International Conference on Internet of Things [Services Society]

In some cases, long battery life may be essential to IoT devices, and early failures of actuators and sensors because of the rapid discharging of battery may lead to unacceptably high replacement costs. Critical to the implementation of this Internet Of Things (IoT) is the design of energy-efficient solutions aiming toward a low consumption current and create a green society. Many IoT devices rely on small, rechargeable batteries, so charging a wireless battery is essential for several reasons. Much research and development are working on how can is powering IoT devices wirelessly. Wireless power transmission technology is the diffusion of microwave power transmission without using any physical support. The vision of future technology is the Internet of Things IoT charging device without wires. The objective of the scope of work is to combine the wireless power technology with a smart house using IoT. In this research paper, we designed and realized a wireless lighting technology using the fundamentals of microwave radiation. We will send microwave energy from (Position 1) to the receiver (Position 2) to turn up an LED lamp 10 W a distance (50 meters). So the proposed prototype takes account of all parameters above to deliver sufficient energy to turn on the LED lamp 10 W wirelessly on the distance of 50 meters at the smart house.

Fifth generation (5G) cellular network, is a promising network that provides user promising quality of services in various applications such as health, banking, education, etc. Security consideration is the most critical element facing the reliance upon this technology. Registration, authentication and key agreement protocols are considered the most important protocols in any cellular network, since the subscriber and the network trust each other and share a symmetric key. Third Generation Partnership Project (3GPP) developed security specifications for the user authentication with the provider. However, many researchers targeted these protocols and concluded that there are still some security issues and proposed authentication and key agreement protocols. On the other hand, blockchain is considered one of the emerging technologies that will have a great impact on our life in the coming days. Blockchain grants its security properties such as authenticity and integrity to various applications such as bitcoin, smart contracts, etc. Therefore, in this paper, we introduce a novel, efficient and secure authentication and key agreement protocol for 5G networks using blockchain. Our security analysis illustrates that the proposed scheme is secure and withstands the known attacks; denial of service, distributed denial of service, man in the middle, hijacking and compromising attacks. Furthermore, our performance evaluation shows that the proposed scheme is more efficient than the current scheme and also other schemes since it preserves the small battery properties of the user equipment and also preserves the bandwidth of the network.

Image recognition is a basic yet necessary task for real time applications. Although there is massive advancement in deep learning models for image recognition, they are all designed for normal images produced by frame-based cameras which are time and power hungry. This work aims to explore the possibility of using event based cameras (EBC) for performing image recognition based on asynchronous events produced in response to changes per pixel rather than waiting for the full-frame image, potentially leading to higher recognition speed and lower power consumption. First, we collect and label raw event data of the MNIST testing dataset using CeleX, a 1 Mega Pixel event-based sensor. Then we evaluate the raw data on a convolutional neural network (CNN) model that is pre-trained using the original full frame image dataset, to calculate the time difference between waiting for a full frame image against asynchronous events. The results demonstrate that using EBC allows us to achieve early image recognition which provides major benefits for real time processing in terms of time and power.

This paper presents a novel supervised on-line fault diagnosis strategy in Heating, Ventilation, and Air conditioning (HVAC) systems for actuator faults using 2D Convolutional Neural Networks. It is based on an efficient 1-Dimensional to 2-Dimensional data transformation that eliminates the need for advanced signals pre-processing. The proposed approach aims to address the limitations found in the previous works in terms of the diagnosis accuracy by adopting the recently evolving topology of the Convolutional Neural Networks. It is developed and validated using simulation data collected for a 3-zone HVAC system simulator using Transient System Simulation Tool (TRNSYS). The proposed approach demonstrates an improved performance when compared to the other data-driven approaches for actuator fault diagnosis in HVAC systems.

