Skip to main content

Anomaly Monitoring

Proactive Anomaly Detection with Our IoT AI Model Deployment Platform​

Anomaly monitoring is a critical feature that helps you detect unusual patterns and behaviors in your IoT devices. By leveraging AI models, you can identify potential issues before they escalate, ensuring the reliability and efficiency of your operations.

How Anomaly Monitoring Works​

Harnessing AI for Real-Time Anomaly Detection​

  1. Data Collection: You add an anomaly monitoring subscription to your IoT devices, and future data points will be screened for anomalies in real-time.

  2. AI Model Analysis: When you choose to add a subscription to an IoT device, a tailor-made model is trained. The model is trained to identify deviations from normal patterns, signaling potential anomalies.

  3. Detection and Alerting: The anomaly monitoring results can be used to trigger downstream functionality, such as sending an email.

Setting Up Anomaly Monitoring​

Steps to Implement Effective Anomaly Monitoring​

Contact us to set up and configure anomaly monitoring for your IoT devices.

Effectiveness of Anomaly Monitoring​

Proven Performance Metrics​

Our anomaly detection system has been rigorously tested on a comprehensive suite of publicly available datasets with labeled anomalies. Here are the performance metrics:

  • Recall: Our system achieves a recall rate of 90%, which means it successfully detects 90 out of 100 actual anomalies. This high recall rate ensures that most anomalies are identified, minimizing the risk of unnoticed issues.

  • Precision: The precision rate of our system is 50%. This indicates that, on average, every second anomaly flagged by our system is a true anomaly. While there may be some false positives, this rate helps balance thorough detection with manageable alert volume.

These metrics represent average performance, and actual results may vary based on the specific characteristics of your data and use case.

Limitations of Anomaly Monitoring​

Numerical Data Requirement​

An anomaly monitoring subscription can be added to any IoT device data stream (temperature, light, CO2, RSSI, etc.) as long as the measurement is represented by numbers.