The primary objective and contribution of this research is the development and design of an artificial intelligence system that will diagnose Chronic Obstructive Pulmonary Disease (COPD) utilizing only the heart signal (ECG) of the patient. In contrast to the traditional way of diagnosing COPD, which requires spirometer tests and a laborious workup in a hospital setting, the proposed system uses the classification capabilities of deep transfer learning and the patient’s heart signal, which provides COPD signs in itself and can be received from any modern smart device. The motivation of this research is that it introduces the first research on automated COPD diagnosis using deep learning and utilizes the first annotated dataset in this field. Since the disease progresses slowly and conceals itself until the final stage, hospital visits for diagnosis are uncommon. Hence, the medical goal of this research is to detect COPD using a simple heart signal before it becomes incurable. Deep transfer learning frameworks, which were previously trained on a general image data set, are transferred to carry out an automatic diagnosis of COPD by classifying patients’ electrocardiogram (ECG) signal equivalents, which are produced by signal-to-image transform techniques. Xception, VGG-19, InceptionResNetV2, DenseNet-121, and "trained-from-scratch" convolutional neural network architectures have been investigated for the detection of COPD, and it is demonstrated that they are able to obtain high performance rates in classifying nearly 33.000 instances using diverse training strategies. The highest classification rate was obtained by the Xception model at 99%. This research shows that the newly introduced COPD detection approach is effective, easily applicable, and eliminates the burden of considerable effort in a hospital. It could also be put into practice and serve as a diagnostic aid for chest disease experts by providing a deeper and faster interpretation of ECG signals. Using the knowledge gained while identifying COPD from ECG signals may aid in the early diagnosis of future diseases for which little data is currently available.
Recent advancements in the automotive industry have led to the design of smart vehicles with high capacity resources for communication, sensing, processing, and storage of data. In the near future, it is envisaged that these resources will be harnessed and utilized to provide cloud services such as storage as a service, computation as a service, and sensing as a service. This paradigm of computing termed vehicular cloud, presents a lot of opportunities for the deployment of delay-sensitive applications in vehicular environment. However, high mobility of vehicles and rapidly changing topology of vehicular networks introduce new challenges such as instability of resources which make the management of resources in vehicular cloud complicated. Therefore, vehicular cloud computing require robust and mobility-aware resource management solutions to be designed. In order to achieve this goal, researchers and engineers need to be abreast with the techniques of resource management in vehicular cloud. This paper presents a detailed survey of resource management tasks in vehicular clouds. A thorough introduction to vehicular cloud is presented initially. Then we identify and examine all the resource management tasks that are carried out in vehicular cloud by classifying them into three different phases: pre-resource assignment phase, resource-assignment phase, and post resource assignment phase. Proposed solutions to resource management challenges in literature at each phase are reviewed in detail. Unlike the existing surveys on vehicular clouds, this study covers all aspects of resource management such as resource brokering, resource demand prediction, resource allocation, and resource scheduling. Following the extensive survey, we also propose frameworks for resource monitoring and resource brokering for vehicular cloud. Finally, open issues and research directions for vehicular cloud resource management are presented.
11 September 2012 in Applied Computational Intelligence and Soft Computing
Navigation planning can be considered as a combination of searching and executing the most convenient flight path from an initial waypoint to a destination waypoint. Generally the aim is to follow the flight path, which provides minimum fuel consumption for the air vehicle. For dynamic environments, constraints change dynamically during flight. This is a special case of dynamic path planning. As the main concern of this paper is flight planning, the conditions and objectives that are most probable to be used in navigation problem are considered. In this paper, the genetic algorithm solution of the dynamic flight planning problem is explained. The evolutionary dynamic navigation planning algorithm is developed for compensating the existing deficiencies of the other approaches. The existing fully dynamic algorithms process unit changes to topology one modification at a time, but when there are several such operations occurring in the environment simultaneously, the algorithms are quite inefficient. The proposed algorithm may respond to the concurrent constraint updates in a shorter time for dynamic environment. The most secure navigation of the air vehicle is planned and executed so that the fuel consumption is minimum.
The demand for computational resources in vehicular environments has increased due to the deployment of numerous intelligent transportation systems in the last decade. The federated vehicular cloud, a variant of vehicular cloud computing where resources embedded in individual vehicles are organized as a single unit to provide cloud services, is considered as an emerging alternative to the conventional cloud platforms for the execution of computationally intensive and delay-sensitive applications. However, the federated vehicular cloud is beset with a capacity-constrained communication channel and limited resource capacity in individual vehicles, leading to challenges in data and resource management. To address these challenges, we propose UniDRM, a unified data and resource management framework for the federated vehicular cloud. The UniDRM organizes vehicles on the road into clusters based on their mobility and resource characteristics, such as resource cost, resource credibility level, resource type, and available resource capacity. The data of computationally intensive tasks are then partitioned using our proposed analytical model and assigned to individual vehicles in the cluster for parallel execution. Three data partitioning and scheduling schemes: time-aware, cost-aware, and reliability-aware, are proposed in this study to execute time-critical tasks, low-cost tasks, and high-security tasks, respectively. Through realistic simulations, a comparative analysis of the proposed partitioning and scheduling schemes is presented.
As the energy becomes the most valuable resource in the wireless sensor networks, to prolong the lifetime of the network, energy-aware protocols are proposed which use the energy minimization techniques and multi-hop paths. However, multi-hop paths and complex calculations at nodes introduce delay and processing overhead. Providing energy-efficiency becomes a scalability problem. Scalability of a routing protocol for wireless sensor networks is mainly affected by the topological changes and by the number of sensor nodes in the network. In large scale networks with a large number of sensor nodes, multiple-sinks (gateways) should be used to provide scalability. In this paper, a novel stateless energy-efficient routing technique, stateless weight routing with multiple sinks (MS-SWR), is introduced for wireless sensor networks. MS-SWR requires no local/global topological information. Property of being stateless avoids the communication overhead and reduces the calculations to minimum. In MS-SWR, reliability is provided by using multiple paths which also reduces the delay to minimum for the time-critical real-time traffic. The performance evaluation of the proposed technique shows that MS-SWR is scalable for large scale wireless sensor networks with multiple sinks.
Integrating mobile devices into Grid technologies and server applications can give ability to command power of supercomputers with a mobile device on one hand and can allow big applications to reach important data anywhere, anytime, on the other. IMOGA is planned to be an example to gather and share data that can be collected by ubiquitous mobile devices which can employ different kind of sensors such as Global Positioning System (GPS), temperature, health monitoring and pollution. In this project location and speed information that is produced by GPS enabled mobile devices such as mobile phones, is used. The developed client application running on mobile devices located in vehicles, such as the mobile phone of the driver, sends location and speed information to the server application in short time intervals via GPRS in the forms of Extended Mark-up Language (XML) like messages. The developed server application, which is preloaded with the highway coordinates via files in Geographic Data Files (GDF) format, locates the street that the vehicle is moving along and the received speed information is recorded along with a timestamp. A display application has also been implemented to calculate average of speeds that any vehicle may have at that very moment and post it on the Internet and WAP. If there is no actual data, i.e. there is no vehicle moving on a specific street, statistical data is utilised to produce such information.