Publications

Abstract : Embedded devices in smart homes have become increasingly vulnerable to numerous security and privacy threats. Over the past few years, devices such as Smart Cameras have been used to launch Distributed Denial of Service (DDoS) attacks wherein attackers exploit weakly configured IoT devices and inject malicious code after discovering their credentials. Conventional methods used to identify anomalies cannot be applied due to the resource-constrained and heterogeneous nature of the IoT devices. To address this, a novel approach that leverages the power consumption of these devices needs to be devised. Towards this goal, a smart home scenario is simulated using Smart Cameras and brute-force, and DDoS attacks were launched to capture variations in power profile. Further, we develop machine learning models to detect anomalies based on the power consumption traces. Our proposed approach achieves an accuracy of 94.04% towards detecting the presence of anomalies. Our analysis reveals that power consumption is a promising factor that can be used to detect anomalies in IoT based smart homes.