Please use this identifier to cite or link to this item:
Title: Resource Inquisition and Optimal Calibration on Internet-of-Things
Authors: Bharti, Monika
Supervisor: Saxena, Sharad
Kumar, Rajesh
Keywords: IoT;Resource Discovery;Searching;Resource Selection
Issue Date: 3-Feb-2020
Abstract: Internet-of-Things (IoT) is an emerging and widely applied research area that envision integration of physical world into digital. For the same, it requires automation of the interconnected resources such that they can sense, process and interpret via Internet-connected infrastructures. A resource is either software or hardware with fundamental characteristics such as physical embodiment, unique identifier, offered service, location, information, operating system and mode of communication. To facilitate communication and computation among the resources, it is required to discover them across distributed systems having distinct communication capabilities and intelligence. It implies that discovery is possible with heterogeneous views like human operators, application softwares and autonomous smart objects. Such views face challenges because resources are generating huge volume of data coupled with mobility and dynamicity. Hence, it would slow down the process of discovery because of issues like heterogeneity, interoperability, lack of standardization, and periodic evaluations. Moreover, these issues limit the resource discovery on IoT paradigm leading to challenges such as to provide answers to knowledge-based queries, to develop models to exchange applications' context, to bridge the gap between device level and across applications, indexing and splitting of multiple parameters, and sophisticated techniques for managing metadata. Thus, it concludes resource discovery as a fundamental challenge for the realization of IoT vision and has gained the researcher's interest all over the world. To address the challenge, resources are required to undergo both retrieval and ranking processes. In short, resource discovery needs organization and analysis of the gathered complex descriptions as data. For the purpose, it requires techniques or algorithms that cover basic fundamental search principles for indexing, clustering, knowledge representation, and content being searched. This will ease in searching and would cover a multitude of functionality and dimensionality. Moreover, due to factors such as noise, limited memory, opportunistic presence of the resources, availability and data integrity, IoT needs to consider various parameters with respect to discovery. The parameters are to automate and access, several dimensions related to query and to understand the search evaluation metrics. In this thesis, a novel clustering technique, namely, Iterative K-means Clustering Algorithm (IKm-CA) and three frameworks, i.e., Context-Aware Search Optimization Framework on Internet-of-Things (\CASOF-IoT), Intelligent Resource Inquisition Framework on Internet-of-Things (IRIF-IoT) and Middleware Approach for Reliable Resource Selection on Internet-of-Things (MARRS-IoT) for resource discovery and selection are proposed. The IKm-CA targets concrete cluster formation using similarity coefficients of vector space model and performs efficient search against matching criteria. It mitigates the problems of selecting erroneous or empty data points to clusters, reduces consumption time, noise factor and the number of clusters that are manually input to the system. Though, the technique eliminates the heterogeneity in data challenge to resource discovery and selection on IoT, yet it has not accounted the issues such as context of the gathered data, scalability and search metrics on IoT. These issues if not resolved, would lead resource discovery and selection to emerge as a non-linear constrained specific problems. Therefore, it is required to optimize them to ease the inter-communication and interaction. For the purpose, CASOF-IoT is introduced that targets knowledge presentation through schema, discovery via multi-modal search algorithm and its optimization through an Iterative Gradient Descent algorithm. The multi-modal search algorithm through keywords, value or spatial-temporal indices perform resource discovery by finding the suited matches as search set from search-space. Though, it performs efficient resource discovery yet it could not address interoperability challenge at various levels like semantic, radio access and context with respect to resource discovery for the automation and authentication of the infrastructure. Hence, another framework, IRIF-IoT is suggested that link resources through usage of semantic description and ontology, their discovery with Semantic Matchmaking Engine using Bipartite Graph (SMEBG) and to access information via web terminal for users. The framework has enhanced the system performance significantly but has not considered trust issue. This issue arises on a global network infrastructure like the IoT due to continuously changing mobility patterns, heterogeneity, interoperability and scalability on the network that imposes restriction to process an optimal decision. Keeping the trust perspective, MARRS-IoT is presented that helps to achieve maximum reliable resource for communication that maintain data integrity via trust evaluations and improve system performance. It dynamically processes information locally as well as globally with less time consumption and minimizes complex searches.
Appears in Collections:Doctoral Theses@CSED

Files in This Item:
File Description SizeFormat 
Resource Inquisition and Optimal Calibration on IoT.pdf2.35 MBAdobe PDFView/Open

Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.