Traffic-Level In-Depth Profiling of Smart Home Events, Devices, and Cross-Device Interactions

The widespread use of Smart Home devices has attracted significant research interest in understanding their behavior within home networks, notably to enhance their security.
Unlike general-purpose computers, these devices exhibit relatively simple and bounded network activity patterns.
The IETF's MUD standard was a leap towards their profiling, providing a format to express those patterns.
However, it lacks expressiveness in modeling communications taking place in an actual Smart Home network, mainly due to its inability to express dependencies between subsequent traffic patterns,
something usual in Smart Homes as part of home automation routines.
We propose a new format, inspired by MUD, overcoming this limitation.
Afterwards, we tackle the problem of the automated generation of MUD-like traffic profiles.
While various works have studied this, they have always considered normal network conditions, overlooking potential hidden patterns that emerge under non-standard conditions. Discovering the latter is crucial to provide an exhaustive profile of the device's traffic.
We propose a mostly automated framework able to extract such patterns in a more exhaustive way than existing works, steadily uncovering more traffic flows, mostly hidden ones.
Additionally, we leverage the latter as a measure of the devices' robustness, and provide guidelines towards enhancing it.
We envision such guidelines could be used in the future to label devices put on the market, in order to provide the customer with a more enlightened choice.