The concept of Scene-to-Speech (STS) is to recognize elements in a captured image or a video clip to speak loudly an informative textual content that describes the scene. The contemporary progression in convolution neural network (CNN) allows us to attain object recognition procedures, in real-time, on mobile handled devices. Considerable number of applications has been developed to perform object recognition in scenes and say loudly their relevant descriptive messages. However, the employment of multiple trained deep learning (DL) models is not fully supported. In our previous work, a mobile application that can capture images and can recognize the objects contained in them was developed. It constructs descriptive sentences and speak them in Arabic and English languages. The notion of employing multi-trained DL models was used but no experimentation was conducted. In this article, we extend our previous work to perform required assessments while using multiple trained DL models. The main aim is to show that the deployment of multiple models approach can reduce the complexity of having one large compound model, and can enhance the prediction time. For this reason, we examine the prediction accuracy for single DL model-based recognition and multiple DL model-based recognition scenarios. The assessments results showed significant improvement in the prediction accuracy and in the prediction time. In the other hand, from the end user aspect, the application is designed primarily for visually impaired people to assist them in understanding their surroundings. In this context, we conduct a usability study to evaluate the usability of the proposed application with normal people and with visually impaired people. In fact, participants showed large interest in using the mobile application daily.

In this work, we present a color image segmentation algorithm considering pairwise color projection. Our approach is based on mode detection of the 2-D histogram. The pairwise projection is chosen by considering the most relevant color components to represent the chromatic information. For this purpose we consider a luminance-chrominance space where the chromatic information is presented by two ones components and other one component presents the achromatic information (called luminance). In this work, we consider the CIE LUV color space which is more even in distribution and more accordant with humans perceive color. The modes detection process consists in the first step of partitioning the 2-D histogram in the UV chromaticity plane using the watershed transform, followed by a mode merging algorithm taking into account both the pixel color distribution in the pairwise color projection and their spatial arrangement in the image. We show on various images that the expected modes are correctly identified.

Crowd counting is applied in many areas including efficient resources allocation and effective management of emergency situations. In this paper, we survey and compare various crowd counting methods. Additionally, we identify the limitations of existing approaches and sketch an agenda for future work to address the identified open research challenges. Furthermore, we present an enhanced deep learning-based solution for crowd counting at bus stops.

Domain Adaptation techniques remain limited in accuracy and robustness due to data sparsity. In this paper, we present a new approach called Domain Adversarial network with Representation Learning (DARL), to improve domain adaptation by introducing an encoding layer as part of DARL model learning. We integrate a Stacked Denoising Autoencoder and Adversarial learning for the domain adaptation process. The advantage of the proposed method is that it can extract descriptive features under noisy conditions while still learning task discriminative features. The encoding under noisy reconstruction ensures both higher accuracy and increased robustness in the learning process. We evaluate DARL using Amazon review data set and the results showed superior accuracy and robustness compared to Domain Adversarial Neural Networks (DANN).

Carbon fiber reinforced composites are promising candidates for building advanced multifunctional structures with superior properties that are suitable for the next generation of automotive and aircraft applications. This study presents an experimental investigation into the effect of the major orientation of the composite hat section on the crushing behavior and load-carrying capacity of a composite double hat structure. The variation in load carrying capacities due various measurement and manufacturing factors can significantly affect the design and thus safety. Therefore, the uncertainty in load carrying capacities is implicitly considered by identifying the maximum and minimum values of the load-carrying capacities at all displacement values. Artificial neural network-based models are then developed and compared using the Mean Squared Error (MSE) measure, with the objective to predict the load-carrying capacity range at each displacement value, which implicitly considers the uncertainty of results. Three samples of each arrangement are statistically analyzed and utilized in the training. The results show that the `X' hat orientation outperforms the `O' hat orientation in terms of load-carrying capacity. On the contrary, the `O' hat orientation outperforms the `X' hat orientation in terms of crash force efficiency with a total value of 0.6 for the former in comparison to 0.5 for the latter. A two layers ANN-models are found best in terms of performance with total RMSE values of 55.5 N for `X' orientation and 515.8 N for `O' orientation.

In this paper, a hybrid approach of combing two machine learning algorithms is proposed to detect the different possible attacks by performing effective feature selection and classification. This system uses Random Forest algorithm for the feature selection to find the most important features combined with Classification and Regression Trees (CART) for the classification of the different attack classes. The proposed system was tested using the UNSW-NB15 dataset and the results show that the proposed method achieves a good performance compared with the existing algorithms.

Detection of images or objects in motion have been highly worked upon, and the detection has been incorporated, required and utilized in applications. A few of the limitations include the lack of computational resources, lack of strategic and methodological data analysis of the observed trained data. The accuracy of detection is dependent on movement, velocity of the objects and Illuminacy. Therefore, it is required that new techniques and strategies of detection are drafted, applied and recognized. In this work, we have worked upon a model based on scalable object detection, using Deep Neural Networks to localize and track people, cars, potted plants and other categories in the camera preview in real-time. The trained model from google inception model has been implemented in an android application. This integrated application works real-time and can be used handy in a mobile phone or any other smart device with minimal computational resources.

