COMPONENT-BASED ARCHITECTURE OF IOT SEARCH ENGINE

Abstract

When the number of IoT devices is growing exponentially, the demand for the search engine of "things" is as big as for the search engine of web pages in the 1990s. Several pilot projects are already running, they differ by their internal architecture and the algorithms for discovery, search, and representation. Early results are as promising as unsettling because they expose the challenges and treats. When web pages are uniform in some extent, the IoT network is highly heterogeneous and dynamic. We argue that because of the nature of IoT, we cannot just reuse the principles of websites search engines. Moreover, the usage of only one algorithm for device discovery or its property search can become inefficient soon. The quality of IoT search can be improved either by using AI and machine learning techniques or dynamically reuse the elements of 2 or more IoT search engines. This paper presents the software architecture of IoT search engines, based on software components, which enable the combination of the advantages of different search engines and their algorithms.