NATO STO

NATO SET-256 explored issues associated with interoperability of discrete tactical sensors in order to build a coalition-based sensor network. NATO-SET 256 conducted an experiment where the team conducted the successful integration of sensors using three separate middlewares running simultaneously on sensors. Specifically, communication to Canada via Open Standards for Unattended Sensors (OSUS), the UK via the SAPIENT protocol, and NATO via the draft STANAG 4789. While specifically addressing the challenge of sensor interoperability, the success raised additional questions that a follow-on ET should explore. Currently we often look at AI enabled systems as standalone entities, but the larger impact of intelligent systems comes through system and information collaboration. Looking forward, there is increasing demand for sensors and platforms that are AI enhanced and can support multi-layered, multi-domain defence applications. The need for intelligent integrated battle command and defence systems blurs the lines between traditional edge and enterprise capabilities pushing more of the intelligence and rapid decision-making process toward the tactical edge where the promise of AI assisted rapid decision making can have the largest impact. Processing technology continues to advance but the limits of size, power weight and data access at the edge dictate that maximum capability is still achieved with collaborative connected systems in which tactical edge systems can reach back to higher level information and systems when necessary and possible. There are of course both technological and operational obstacles to development and deployment of such complex interoperable systems. For example, one obstacle to true interoperability is trust, and that applies both machine to machine and human to machine. What can we do to increase trust between multi-national systems so users can trust the tasking and data output. How do we move higher level processing like fusion and AI/ML to the sensor / edge side? How is AI/ML and edge fusion best distributed in a complex system and workflow? Can we make information extraction and fusion more collaborative so that both human and AI agents can share information and tasking to provide better information and faster decision making? How can such systems reliably work in degraded and contested environments? Finally, in addition to technical hurdles, changes must be made in policy, joint and multinational operations doctrine, concepts of operation, and other impactful areas to remove roadblocks that hinder the deployment of higher-level automation, collaboration, and information fusion.