the operational environment. This study is a first systematic attempt at objective prediction of potential drone accident events from operator's physiological data in (near-) real time. Findings will lay the foundation for creating automated intervention systems for drone operations, ultimately leading to safer jobsites.

Race logic is a relative timing code that represents information in a wavefront of digital edges on a set of wires in order to accelerate dynamic programming and machine learning algorithms. Skyrmions, bubbles, and domain walls are mobile magnetic configurations (solitons) with applications for Boolean data storage. We propose to use current-induced displacement of these solitons on magnetic racetracks as a native temporal memory for race logic computing. Locally synchronized racetracks can spatially store relative timings of digital edges and provide non-destructive read-out. The linear kinematics of skyrmion motion, the tunability and low-voltage asynchronous operation of the proposed device, and the elimination of any need for constant skyrmion nucleation make these magnetic racetracks a natural memory for low-power, high-throughput race logic applications., Comment: 9 pages, 3 figures, submitted for review

Contamination of electroencephalogram (EEG) signals due to natural blinking electrooculogram (EOG) signals is often removed to enhance the quality of EEG signals. This paper discusses the possibility of using solely involuntary blinking signals for human authentication. The EEG data of 46 subjects were recorded while the subject was looking at a sequence of different pictures. During the experiment, the subject was not focused on any kind of blinking task. Having the blink EOG signals separated from EEG, 25 features were extracted and the data were preprocessed in order to handle the corrupt or missing values. Since spontaneous and voluntary blinks have different characteristics in terms of kinematic variables and because the previous studies’ control setup may have altered the type of blink from spontaneous to voluntary, a series of statistical analysis was carried out in order to inspect the changes in the multivariate probability distribution of data compared to the previous studies. Statistical significance shows that it is very likely that the blink features of both voluntary and involuntary blink signal are generated by Gaussian probability density function, although different than voluntary blink, spontaneous blink is not well discriminated with Gaussian. Despite testing several models, none managed to classify the data using only the information of a single spontaneous blink. Thereby, we examined the possibility of learning the patterns of a series of blinks using Gated Recurrent Unit (GRU). Our results show that individuals can be distinguished with up to 98.7% accuracy using only a reasonably short sequence of involuntary blinking signals.

OBJECTIVE Cortical resilience can be defined as the ability of the cortex of the human brain to rebound efficiently from perturbation. This concept is important in both research and clinical practice contexts. However, no direct measure of cortical resilience exists. Inhibitory variants of repetitive transcranial magnetic stimulation, such as continuous theta burst stimulation, provide a standardized method for inducing a perturbation; when coupled with the assessment of recovery rate from the perturbation, such a paradigm might provide a standardized measure of cortical resilience. This article describes a standardized method for quantifying cortical resilience using theta burst stimulation protocols. METHODS A descriptive overview of a method for assessing cortical resilience is presented. Links are drawn between critical facets of the resilience construct and the protocol described. RESULTS The Cortical Challenge and Recovery Test (CCaRT) method makes use of existing stimulation parameters and cognitive testing paradigms to provide a flexible and conceptually meaningful measure of cortical resilience. CONCLUSIONS The CCaRT paradigm is potentially useful in research and contexts where cortical resilience is to be measured. The CCaRT paradigm has applicability to epidemiological studies and laboratory experimentation as well as diagnostic practice and clinical trial outcome measures.

The burgeoning growth of Big Data not only matures and improves the data management efficiency and useful information extraction techniques, but also motivates the computational science researchers to come up with a new method or solution that can be repurposed for problems across the domain. Computational Sustainability joins this movement for a transferrable computational technique for sustainable development and a better future. Internet-of-energy (IoE) - leveraging IoT to smart grids associated with advanced analytics - is one of the prominent efforts in this regard. This paper presents a qualitative analysis on the elements of the energy and power management ecosystem in the United States. This qualitative study includes the Grid Overview of the United States; Weather and Climate and its impact on the entire energy generation and consumption dynamics; Peak Load Forecasting and its techniques and burgeoning challenges; Variable Renewable Energy, its reliability challenges and how we can take advantage of this variability; Commodity Prices and its criticality; Energy Disaggregation and its impact on consumption-awareness; and Generation Expansion and Decision Analysis. Besides, IoE integration, associated trade-offs, challenges, research opportunities and transferable computational techniques are addressed in this communication. Furthermore, schematics and quantitative analysis are presented in support of this study.

