Posted On: Sep 14, 2022

We are excited to introduce two new enhancements to Amazon IoT Device Defender ML Detect, Custom Metrics and Dimensions support. ML Detect now supports monitoring of custom metrics allowing you to evaluate operational health parameters that are unique to your fleet. Also, in addition to setting static alarms manually with Rules Detect, you can now use machine learning to automatically learn your fleet's expected behaviors on custom metrics. Further, with the new Dimensions filter support for ML Detect, you can define attributes to evaluate more precise metrics in your ML security profile.

In this release, custom metrics on ML Detect supports the number-type, such as device’s connection signal strength or percentage of CPU usage, while the dimensions feature provides support for MQTT-topic-filter on four device cloud-side metrics (number of messages received, message byte size, number of messages sent, and number of authorization failures). To get started with monitoring custom metrics, you can setup a device side agent with our sample agent in Python or use Amazon IoT Device SDK in C++. The custom metrics and dimension capabilities are available in all Amazon Web Services regions where Amazon IoT Device Defender ML Detect is available, including the Amazon Web Services China (Beijing) Region, operated by Sinnet and the Amazon Web Services China (Ningxia) Region, operated by NWCD. For more information, refer to Amazon IoT Device Defender documentation.