The AGT IoT Analytics apply a large variety of analytics engines to IoT data sources in order to understand the behavior of people and things. Purpose-built to handle the complexities of IoT data, these analytics can be applied to all types of sources such as video, audio, location, data from wearable sensors and more.
Analytics Hierarchy for More Actionable Decision Support
The AGT IoT Analytics platform is not just a collection of analytics engines and modules; it operates as a hierarchy of increasing semantic content and meaning:
- Foundational analytics deal with sensor level data such as advanced signal processing, feature extraction, advanced data filtering, multi-sensor fusion, noise reduction, data normalization and cleansing. These are essential analytics for IoT as they handle sensor data directly and “prepare” them for the next tier or layer in the hierarchy.
- Insight analytics extract activities, patterns and behaviors by using data fusion, classification, clustering, activity recognition, time series analysis, object recognition and object tracking and other techniques. The outputs of these modules deliver actual new insights from the data.
- Action analytics serve as decision support components and include predictive models (both physics based and data driven), what-if simulation, optimization and more. These support decision makers as they determine an informed course of action.
Powerful Engines with Domain-Specific Applications
The AGT IoT Analytics consist of powerful engines that generate insights common to many domains and industries. In addition, they support domain-specific analytics created both by AGT and third parties. Developers can pick and choose solution-relevant analytics from each tier or layer and to connect them together so that very complex behavior can be inferred from relatively simple individual sensor readings.
For example, AGT is developing analytics for sports and entertainment. These analytics include:
- Motion and physiological parameter analytics (applied to the wearable sensors)
- Behavior analytics which extract context from the data (such as game, general mood/excitement level of audience, etc.) and from individuals or groups (for example, based on specific user profiles or on current activity) by fusing real-time sensor data and historical data.
- A True Fan Index analytic model that incorporates behavior, game context, activity and mood at any moment in the game.
- Content creation analytics, which generate insights from data (real time/sensor, historical) that could be of interest to the target audience (e.g., electronic Gameday programs, game statistics, replays, etc.)
IoT data is very different from transaction-oriented business data that support most decisions. Coming from millions of sensors and sensor-enabled devices, IoT data is more dynamic, heterogeneous, unstructured and unprocessed than typical business data. It is generated at huge scale and is time-sensitive; IoT analytics may only be useful for a short period such as the duration of an event. AGT delivers the sophisticated, IoT-specific analytics that IoT technologies demand.