Researchers face difficulties in finding relevant papers to their research interest as the number of scientific publication is rapidly increasing on the web. Scientific paper recommenders have emerged as a leading solution to help researchers by automatically suggesting relevant and useful publications. Several approaches have been proposed on improving recommender systems. However, most existing approaches depend on priori user profiles, and thus they cannot recommend papers to new user. Furthermore, the existing approaches utilize non-public contextual information, and thus it cannot adequately find similarities between papers due to copyright restrictions. Also, the existing approaches consider only single level paper-citation relation to identify similarities between papers. Considering the above challenges, this paper presents a collaborative filtering based recommendation approach for scientific papers that does not depend on priori user profiles and which utilizes only public contextual information. Using citation context, we utilized 2-level paper-citation relations to find hidden associations between papers. The rational underlying this approach is that, two papers are co-occurred with same cited paper(s) and two papers are co-occurring with same citing paper(s) are significantly similar to some extent. To evaluate the performance of the proposed approach, publicly available datasets are used to conduct extensive experiments. The experimental results demonstrate that the proposed approach has significantly outperforms the baseline approaches in terms of precision, recall, F1, mean average precision, and mean reciprocal rank, which are commonly used information retrieval metrics. The novelty of this study is that, with the proposed approach, researchers are able to find relevant and useful publications over the internet regardless of their previous research experiences and research area.

Magnetic Topological Textures (MTTs) have the potential to become the building block of future memory and logic-in-memory devices due to their compact size and non-volatility. However, these MTTs are used in conjunction with the CMOS control circuit to design memory and logic-in-memory devices. Micromagnetic and SPICE simulations need to be carried out separately for MTT and CMOS, respectively, to evaluate the performance of these hybrid devices. Thus, a SPICE model that can emulate the physical behavior of the MTT would allow the designers to evaluate the performance of these hybrid devices accurately and enable design space exploration. In this paper, we build a compact and modular SPICE model of MTT nucleation, motion, detection, and annihilation on the racetrack by extracting parameters from micromagnetic simulation tools. We also build a CMOS-MTT generic racetrack circuit model in SPICE and validate the model’s accuracy by comparing it with micromagnetic simulation results. As we can perform the simulations fully within SPICE while capturing the essential physics of the MTTs on the racetrack, we enable simulation runtime acceleration by a factor of > 103 and > 106 times approximately compared to the micromagnetic simulations carried out with MUMAX3 and OOMMF. We envision that the proposed model will enable design and application explorations and device-circuit co-simulation, co-design, and co-optimization for the emerging MTT-based computing systems.

Detecting modes of transportation through human activity recognition is important in the effective and smooth operation of smartphone applications or similar portable devices. However, the effectiveness of such tasks depends on the nature and type of data provided, and it can often become quite challenging. “SHL recognition challenge 2021” is an activity recognition challenge that aims to detect eight modes of locomotion and transportation. The dataset in this challenge was based on radio data, including GPS reception, GPS location, Wi-Fi reception, and GSM cell tower scans. The objective was to create a model that was able to recognize the modes in a user-independent manner. In this paper, our team (Team Nirban) gives an appropriate summarization of our methods and approach to the challenge. We processed the data, extracted various features from the dataset, and selected the best ones, which helped our model to be generative and user-independent. We exploited a classical machine learning based approach and achieved 93.4% accuracy and 89.6% F1 score on the training set using 10-fold cross-validation, as well as 62.3% accuracy on the provided validation set.

Annotation problem in Human Activity Recognition (HAR) has always been a major concern. The main problem of human-based annotation is that it often leads to misclassification due to uneagerness, hectic working and living conditions, misleading emotional fluctuation, etc. A solution for that can be the use of a fully automated annotation system. Here, we proposed a unique method for developing a pre-trained model which will be used as the auto annotation tool, using two completely different benchmark Activities of Daily Living (ADL) in the wild data sets. Our proposed auto labeling system has been trained in varied conditions which reduce the bias from location, people, and device, and makes our system very robust. Here, we used a unique mixture of augmentation, scaling, and boosting to increase accuracy, reduce overfitting, and handle imbalanced class problems. The accuracy achieved for auto annotation in case of data set 1, data set 2, and combined data set are 92%, 87%, and 73% respectively. Also, performance evaluation has been done using precision, recall, f1 score, and ROC curve, and the results achieved are promising.

