Artificial Intelligence for Cybersecurity in Smart Homes: An In-Depth Exploration of AI Techniques and Research Opportunities

Authors:
Preprint
Publisher website
Google Scholar
Abstract
The proliferation of connected smart homes presents a conundrum to society - security versus convenience. As organisations invest heavily in developing products and services to automate domestic tasks, consumers are increasingly being lured towards unsecured devices that often offer enticing and immediate benefits. These devices range from security cameras to climate control systems, intelligent lighting, and more. The evolution of artificial intelligence offers a potential answer to identifying cyberattacks and vulnerabilities in smart home environments. This study intends to establish the existing architectures, frameworks, techniques, data sources and environments to which the current state of the art has contributed. To that extent, a systematic literature review (SLR) process has been carried out across four journal databases - SCOPUS, ACM, IEEE Xplore and Science Direct. 390 papers were identified in the selection stage, 131 progressed to the extraction stage, where 90 were finally accepted. The findings show that significant efforts have been made to test the detection accuracy of various artificial intelligence techniques using standard open datasets of known attacks, or simple datasets from prototype testbeds. Attempts were also made to identify zero-day vulnerabilities through anomaly detection techniques, although the breadth and depth of the datasets employed are limited. The summary of this review reveals the gaps across datasets, techniques, and missing external data context from sources, such as the National Vulnerability Database (NVD), amongst others. These present promising opportunities for research in the field.

This publication has 50 references indexed in Scilit: