Smart Home Crawler: Towards a framework for semi-automatic IoT sensor integration

TL;DR

This paper introduces the Smart Home Crawler Framework, which simplifies and accelerates the integration of heterogeneous IoT sensors in smart homes using semantic abstraction and machine learning.

Contribution

It presents a novel framework that provides semantic abstraction and machine learning-based device linking to streamline sensor integration in smart homes.

Findings

Prototype demonstrated at ICT 2018

Provides semantic abstraction for heterogeneous sensors

Uses machine learning for device linking

Abstract

Sensor deployments in Smart Homes have long reached commercial relevance for applications such as home automation, home safety or energy consumption awareness and reduction. Nevertheless, due to the heterogeneity of sensor devices and gateways, data integration is still a costly and timeconsuming process. In this paper we propose the Smart Home Crawler Framework that (1) provides a common semantic abstraction from the underlying sensor and gateway technologies, and (2) accelerates the integration of new devices by applying machine learning techniques for linking discovered devices to a semantic data model. We present a first prototype that was demonstrated at ICT 2018. The prototype was built as a domainspecific crawling component for IoTCrawler, a secure and privacy-preserving search engine for the Internet of Things.