Data conversion from analog to digital and time domains and data transfer costs are the main bottlenecks in designing energy & area efficient near-sensor processing architectures. In this paper, we propose a temporal Processing in Memory architecture that addresses both data conversion and data transfer costs. We develop a novel analog to time encoding method that translates the incoming analog signal directly into a time-coded value, reducing data conversion costs. To do this, we exploit the linear kinematics of skyrmions in magnetic racetracks and propose a skyrmion-based non-volatile temporal memory architecture that employs time encoding. We also reduce the data transfer costs by processing the temporally stored data during the memory read operation. A read unit shared between the temporal memory cells acts as the processing element (PE) in our proposed design. Thus, the proposed approach enables the reduction of both data conversion and data transfer costs achieving < 340 fJ/conversion and approximately 530-660 fJ/operation.

Rapid increase in scholarly publications on the web has posed a new challenge to the researchers in finding highly relevant and important research articles associated with a particular area of interest. Even a highly relevant paper is sometimes missed especially for novice researchers due to lack of knowledge and experience in finding and accessing the most suitable articles. Scholarly recommender system is a very appropriate tool for this purpose that can enable researchers to locate relevant publications easily and quickly. However, the main downside of the existing approaches is that their effectiveness is dependent on priori user profiles and thus, they cannot recommend papers to the new users. Furthermore, the system uses both public and non-public metadata and therefore, the system is unable to find similarities between papers efficiently due to copyright restrictions. Considering the above challenges, in this research work, a novel hybrid approach is proposed that separately combines a Content Based Filtering (CBF) recommender module and a Collaborative Filtering (CF) recommender module. Unlike previous CBF and CF approaches, public contextual metadata and paper-citation relationship information are effectively incorporated into these two approaches separately to enhance the recommendation accuracy. In order to verify the effectiveness of the proposed approach, publicly available datasets were employed. Experimental results demonstrate that the proposed approach outperforms the baseline approaches in terms of standard metrics (precision, recall, F1-measure, mean average precision, and mean reciprocal rank), indicating that the proposed approach is more efficient in recommending scholarly publications.

Energy efficiency plays a significant role for the network lifetime extension in wireless sensor networks (WSNs). On the other hand, higher priority information requires Quality of Service (QoS) to ensure their delivery with minimized delay. Furthermore, WSN nodes are low powered devices powered by a small battery where the capacity is limited. Therefore, there is a need to establish an energy-efficient Medium Access Control (MAC) protocol with minimum packet delays. In this paper, we propose an adaptive energy-efficient synchronous Quality of Service (QoS) MAC protocol, called Timeout Multi-priority based MAC (TMPQ-MAC) protocol. TMPQ-MAC proposed an adaptive energy-efficient MAC protocol which can dynamically extend and end the active time to handle the load variations in the network. The TMPQ-MAC also considers four different packet priorities so that data packets with urgent information get transmitted with minimum delay. Finally, the performance of the protocol is evaluated through simulations and provided a comparison with Multi-priority QoS MAC (MPQ-MAC), QoS Aware Energy-Efficient MAC (QAEE-MAC) and Energy Harvested Receiver Initiated MAC (ERI-MAC) protocols. The simulation results show that the TMPQ-MAC improves the energy consumption at the sender nodes and the average packet delays by up to 7.8% and 7.9%, respectively over other protocols.

We envision a unique social interaction system, ‘users-as-beacons’ built upon Bluetooth Low Energy (BLE) beacon technology, that could provide potential privacy benefits. It leverages BLE to employ the user devices to act as mobile beacons. Its potential applications include community-based social networking, localized advertising, and instant reviewing. To evaluate the potential for this system and inform design, we conducted an exploratory interview study of 27 participants of a hypothetical localized content-creating system. Using a design prototype and multiple scenarios as prompts, we asked questions regarding users’ perceptions of the potential benefits and challenges of a users-as-beacons system, focusing in particular on their privacy concerns and needs. Our results indicate that users do perceive the benefit of increased trustworthiness of user-beacons, but do not have expectations of greater location or behavioral tracking privacy. We highlight multiple design challenges of this system in supporting the trustworthy, relevant, and timely sharing of posts between people in a community.

