Big Data Anomaly Detection Engine
Learn what is normal, and detect abnormal events in big data in real time
The increasing amount and variety of sensors deployed across various IoT applications presents us with a significant challenge: How can we process, quickly and accurately, exceptionally large volumes of disparate data from different data types?
AGT has developed a flexible analytics engine that facilitates the learning of normal behaviors in complex, sensor-rich environments in order to detect abnormal events without the need for rules configuration. This enables efficient system operation through “management by exception”.
Insights generated from the learned model of normal behavior are also used to optimize operational processes.
Reduce number of alerts. Determine root causes.
The traditional approach to the automatic monitoring of complex environments is to use a rules-based engine, which triggers alerts if and when certain thresholds, which are configured manually, are exceeded. But how do current systems learn normal from abnormal events? They don’t, and simply adding rules is not a scalable approach. These systems lack data fusion and learning capabilities, and therefore fail to get at the root cause of events.
AGT’s universal and flexible big IoT data analytics engine works with large amounts of data, automatically learning the normal behaviors of complex systems and detecting abnormal events. Our analytics engine can be applied across a range of different use cases and data types, and reduces the number of manual interventions necessary to monitor and manage complex systems and optimize operational processes.
Our big data analytics engine is deployed in installations that contain thousands of sensors, including, for example, in the traffic domain (incident management) and energy grid operations. Our engine is additionally used to analyze social media data, energy consumption patterns, video data and security logs.