Artificial Intelligence Software Agents in Air and Missile Defence

AOS’s employment of an agent-based approach is unique amongst the decision support tools that RAAF has been exposed to so far. All other such tools that have been demonstrated to … are reliant upon a combination of real-time information and a database of stored, previously captured data. The agent-based approach appears to provide some advantages, in its lack of requirement for stored data, and an ability to still provide decision support in situations that are dissimilar from those previously observed and catalogued.

As one RAAF expert commented above, AOS has just completed a demonstrator for the RAAF of how Artificial Intelligence Software Agents can assist Air Defence Operators to detect and assess unexpected and unidentified aircraft approaching Australia.


IBA offers:

  • “Personalised” attention to every track detected, unlike a human it doesn’t lose concentration, so that no tracks are overlooked.
  • Agents continuously monitor the aircraft performance, and immediately notice, and can alert the operator, if there is a deviation, e.g., an aircraft apparently routinely on track descending rapidly due to a depressurisation.
  • The operator’s concentration is then focused solely on the “outliers”, without the need to switch attention between frequent “scans” of the situation and resolving unexplained tracks.

With the rapid increase in air travel in Asia the amount of traffic to Australia’s north makes it difficult to manually detect and recognise potential intruders.


AOS’s Intelligent Battlespace Advisor™, or IBA, has been designed to detect:

  1. Uncooperative intruders that are detected from their radar/radio emissions.
  2. Uncooperative intruders flying under passenger aircraft, to conceal themselves.
  3. Cooperative spoofing, where an intruder copies another’s identity.
  4. Cooperative intruder with no flight plan enters Australia’s airspace.
  5. Coordinated action among several aircraft.
  6. Careful monitoring of selected areas at times of high importance, such as Olympic or Commonwealth Games or international meetings.

IBA allocates one intelligent software agent to each track it detects

So, if there are 300 tracks, then 300 agents are created, one monitoring each to determine its identity and intention. If IBA detects an aircraft that appears to be acting abnormally, e.g., it is 45 min early, then it investigates why, and only alerts the operator if the abnormality cannot be explained.