Finding a means of clean and free source of energy has become vital in this age and era. With rapidly shifting technological advancements, the necessity for harvesting energy is a top priority. Moreover, the time has come to utilize the core of humanity, the human body itself, for greater purposes. In this report, thorough and proper research has been carried out to learn extensively about human gait and biomechanical movement and the means of extracting clean energy from that. Proper research along with practical experiments including that on different human test subjects of different body mass index, weight, and gender has been carried out in order to find the forces acting on the plantar region considering different test subjects of varying weight, age, gender, and body mass index to find out plantar pressure distribution and the precise distribution of forces acting. A working prototype has then been modelled and devised based on the findings from the research. The prototype has been made from 3D printing, a combination of piezoelectric materials is then tested and the results are tabulated which are provided herewith.

Data collection using unmanned aerial vehicles (UAVs) in construction and heavy civil projects is subject to compliance with strict operational rules and safety regulations. Both the US Fed...

The use of wireless sensors is increasing day by day. Different types of wireless sensors are being used in fisheries sectors to monitor the water quality, growth of the fish, and health of the fish. Due to a brupt changes in water quality parameters, the rapid outbreak of fish disease has become a significant constraint for this sector's sustainability. Development of an early monitoring system of fish culture parameters through high resilience and efficiency wireless sensor networks (WSN) effectively assess water quality regulators instantly and thus take proper actions for sustainable management of freshwater resources. However, current observation systems only consider the data from a single sensor. We designed a data fusion model using Dempster-Shafer theory (DST) to fuse the monitoring sensor data from different sensors to calculate the fish's sustainable environment. Moreover, we evaluated our monitoring system results for different scenarios using the standard performance metrics, i.e., specificity, sensitivity, a ccuracy, and F-Score were calculated using the True Positives (TP), False Positive (LIP), True Negative (TN), and False Negative (LIN) values. Our model's finding shows that fusing data from different sensors provide a more accurate result for monitoring the water's sustainability.

Security is the basic need of hour and home automation is useful for the ease of people. There are many security systems which are being used in industry, but these are too expensive and difficult to use. There is another issue in the industry due the sequential processing of microcontrollers, which delays the work and consume more energy. In this paper we are focusing on a security system to reduce security threats by using advanced equipment such as FPGA programmed by VHDL language. Field Programmable Gate Array (FPGA) based security system can be utilized as a door lock in hotels, banks and everywhere etc. When a Radio Frequency Identification Device (RFID) tag is scanned by an its reader, the door lock will open if the tag information is already stored in the database of FPGA otherwise alarm will ring. Moreover, after opening the door lock, light or fan will automatically be turned on according to the need of that specific user. This project is a combination of two features i.e.; security and automation. Both features are basic needs in market. Our project can be commercialized in hotels, restaurants, schools, colleges, universities and many other places.

Facebook is the most popular social media platform all over the world and a huge data repository that can be analyzed to predict events as people express their thoughts on this platform. There are some events that often turn into potential crises and these crises might be avoided if we can predict it earlier. This work aims to predict events that are happening or going to happen by analyzing Bengali Facebook posts and help authorities to take proper security arrangements. In this work, the Naive Bayes Classification model has been used to classify posts as an event. Tokenization, Stopwords removing have been used for data pre-processing. Event phrase matching and 2-Layer Filtering have been used for features collection. Sentiment scores have been measured using Valence Aware Dictionary and Sentiment Reasoner(VADER). Finally, this work predicts four types of events(protesting, celebrating, religious, and neutral) showing 87.5% accuracy.

