IoT-NM: Monitoring as a Service for IoT Networks

The rapid proliferation of IoT devices in the past years has led to a growing demand -and offer- for monitoring solutions. Focusing on solutions that provide indirect monitoring through the network, i.e., that analyze traffic data to monitor devices and application, we observe that measurements and analytics are often generic and arbitrary, and rarely tailored to the specifics of the devices and applications they monitor. The difficulty of creating ad-hoc solutions is not only due to the diversity of IoT devices and applications, but also to the difficulty of deploying such features: acquiring the right data for a new monitoring need requires manually configuring the different network equipment to extract this data. Without this, monitoring is constrained to a fixed set of input features. In this paper, we present a novel platform architecture model for network traffic monitoring that removes this limitation. It allows network administrators to dynamically deploy new monitoring functions with no manual intervention. On demand, it automatically installs arbitrary data acquisition methods in network equipment, instantiates monitoring functions in dedicated a compute node, and transmits the extracted data between the two. This way, monitoring functions can be implemented by data analysts with limited networking knowledge, and deployed at no cost by network administrators. In addition to the above model, we provide a functional proof of concept: IoT-NM. We provide empirical results showing that our solution can manage high network load with little to no impact on routing performance.