In this paper, we propose an automatic electronic gate system with authenticity confirmation from 3 individual modules, viz., car make and model, license plate and face detection. The ultrasonic sensor detects the stopped cars and initiate the camera to capture the car image. Connected Components identification and Optical Character Recognition (OCR) algorithms are performed to recognize the characters and numbers in the car plate. The car make and model detection algorithm uses feature extraction algorithms based on Difference of Gaussians (DoG) detector and Scale Invariant Feature Transform (SIFT) descriptor. The Euclidian distance measure identifies the suitable match for a query image with the one in database to determine the car make and model. The face detection of driver is carried out to ensure the security of intended premise of the system deployment. It is based on the Viola Jones algorithm. In the final stage, matching algorithms are applied to decide whether the image of the car plate, make and model and the face does not conflict with the details stored in the database. The smart electronic gate will open and let the car enter when the authenticity is confirmed. The system is evaluated by building a prototype electronic smart gate using Logitech C920 HD pro webcam, Toy car and LEGO NXT Motor Controlled Gate. Moreover we built a preliminary database from the images captured by surveillance cameras for make and model recognition of Qatar cars. Experimental evaluation on this dataset delivered an accuracy of 75%.

With the advancement in personal smart devices and pervasive network connectivity, users are no longer passive content consumers, but also contributors in producing new contents. This expansion in live services requires a detailed analysis of broadcasters' and viewers' behavior to maximize users' Quality of Experience (QoE). In this paper, we present a dataset gathered from one of the popular live streaming platforms: Facebook. In this dataset, we stored more than 1,500,000 live stream records collected in June and July 2018. These data include public live videos from all over the world. However, Facebook live API does not offer the possibility to collect online videos with their fine grained data. The API allows to get the general data of a stream, only if we know its ID (identifier). Therefore, using the live map website provided by Facebook and showing the locations of online streams and locations of viewers, we extracted video IDs and different coordinates along with general metadata. Then, having these IDs and using the API, we can collect the fine grained metadata of public videos that might be useful for the research community. We also present several preliminary analyses to describe and identify the patterns of the streams and viewers. Such fine grained details will enable the multimedia community to recreate real-world scenarios particularly for resource allocation, caching, computation, and transcoding in edge networks. Existing datasets do not provide the locations of the viewers, which limits the efforts made to allocate the multimedia resources as close as possible to viewers and to offer better QoE.

This paper proposes a data storage method that cost-efficiently manages a massive amount of short lifetime data continuously generated typically by IoT devices (called Live Data, e.g., surveillance camera images) considering real-time data retrieval. Along with the rapid growth of the IoT, many users expect to shift to an Open IoT in which they can share enormous amounts of Live Data. To efficiently use Live Data, it is important to use both hot and cold data storage. Data should be stored in cold storage, which is less expensive but has slow access speeds than hot storage, only if the data are unlikely to be used. The proposed method virtualizes distributed hot storage to provide two data placement mechanisms; (1) an original-data-only optimized allocation according to demand by migrating original data to the appropriate distributed hot storage, and (2) integrated data replacement in the case of capacity overflow by migrating data to the distributed hot storage storing less useful data. As a result of a performance evaluation, compared with a cache algorithm often used for traditional data, the proposed method can reduce operational costs (transmission costs for data migration and retrieval and miss-hit penalty costs) when the ratio of single miss-hit penalty costs to 1-hop transmission costs is more than 6.

The Electronic Health Records (EHR) sharing system is the state of art for delivering healthcare. The tools within the system can predict outcomes during the patient's lifetime, monitor how effective are treatments, track therapies and detect human errors. For main industry stakeholders, the priority is to ensure integrity and interoperability across the care continuum. Achieving this priority is still challenging considering an enormous amount of EHR data, security issues, and heterogeneity of healthcare information systems. To overcome these challenges, this work proposes BiiMED: a Blockchain framework for Enhancing Data Interoperability and Integrity regarding EHR-sharing. The proposed solutions include an access management system allowing the exchange of EHRs between different medical providers and a decentralized Trusted Third Party Auditor (TTPA) for ensuring data integrity. This work establishes a foundation for further research on dynamic data interoperability and integrity verification in a fully decentralized environment.

