Context‐aware search optimization framework on the internet of things

Abstract

The resource discovery on IoT paradigm requires to be efficient with respect to modeling, storage, processing, and validation of the gathered data. These requirements face challenges like interoperability, heterogeneity, etc, with respect to exponentially growing interconnected resources across distinct application domains and drastically changing search metrics. It leads resource discovery to emerge as a non-linear constrained-specific problem that need to be linearized for its optimization with reduced complexity. Keeping the perspective, a context-aware search optimization framework on the internet of things is introduced, which targets knowledge presentation through schema, discovery via a 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 performs resource discovery by finding the suited matches as a search set from a search-space. The search set is further evaluated via the iterative gradient descent algorithm for optimization through the usage of iterative and convergence properties of the gradient descent. The search efficiency is tested using various objective functions and resources on MATLAB and is compared with Newton and Quasi-Newton methods. The obtained results depict the efficiency of the algorithm graphically with reference to the searching time, such as validate the system performance. KEYWORDS electronic toll plaza, Internet of Things, iterative gradient descent, multi-modal search, resource discovery, search optimization INTRODUCTIONThe Internet of Things (IoT) paradigm is expected to have almost 212 billion resources, as reported by Gartner, 1 which require to be identifiable to communicate and to interact with each other. These resources are defined with context, ie, a set of inter-related events that share logical and timing relation among each other to provide relevant information such as location and type to characterize them. 2 The events belong to various stages such as data collection, processing, modeling, storage, reasoning, and validation. Each stage with respect to context has their own needs, models, and techniques, which are to be studied thoroughly.For the Context Collection, the IoT platform requires various type of sensors that are compatible with different applications. For example, a temperature tracking system is designed that uses a client-server architecture having a Raspberry Pi and system-on-a-chip (SoC) device. It senses the temperature and forwards it to the end user using an Android mobile application. 3 Another novel service, Social Internet of Things framework (SIoT)is proposed to enhance intelligence, for dynamic IoT service discovery in smart spaces, and is based on a cognitive reasoning approach. 4Moving to the Context Modeling, there exist various techniques such as key-value, mark-up scheme, graphical, logic-based, and ontology-based to model the gathered data. The key-value-based modeling technique creates text or binary file a...