الفهرس | Only 14 pages are availabe for public view |
Abstract The basic idea of Internet of things concept is the pervasive presence around us of a variety of things or objects. Internet of things major concern not only the huge number of devices connected to the network but the heterogeneity of these devices as it will surround users with more perceptive and intelligent information. Rapid development of sensor networks and radio frequency identification (RFID) has led to a tremendous number of devices connected to the Internet sharing their data over time with each other. These sensors data are often processed and transformed to context. Context is any information that can be used to characterize the situation of an entity. Inaccurate or noisy devices often produce inconsistent context data. Several techniques have been proposed to solve inconsistency in contexts. This Thesis Presents a new approach based on machine learning algorithms to solve context inconsistencies, first we build a simulation environment of data ware house to generate context dataset, and then we apply random errors on the context dataset. Second we apply existing resolution methods on inconsistent contexts and select best resolution methods. After building a resolution data set of best resolution methods alongside their situations we apply our machine learning algorithm on the resolution data set and build a model that will be used in future prediction of resolution method. Our purpose of this study is to improve inconsistency resolution accuracy and prevent removing un correct contexts inconsistencies, also find most of undetected inconsistencies. The organization of the thesis is as follows: Chapter 1 starts with a history of internet and the new paradigm internet of things changing our perceptive to data and devices connected to the internet. Then a brief introduction was introduced to middleware and context awareness functionality. Moreover data mining was discussed with relation to internet of things. Finally research objectives are discussed, and Finally, the thesis organization is given Chapter 2 briefs background of context Inconsistency and resolution strategies Chapter 3 gives a detailed description of the proposed work. Chapter 4 gives a detailed description of the proposed method based on machine learning modeling. Chapter 5 discusses the Experimental results of the proposed methods and comparative study. First a simulator was built to gather inconsistency resolution data set. Then the proposed machine learning approach was applied on this data set. Finally gives conclusions from proposed methods and future works. |