Modern applications for cultural content enrichment and management require low delay image retrieval methods in large databases. Classical image retrieval methods are suitable for certain applications but are also known to lack the ability of generalization and their use for low delay applications for retrieval tasks is not investigated for the context of cultural heritage. As a potential improvement, we propose a new approach for large scale image retrieval that uses a pre-trained CNN as a global features extractor and a clustering model trained on these features to regroup similarly looking images. For retrieval, this model quickly identifies the closest cluster and then, the matching is only carried out for the images of the selected cluster. As a result, our approach does not require indexation and the preliminary results show that it is suitable for real-time and low delay applications as in the matching step, no heavy processing is required. Some of the suitable applications include quick image search and digital rights management.

Data missing is a vitally important issue that influences the classification results in medical field. This paper proposes an improved support vector machine (SVM) imputation algorithm by using strategies of pre-imputation, multiple iteration and grid search (IG-SVMI). Based on the experimental performance, nine UCI datasets and two real datasets are used to compare the proposed algorithm with four existing imputation algorithms (RFI, KNNI, CCMVI and orthogonal coding SVMI). The datasets are considered into two types of originally containing missing value and randomly auto-generating missing of complete dataset. Classification accuracy and NRMSE are used as parameters to judge the efficient of the proposed IG-SVMI algorithm. The experiments have shown that the proposed IG-SVMI algorithm can achieve better results than the benchmark approaches.

A sleepy driver is arguably much more dangerous on the road than the one who is speeding as he is a victim of microsleeps. Automotive researchers and manufacturers are trying to curb this problem with several technological solutions that will avert such a crisis. This article focuses on the detection of such micro sleep and drowsiness using neural network-based methodologies. Our previous work in this field involved using machine learning with multi-layer perceptron to detect the same. In this paper, accuracy was increased by utilizing facial landmarks which are detected by the camera and that is passed to a Convolutional Neural Network (CNN) to classify drowsiness. The achievement with this work is the capability to provide a lightweight alternative to heavier classification models with more than 88% for the category without glasses, more than 85% for the category night without glasses. On average, more than 83% of accuracy was achieved in all categories. Moreover, as for model size, complexity and storage, there is a marked reduction in the new proposed model in comparison to the benchmark model where the maximum size is 75 KB. The proposed CNN based model can be used to build a real-time driver drowsiness detection system for embedded systems and Android devices with high accuracy and ease of use.

In this paper, we propose a new approach to generate descriptive images from simple Arabic text of stories for children to build a flexible and rich multimedia repository. To make it successful, we need to transform the actors and objects and their semantic relationships from the text in compact and structured entities. We use the scene graph representation that considers objects together with their attributes and relationships. In fact, this representation has been proven to be useful across a variety of vision and language applications. We extend, thereby the notion of image generation to our approach as the task of constructing new scenes based on a set of toolboxes and a set of generated images for common textual descriptions. Our preliminary results show that the image repository is very efficient, saving then huge storage space. Besides, the composed scenes using the generated images could be customized providing a fair understanding of the main actors/objects of the stories as well as a coherent visual layout of the scenes.

Recently, Wireless Sensor/Actuator Networks (WSAN) became a major innovation in the modern world. This technology consists of two main parts, a group of sensors and actuators. Sensors act as an information gathering tool which collects data about the physical world like heat and humidity then communicates this data with other nodes on the network. Upon receiving the collected data, actuators can then perform actions that interact with the physical world according to the collected data. This type of interaction between devices and the physical world sparked new horizons in IoT and other smart devices. Therefore, some applications based on WSAN can automate specific tasks like controlling temperature of the room in a smart home. But as technology advances, WSAN based applications are implemented into more mission critical systems where QoS is very crucial and cannot tolerate packet loss or delays. Hence, QoS have become an important factor when discussing mission critical WSAN, since such systems can be life-threatening in many cases. Therefore, in this paper, we will provide a comprehensive overview of state-of-the-art approaches, protocols and applications that contributed in enhancing QoS in WSAN. Furthermore, we will provide our own comments on the shortcomings of the proposed solutions and discuss our take on how the discussed techniques can be further improved.