Parkinson’s disease(PD) is a neurological condition that is dynamic and steadily influences the movement of the human body. PD influences the central apprehensive system which happens because of the hardship of dopaminergic neurons brought about in a neuro-degenerative incubation. The patients who have PD usually suffer from tremor, unyielding nature, postural shifts, and lessen in unconstrained advancements. There is no particular diagnosis process for PD. PD varies from one person to another person depending on the situation and the family history.Magnetic Resonance Imaging (MRI), Computed Tomography (CT), ultrasound of the brain, Positron Emission Tomography (PET) scans are common imaging tests to figure out this disease but these tests are not particularly effective. In this research, several tests are run on two types of data group - control and PD affected people. The dataset is collected from the Parkinson’s Progression Markers Initiative (PPMI) repository. Then MRI slices are processed from the selected data group into the CNN models. Three different Convolutional Neural Network (CNN) architectures are used in this work to extract features from the data group. The CNN models are InceptionV3, VGG16 and VGG19. These models are used in this research to compare and to get better accuracy. Among these models, VGG19 worked best in the dataset because the accuracy for VGG19 is 91.5% where VGG16 gives 88.5% and inceptionV3 gives 89.5% accuracy on detecting PD.

Human activity recognition (HAR) is a wide field of study which identifies a person's specific movement or behavior based on sensor data. Recognition of human behavior is the origin of many technologies, such as those concerned with personal biometric signatures, sports training, digital computing, security, health and fitness tracking, ambient-assisted living and management. Studying recognition of human activity shows that researchers are mostly interested in human everyday activities. HAR models input is the reading of the raw sensor data, and output is the prediction of the movement activities of the user. The HAR framework is becoming an evolving discipline in intelligent computing applications in the field of pervasive computing. In our study, we applied several machine learning algorithm along with some preprocessing techniques to identify which algorithm performs better in dataset acquired from the WISDM laboratory, which is available in public domain. The experiment shows that the highest accuracy is achieved in phone accelerometer data using Principal Component Analysis (PCA) with Random Forest (RF) than any other algorithm and preprocessing techniques in terms of human activity recognition. This experiment will help perform more work on the basis of implementing classification and preprocessing techniques to identify human activities.

Bengali is one of the most spoken languages in the world. Nowadays, many people share their thoughts, emotions, and ideas through social media using the Bengali language. Facebook comments are one of the sources of sharing opinions among them. The goal of this study is to classify positive and negative emotions accurately with the help of the Facebook comments dataset. For this work, different deep learning models called Stacked Long short-term memory (LSTM), Stacked LSTM with 1D convolution, CNN with pre-trained word embeddings and, RNN with pre-trained word embedding have been used. The RNN model with pre-trained word embeddings has performed the best among all models with 98.3% accuracy. Other models do not perform well as the RNN model.

Human activity recognition has important applications in healthcare, human-computer interactions and other arenas. The direct interaction between the nurse and patient can play a pivotal role in healthcare. Recognizing various activities of nurses can improve healthcare in many ways. However, it is a very daunting task due to the complexities of the activities. "The 2nd Nurse Care Activity Recognition Challenge Using Lab and Field Data'' provides sensor-based accelerometer data to predict 12 activities conducted by the nurses in both the lab and real-life settings. The main difficulty of this dataset is to process the raw data because of a high imbalance among different classes. Besides, all activities have not been performed by all subjects. Our team, 'Team Apophis' has processed the data by filtering noise, applying windowing technique on time and frequency domain to extract various features from lab and field data distinctly. After merging lab and field data, the 10-fold cross-validation technique has been applied to find out the model of best performance. We have obtained a promising accuracy of 65% with an F1 score of 40% on this challenging dataset by using the Random Forest classifier.

CKD is a serious reason of demise and disability. It was the 27th focal reason in 1990 and became 18th focal reason in 2010. Near about 1 million people lose their life in 2013. In spite of that, people of developing countries are being affected by CKD. We analyzed the data of CKD patient and proposed a system from which it will be possible to predict the risk of CKD. We have used 455 patients' data. Online data set which is collected from UCI Machine Learning Repository and real time dataset which is collected from Khulna City Medical College are used here. We used Python as a high-level interpreted programming language for developing our system. We trained the data using 10-fold CV and applied Random forest and ANN. The accuracy achieved by Random forest algorithm is 97.12% and ANN is 94.5%. This system will help to predict early disclosure of chronic kidney diseases.

