Behold! It is I, Monitron!

The names AWS uses for its products range from the excruciatingly literal (eg “Amazon Managed Streaming for Apache Kafka”) to the obscure, but cool-sounding (“Athena”, “Kinesis”). Sometimes they’re just confusing, sounding like they do something they don’t (eg “AppSync”, “Artifact”).

Every now and then though, AWS hit on names that communicate a sense of fun. “Snowcone” is a good recent example, and we can now add to that list “Monitron”, AWS’s complete IoT+ML solution for anomaly detection in industrial machinery.

The hardware even looks like it has a face!

Monitron means you no harm. For perspective, the smaller units are about the size of a matchbox.

In the Amazon Monitron Starter Kit you will find the Monitron Gateway plus 5 matchbox-sized Monitron Sensors. The sensors consist of a triaxial accelerometer, which measures vibration in 3D space, and a temperature sensor. Each sensor is powered by a battery that Amazon estimate will last for 3 years, based on a reporting frequency of once per hour. The reporting frequency cannot be changed, though I’m sure that’s a feature customers will request.

The sensors are designed to be glued permanently on to the equipment you wish to monitor, and from there will communicate their readings back to the gateway, which will forward them to AWS.

The gateway connects to the Internet via WiFi (that’s the only option) and the sensors connect to the base station using Bluetooth 5, which offers significantly better performance over Bluetooth 4 — a four-fold increase in range (up to 40m indoors) and a significant increase in data throughput.

Once your data is in AWS, it can be viewed using the Monitron app, which provides a historical view of sensor data and well as a way to access information about any alarms raised.

Monitron detects abnormal machine operating states by analyzing vibration and temperature signals using the ISO 20816 standards for vibration, as well as machine learning-enabled models. Customers can improve the accuracy of the ML anomaly detection over time by providing feedback to Monitron.

IoT or ML? So is this product an IoT product or a Machine Learning product?

In reality Monitron is the logical combination of both technological streams, with both IoT and ML technologies crucial to making a functional product.

That said, in the AWS product catalogue Monitron sits under Machine Learning. This perhaps tells us that AWS is less interested in positioning itself as a leader in IoT than it is in emphasising the advances it has made in bringing ML to the problem space of industrial equipment monitoring.

This article was written by Guy Morton, Solutions Architect at AC3 and AWS APN Ambassador.