Author(s): Diallo, Ousmane
Date: 2014
Persistent ID: http://hdl.handle.net/10400.6/3301
Origin: uBibliorum
Project/scholarship: info:eu-repo/grantAgreement/FCT/6817 - DCRRNI ID/PEst-OE%2FEEI%2FLA0008%2F2011/PT; info:eu-repo/grantAgreement/FCT/6817 - DCRRNI ID/PEst-OE%2FEEI%2FLA0008%2F2013/PT;
Subject(s): Bases de dados; Sistemas de bases de dados; Gestão de bases de dados; Bases de dados em tempo real; Rede de sensores sem fios; Rede de sensores sem fios - Gestão de dados
Technological advances in recent years have allowed the maturity of Wireless Sensor Networks (WSNs), which aim at performing environmental monitoring and data collection. This sort of network is composed of hundreds, thousands or probably even millions of tiny smart computers known as wireless sensor nodes, which may be battery powered, equipped with sensors, a radio transceiver, a Central Processing Unit (CPU) and some memory. However due to the small size and the requirements of low-cost nodes, these sensor node resources such as processing power, storage and especially energy are very limited. Once the sensors perform their measurements from the environment, the problem of data storing and querying arises. In fact, the sensors have restricted storage capacity and the on-going interaction between sensors and environment results huge amounts of data. Techniques for data storage and query in WSN can be based on either external storage or local storage. The external storage, called warehousing approach, is a centralized system on which the data gathered by the sensors are periodically sent to a central database server where user queries are processed. The local storage, in the other hand called distributed approach, exploits the capabilities of sensors calculation and the sensors act as local databases. The data is stored in a central database server and in the devices themselves, enabling one to query both. The WSNs are used in a wide variety of applications, which may perform certain operations on collected sensor data. However, for certain applications, such as real-time applications, the sensor data must closely reflect the current state of the targeted environment. However, the environment changes constantly and the data is collected in discreet moments of time. As such, the collected data has a temporal validity, and as time advances, it becomes less accurate, until it does not reflect the state of the environment any longer. Thus, these applications must query and analyze the data in a bounded time in order to make decisions and to react efficiently, such as industrial automation, aviation, sensors network, and so on. In this context, the design of efficient real-time data management solutions is necessary to deal with both time constraints and energy consumption. This thesis studies the real-time data management techniques for WSNs. It particularly it focuses on the study of the challenges in handling real-time data storage and query for WSNs and on the efficient real-time data management solutions for WSNs. First, the main specifications of real-time data management are identified and the available real-time data management solutions for WSNs in the literature are presented. Secondly, in order to provide an energy-efficient real-time data management solution, the techniques used to manage data and queries in WSNs based on the distributed paradigm are deeply studied. In fact, many research works argue that the distributed approach is the most energy-efficient way of managing data and queries in WSNs, instead of performing the warehousing. In addition, this approach can provide quasi real-time query processing because the most current data will be retrieved from the network. Thirdly, based on these two studies and considering the complexity of developing, testing, and debugging this kind of complex system, a model for a simulation framework of the real-time databases management on WSN that uses a distributed approach and its implementation are proposed. This will help to explore various solutions of real-time database techniques on WSNs before deployment for economizing money and time. Moreover, one may improve the proposed model by adding the simulation of protocols or place part of this simulator on another available simulator. For validating the model, a case study considering real-time constraints as well as energy constraints is discussed. Fourth, a new architecture that combines statistical modeling techniques with the distributed approach and a query processing algorithm to optimize the real-time user query processing are proposed. This combination allows performing a query processing algorithm based on admission control that uses the error tolerance and the probabilistic confidence interval as admission parameters. The experiments based on real world data sets as well as synthetic data sets demonstrate that the proposed solution optimizes the real-time query processing to save more energy while meeting low latency.
Fundação para a Ciência e Tecnologia
This work has been partially supported by the Instituto de Telecomunicações, Next Generation Networks and Applications Group (NetGNA), Portugal, by National Funding from the FCT – Fundação para a Ciência e Tecnologia through the PEst- OE/EEI/LA0008/2011 and Pest-OE/EEI/LA0008/2013 projects, and by the AAL4ALL (Ambient Assisted Living for All), project co-financed by the European Community Fund FEDER through COMPETE – Programa Operacional Factores de Competitividade.