Renal cell cancer nuclei segmentation from histopathology image is an important step in cancer diagnosis and treatment. However, an automatic cell nuclei segmentation in medical application is a difficult task. Particularly a large amount of annotated data is required for deep learning network. In this work, synthetic but annotate renal cell nuclei data are generated based on non-synthetic data reference. Furthermore, background texture is created according to reference image. The Random shapes of nuclei polygon are created where polygon regions are filled up with foreground nuclei color. The synthetic cell nuclei patches are refined with speeded-up robust feature selection algorithm. The non-synthetic reference patches are considered to assign a score for each synthetic patch where synthetic patches with highest scores are collected. The synthetic patches with corresponding masks are utilized to train U-net segmentation network, which provides better performance than existing methods. In this approach, manual data annotation does not require, which is time-consuming and expensive. The renal cell cancer histopathology data has been collected from Cancer Image Archive, USA which is used to produce non-synthetic reference patches. The proposed framework is validated with Kidney Renal Clear Cell Carcinoma (KIRC) datasets. We have achieved average 0.923 cell nuclei segmentation accuracy from histopathology patches where accuracy provides better segmentation result than other existing methods.

The paper focuses on the application of fuzzy logic as a Multi-Criteria Decision Making (MCDM) method to develop an idea of a managerial tool that can predict accidents in different industries. Accidents in industries are not directly dependent on the amount of expenses in different relevant sectors. Fuzzy logic enables us to quantify the opinion of experts with many years of experience and combine those quantified opinions with other parameter affecting the occurrence of accidents in industries which in this case are the amount of expenses in different relevant sectors. This combined assessment generates a relationship between the occurrence of accidents and the amount of expenses in different relevant sectors. The tool will help management to mitigate accidents which is more preferable than reacting to accidents.

Non-invasive methods in the assessment of health parameters are important for various kinds of patient, but only few expensive and inconvenient solutions are available. Investigations of non-invasive methods beginning with simple red-green-blue (RGB) imaging extending to hyperspectral camera-based image have encountered portability, usability, and reliability issues. This paper presents a smartphone-based non-invasive hemoglobin level prediction model that addresses portability, accuracy and ease-of-use problems by taking advantage of the built-in high-resolution camera, significant computation ability, storage, and communication facility of current smartphones. In this research work, RGB information of a 10-second fingertip videois used to convert into different color spaces including hue (H), saturation (S), value (V), lightness (L), a, b (a and b for the color dimensions) and gray (g). Later, features are extracted from all the combinations of ten different colors and applied to a Partial Least Squares (PLS) algorithm. We have evaluated the best color combination to generate a prediction model with the data collected from American and Bangladeshi patients. Our model with a specific pixel color combination achieves a reliable accuracy.

The concept of maintenance has advanced significantly over the last decades from a reactive service activity towards a pro-active one with more C-level attention. Among them, predictive maintenance has become a widely used term in industrial arena and academic research. This is being developed constantly by engineers and researchers based on monitoring historical data, modeling, simulation, and failure probabilities to predict fault and system deterioration over their useful life. Generally, the effective lifetime of machines depends on available and accessible data. However, certain unexpected situations may arise that are hard to predict, such as shock damage and unwanted degradation of the tool. Researchers are still working on understanding these problems and their effects on predictive maintenance. This paper presents an overview of condition-based predictive maintenance solutions that aim to avoid unexpected and unplanned failures during the manufacturing and operational process based on advanced data analytics. Furthermore, a brief illustration & discussion is presented on the challenges and opportunities of condition-based predictive maintenance and conclude a summary on future research.

Reducing number of forwarding nodes is the main focus of any broadcasting algorithm designed for ad-hoc wireless networks. All reliable broadcasting techniques can be broadly classified under proactive and reactive approaches. In proactive approach, a node selects a subset of its neighbors as forwarding node and announces the forwarding node list in the packet header during broadcast. On the other hand, no such forwarding list is generated in reactive approach. Rather, a node (cognitively) determines by itself whether to forward the packet or not based on neighbor information. Dominant pruning and Self-pruning are two example techniques that fall under proactive and reactive approach respectively. Between the two methods, dominant pruning shows better performance than self-pruning in reducing number of forwarding nodes as they work with extended neighbor knowledge. However, appended forwarding node list increases message overhead and consumes more bandwidth. As a result, the approach becomes non-scalable in large networks. In this paper, we propose a reactive broadcasting technique based on self-pruning. The proposed approach dubbed as “Improved Self-pruning based Broadcasting (ISB)” algorithm completes the broadcast with smaller packet header (i.e., with no overhead) but uses extended neighbor knowledge. Simulation results show that ISB outperforms dominant pruning and self-pruning. Furthermore, as the network gets more spread and denser, ISB works remarkably well.

