AWS IoT Events
An event-driven service that detects patterns in IoT sensor data and automatically executes alerts and actions based on defined conditions
Overview
AWS IoT Events is a service that monitors telemetry data sent from IoT devices and applications, automatically triggering alerts and actions when defined conditions or patterns are matched. You define detector models as state machines to visually model device state transitions. For example, you can build complex logic without code - such as raising an alarm when a temperature sensor exceeds a threshold for 5 minutes, then sending an emergency shutdown command if it continues for 10 minutes. It integrates with IoT Core, IoT SiteWise, SNS, Lambda, and more to automate workflows from detection to response.
State Machine Design with Detector Models
The core of IoT Events is the detector model, which defines device states as a state machine. Each state has OnEnter, OnInput, and OnExit event handlers, with state transition conditions written as logical expressions. For example, in factory temperature monitoring, you define three states - Normal, Warning, and Critical - with the model transitioning to Warning when temperature exceeds 60 degrees and to Critical above 80 degrees. Detector models operate independently per instance, so a single model can individually monitor thousands of sensors. Instances are isolated by key value (e.g., device ID), with each device maintaining independent state. Timer functionality enables delayed logic that executes actions only when a specific state persists for a set duration, preventing false alarms from temporary spikes. Model definitions are managed in JSON format, making version control and test environment validation straightforward.
Simplified Monitoring with Alarm Models
While detector models build complex state machines, alarm models are a lightweight mechanism specialized for simple threshold monitoring. They set thresholds against a single input value and transition to an alarm state when exceeded - a simple model that is easy to configure and suitable for PoCs and small-scale monitoring. Alarm models integrate directly with IoT SiteWise asset properties, making them ideal for industrial IoT sensor value monitoring. Alarm severity levels (Critical, Major, Minor, Low) can be configured, enabling routing to different notification targets based on level. Acknowledge and Reset workflows are built in, allowing operators to record manual alarm acknowledgment. In practice, a staged approach is effective: start with alarm models for basic threshold monitoring, then migrate to detector models when compound conditions become necessary.
Integration with Other IoT Services and Operational Design
IoT Events functions as the event processing layer within the AWS IoT ecosystem, working with data collection (IoT Core), storage (IoT Analytics), and visualization (IoT SiteWise) to form a complete IoT pipeline. The most common configuration receives input directly from IoT Core's rule engine, evaluating messages published to MQTT topics in real time. Action output destinations include SNS (email/SMS notifications), Lambda (custom logic execution), IoT Core (sending commands to devices), and SQS (asynchronous processing queues). An operational consideration is that detector model evaluation rates have per-second message limits, and Service Quotas increase requests are needed when large numbers of devices send data at high frequency. For costs, pricing is based on message evaluation count, so a design that filters unnecessary data at the IoT Core rule engine before sending to IoT Events is cost-efficient.