It’s been tremendously exciting to see how rapidly the DIFI standard has become so widely adopted for ground segment operations across earth observation (EO), remote sensing (RS) and TT&C applications. Its presence in satcom, milsatcom and mission download networks continues to expand and has become foundational across Ground-as-a-Service (GaaS) implementations.
Now we’re seeing expansion in another critical mission set, RF analysis and SDA applications, where DIFI is contributing to the creation of AI and machine learning (ML) capabilities.
Using DIFI, raw signals can be fed directly into private or public clouds very near the point of signal capture where they can be used to train AI and machine language analysis for signals intelligence. This is particularly cost effective for use with public cloud infrastructure that has global reach. Public clouds typically mainly charge for compute and egress of data, less so for ingress fees or the networking of the process results.
It’s broadly acknowledged that training is the greatest hurdle to effective AI adoption. Using a common data standard to ingest the data brings scale benefits to AI training and performing that training in the cloud while limiting data egress brings economic scale.
Once created, these AI/ML algorithms can then be deployed at scale across an entire set of deployed sensors in local public and private clouds. Relevant results and conclusions can then be output to local and centralized operations teams, taking full economic advantage of the cloud network. As satcom continues to expand its reach with NGSO constellations and very High Throughput Satellite (vHTS) architectures, signal monitoring at scale will be a necessity for both quality of service assurance and SDA applications.