Advancement in transportation system has boosted speed of our lives. Meantime, road traffic accident is a major global health issue resulting huge loss of lives, properties and valuable time. It is considered as one of the reasons of highest rate of death nowadays. Accident creates catastrophic situation for victims, especially accident occurs in highways imposes great adverse impact on large numbers of victims. In this paper, we develop an intelligent accident detection, location tracking and notification system that detects an accident immediately when it takes place. Global Positioning System (GPS) device finds the exact location of accident. Global System for Mobile (GSM) module sends a notification message including the link of location in the google map to the nearest police control room and hospital so that they can visit the link, find out the shortest route of the accident spot and take initiatives to speed up the rescue process., The 3rd IEEE International Conference on Telecommunications and Photonics (ICTP) 2019

We propose a method to tackle the problem of mapless collision-avoidance navigation where humans are present using 2D laser scans. Our proposed method uses ego-safety to measure collision from the robot's perspective while social-safety to measure the impact of our robot's actions on surrounding pedestrians. Specifically, the social-safety part predicts the intrusion impact of our robot's action into the interaction area with surrounding humans. We train the policy using reinforcement learning on a simple simulator and directly evaluate the learned policy in Gazebo and real robot tests. Experiments show the learned policy can be smoothly transferred without any fine tuning. We observe that our method demonstrates time-efficient path planning behavior with high success rate in mapless navigation tasks. Furthermore, we test our method in a navigation among dynamic crowds task considering both low and high volume traffic. Our learned policy demonstrates cooperative behavior that actively drives our robot into traffic flows while showing respect to nearby pedestrians. Evaluation videos are at https://sites.google.com/view/ssw-batman, Accepted in ICRA 2020

Public speaking anxiety (PSA) ranks as a top social phobia across the world caused by various confounding factors. Motivated by the inherent data sparsity and lack of annotations in human-related applications, we propose unsupervised learning techniques to estimate PSA from speech signals. The labeled source data come from the publicly available CREMA-D dataset, while the unlabeled target data come from real-life public speaking tasks. Since fear is one of the major factors of PSA, the goal of this study is to build fear-specific representations from the labeled source data to estimate the degree of fear in the target data, and examine the extent to which the latter is associated with anxiety during the public speaking encounter. Transfer learning is performed through the domain-adversarial neural network (DANN) and Wasserstein generative adversarial network (WGAN). Results indicate that the proposed unsupervised fear- specific estimates can detect public speaking anxiety with Pearson’s correlation coefficient of 0.28 (p

Electromyography (EMG) is a biomedical signal which is recorded from muscle. It can detect muscle movements. By using HMI (human machine interfacing), EMG can be used to control prosthetic arms by an amputee individual. The available prosthetic arms and EMG signal recorders both are very expensive. The price of commercial prosthetic arm starts from 100,000 BDT. So, an affordable prosthetic arm is required for the people of Bangladesh. In this project an EMG recorder circuit along with a prosthetic arm was designed. The arm was printed using 3D printer and the recorder circuit was also printed in a circuit board (PCB) and built with available components. The 3D printed prosthetic arm was interfaced with the EMG recorder by Arduino microcontroller. The recorded signal was processed in both analog and digital domain. The arm was tested by an amputee individual who was able to control the prosthetic arm by muscle contraction which was detected by the EMG recorder. To minimize the cost, this project was completed using locally available resources. Total cost of this project was within 4,000 BDT. People who are amputee can use this 3D printed prosthetic arm to improve their quality of life.

