ATLAS’ vision is to guarantee the US Government (USG) access to space through the ability to leverage a global hybrid network of USG and commercial antennas using a highly innovative machine learning scheduler of schedulers. Built on ATLAS’ commercially available product, the prototype for this effort is a data analytics engine to support the ATLAS Cognitive Constellation Management Scheduler. Leveraging the power of data science and machine learning to facilitate decision making in the scheduler, the unified analytics engine prototype will transform raw data from billions of data points to increase network performance measured by the results of the taskings enhancing reliability and resiliency.
The main goal of this project is to leverage streaming and real-time data, in addition to billions of historical data points and state of the art data analytics to create detailed performance metrics per site, per customer, per satellite and even per hardware unit, per software patch, and per configuration. The ability to slice and dice per site, customer, etc., tied in with modern approaches to outlier identification, anomaly detection, and time series analysis can empower a suite of insights. This includes use cases such as hardware failure prediction for maintenance optimization and downlink throughput maximization.
The proposed solution will improve the efficiency and resiliency of the satellite communications infrastructure by enabling NASA to smartly leverage their networks and a global hybrid network of satellite communications ground stations. This product will apply data science and machine learning to support decision making in the cognitive scheduler to facilitate RF antenna utilization and reliability.
Commercial applications are very similar to USG applications in that the demand for data from space is increasing, ground stations are limited, and adding capacity is costly. Maximizing the use of existing infrastructure offers significant cost savings and effective use of underutilized assets. Therefore, the non-NASA applications of this technology are essentially the same.