Detecting Application-Level Associations Between IoT Devices Using a Modified Apriori Algorithm

Abstract

Internet of Things (IoT) for home systems enables new functionalities and results in significant conveniences. However, their reliance on a stable, continuous Internet connectivity reduces their overall reliability. Losing connectivity to the Internet, for many of these devices, translates to the cessation of even the most basic of functionalities (e.g., being able to turn on the light, even from within the house). A possible solution to this problem is to shift some functionalities done by cloud-based servers to the edge (e.g. the home router), but doing so conveniently would necessitate the ability to dynamically identify pairs of IoT devices (usually sensor-actuator pairs) that communicate (their associations), and the rewriting/rerouting of packets or messages between those devices. The first problem is also known as the association detection problem, and is where this paper makes a contribution. We describe a solution to the association detection problem using a modified Apriori algorithm and a method to create its input from network traffic, and then revise the solution to respond to fluctuating network conditions. The final design accurately discovers sensor-actuator pairs using a simple approach with low computational complexity, and with only the hardware addresses of monitored IoT devices as its starting knowledge.