SiC based Solid State Circuit Breakers (SSCBs) have recently entered into the market at the low voltage end of the network, and their adoption is slowly moving upstream as higher voltage offerings mature. Unavailability of commercial medium voltage SiC devices and their low current handling capability and high dv/dt hinder medium voltage SSCB designs. However, series connection of low voltage devices can meet the required voltage level if dynamic and static voltage balancing is possible. In this paper, an active voltage clamping method is proposed to achieve balanced dynamic turn-off voltage sharing between cascaded devices (CREE 1.2kV/30A SiC MOSFETs). Also, this method minimizes the number of components, overall switching losses and delay time by avoiding bulky and expensive RC snubbers. Different possible reasons behind dynamic voltage unbalancing and the proposed balancing technique have been verified in the LTspice simulation software. In addition, experimental results have been carried out to validate the effectiveness of the proposed technique.

Human activity recognition is one of the important research topics in machine learning. Different algorithms have been proposed for human activity recognition in several years. But nurse care activity recognition is a new section in machine learning which mainly focused on physical activity recognition. We are the members of the team "Data Digger" accepted a challenge named "Nurse Care Activity Recognition Challenge". We used Carecom nurse care activity dataset for that challenge [14]. We compare several machine learning algorithms for the dataset. The proposed method was trained and evaluated on Carecom nurse care activity recognition dataset and got 69% accuracy after splitting the dataset.

Computer Vision has become the poster child for Deep Learning. The image classification accuracy of convolutional neural nets on benchmark data sets has increased every year since their inception. This has been aided with advances in feature fusion. The increase in the availability of imagetext occurrence has lead to text augmented feature spaces that have lead to higher accuracy in in image classification tasks. However, these works are limited to instances where text is readily available. This study presents an approach to featurize text within natural images with the goal of augmenting image features for image classification tasks. Text extraction and featurization in natural images is a challenging task due to challenges in reliable text localization and OCR results, both being impeded by the variability in image text and errors in OCR. We overcome these challenges by implementing a novel bounding box concatenation algorithm and a novel feature boosting algorithm. The result is a pipeline that encodes an image into a text feature space. Classifiers trained on the text based feature space have comparable accuracy to the state of the art Convolutional Neural Nets (CNN's) while being significantly inexpensive computationally. Moreover, the augmentation of text features to image features generates a hybrid feature space with a higher information content for a classification problem when compared to a feature space comprised exclusively of image features. Thus, we see a rise in classification accuracy across all state of the art machine learning algorithms.

Bluetooth Low Energy (BLE) beacons are an emerging type of technology in the Internet-of-Things (IoT) realm, which use BLE signals to broadcast a unique identifier that is detected by a compatible device to determine the location of nearby users. Beacons can be used to provide a tailored user experience with each encounter, yet can also constitute an invasion of privacy, due to their covertness and ability to track user behavior. Therefore, we hypothesize that user-driven privacy policy configuration is key to enabling effective and trustworthy privacy management during beacon encounters. We developed a framework for beacon privacy management that provides a policy configuration platform. Through an empirical analysis with 90 users, we evaluated this framework through a proof-of-concept app called Beacon Privacy Manager (BPM), which focused on the user experience of such a tool. Using BPM, we provided users with the ability to create privacy policies for beacons, testing different configuration schemes to refine the framework and then offer recommendations for future research.

In-vivo observing of patient's heart mechanics by means of seismocardiography (SCG) was recently rediscovered and presented (Demonstrated). Most of the time, seismocardiography is recorded utilizing 2D or 3D accelerometers set on patient's sternum. This research represents the ability and performance development of a PVDF piezoelectric film transducer by placing them in multiple positions on the subject to record real time mechanical movement of the heart (seismocardiogram and heart sounds) and furthermore of the breath related thorax movement. Patient's ECG signal was also recorded simultaneously to give a time reference of the cardiovascular movement and better assessment of SCG signal acquisition. The piezoelectric sensor has been able to record respiratory signal, seismocardiography and heart sounds all along. These signals later have been processed on MATLAB by constructing simple digital filter. The seismocardiogram waveform gives off an impression of being reliant on the different periods of the respiratory cycle. The recorded signals of piezoelectric sensor show important data such as pulse rate, respiratory cycle (Inhale and exhale), timing of mechanical occasion inside the heart cycle, and so forth.