Decoding human activity accurately from wearable sensors can aid in applications related to healthcare and context awareness. The present approaches in this domain use recurrent and/or convolutional models to capture the spatio-temporal features from time-series data from multiple sensors. We propose a deep neural network architecture that not only captures the spatio-temporal features of multiple sensor time-series data but also selects, learns important time points by utilizing a self-attention mechanism. We show the validity of the proposed approach across different data sampling strategies on six public datasets and demonstrate that the self-attention mechanism gave a significant improvement in performance over deep networks using a combination of recurrent and convolution networks. We also show that the proposed approach gave a statistically significant performance enhancement over previous state-of-the-art methods for the tested datasets. The proposed methods open avenues for better decoding of human activity from multiple body sensors over extended periods of time. The code implementation for the proposed model is available at https://github.com/isukrit/encodingHumanActivity.

The sudden traffic slowdown especially in fast scrolling roads and highways characterised by a scarce visibility is one of the major causes of accidents among motorised vehicles. It can be caused by other accidents, work-in-progress on roads, excessive motorised vehicles especially at peak times and so on. Typically, fixed traffic sensors installed on roads that interact with drivers’ mobile APP through the 4G network can mitigate such a problem, but unfortunately not all roads and highways are equipped with such sensors. In this paper, we discuss a possible alternative solution for addressing such an issue considering mobile traffic sensors directly installed in private and/or public transportation and other volunteer vehicles. In such a scenario a fast real-time processing of big traffic data is fundamental to prevent accidents. In particular, we discuss an IoT Cloud system for traffic monitoring and alert notification based on OpenGTS and MongoDB. Our IoT Cloud system, besides for private drivers, it is very useful for drivers of critical helpful service such as ambulances. Experiments prove that our system provides acceptable response times that allows’ drivers to receive alert message in useful time so as to avoid the risk of possible accidents.

Recent advancements in wireless communication technologies have led to the development of practical applications such as infrastructure monitoring, earthquake monitoring, environment monitoring, remote healthcare monitoring etc. In these applications, we may integrate the Wireless Sensor Networks (WSNs) with Internet of Things (IoT), where sensor nodes join the internet dynamically and use it to collaborate and accomplish their tasks. In e-healthcare system, wireless medical sensor nodes continuously monitor and collect the physiological information and transmit the same to healthcare professionals through IoT terminals. To overcome the limitations in communication and computational capabilities in sensors and to attain privacy in healthcare wireless medical sensor networks (HWMSNs), few signature schemes were proposed in literature. However most of these schemes are insecure against various attacks. Hence to ensure privacy and security in medical data, in this paper, we proposed an efficient pairing-free aggregate signature scheme for HWMSNs in certificateless system. We exploit aggregation technique in certificateless system without using pairings to reduce computational complexity and transmission overhead in transferring the data. Our scheme achieves full aggregation to improve communicational efficiency and performance analysis shows that the proposed scheme is more efficient.

This study addresses the problem of the automatic detection of disease states of the retina. In order to solve the abovementioned problem, this study develops an artificially intelligent model. The model is based on a customized 19-layer deep convolutional neural network called VGG-19 architecture. The model (VGG-19 architecture) is empowered by transfer learning. The model is designed so that it can learn from a large set of images taken with optical coherence tomography (OCT) and classify them into four conditions of the retina: (1) choroidal neovascularization, (2) drusen, (3) diabetic macular edema, and (4) normal form. The training datasets (taken from publicly available sources) consist of 84,568 instances of OCT retinal images. The datasets exhibit all four classes of retinal disease mentioned above. The proposed model achieved a 99.17% classification accuracy with 0.995 specificities and 0.99 sensitivity, making it better than the existing models. In addition, the proper statistical evaluation is done on the predictions using such performance measures as (1) area under the receiver operating characteristic curve, (2) Cohen’s kappa parameter, and (3) confusion matrix. Experimental results show that the proposed VGG-19 architecture coupled with transfer learning is an effective technique for automatically detecting the disease state of a retina.

This paper presents a novel, wearable navigation system for visually impaired and blind pedestrians that combines a global positioning system (GPS) for user outdoor localization and tactile-foot stimulation for information presentation. Real-time GPS data provided by a smartphone are processed by dedicated navigation software to determine the directions to a destination. Navigational directions are then encoded as vibrations and conveyed to the user via a tactile display that inserts into the shoe. The experimental results showed that users were capable of recognizing with high accuracy the tactile feedback provided to their feet. The preliminary tests conducted in outdoor locations involved two blind users who were guided along 380–420 m predetermined pathways, while sharing the space with other pedestrians and facing typical urban obstacles. The subjects successfully reached the target destinations. The results suggest that the proposed system enhances independent, safe navigation of blind pedestrians and show the potential of tactile-foot stimulation in assistive devices.

Convolutional Neural network (CNN) based spinal cord disease prediction has emerged as a reliable model in medical imaging applications. Spinal cord injury (SCI) detection is one of the major problems for disorder segmentation and classification. Traditionally, radiologists analyze SCI images manually in order to detect abnormal spinal disorders. Manual interpretation of high dimensional feature space makes it difficult to predict the exact category and level of severity. On the other hand, deep learning framework helps to diagnose accurately and quickly. Deep learning approach is used to classify normal abnormal SCI images automatically. This paper presents a deep learning framework for helping diagnose SCI features based on the segmentation process. In this paper, a novel CNN-deep segmentation based boosting classifier is applied on sensor SCI image data. A real-time wearable sensor is used to capture the spinal cord disorder data with different shapes and orientations. Experimental results show that the present CNN-deep segmentation based boosting classifier has high computational SCI disorder prediction compared to the existing CNN based classifiers. Experimental results proved that the present model has better performance than the existing spinal cord injury detection models in terms of true positive rate; TP=0.9859, Accuracy=0.9894, and Error rate=0.019 are concerned.