The Missile Defense Agency is rolling out a software-focused upgrade to improve the midcourse discrimination capability of its Upgraded Early Warning Radars (UEWR), part of the Advanced Object Classification (AOC) effort. The UEWR network — composed of five long-range phased-array radars located in the United States, Greenland and the UK — has long provided strategic detection and tracking for the Ground-Based Midcourse Defense system, but distinguishing real warheads from decoys in the midcourse phase remains a persistent technical challenge. Rather than pursuing hardware changes, the MDA is investing in advanced signal-processing techniques and machine-learning classification algorithms intended to better separate threat objects from benign signatures during the flight phase when targets are in space and countermeasure deployment is likely. The AOC program has been fielding iterative software releases; AOC 1.0 delivered initial classification improvements and has been exercised at selected UEWR sites. The next release, AOC 1.1, will incorporate more refined feature extraction, improved clutter suppression, and enhanced probabilistic models to raise discrimination confidence within the 20-minute midcourse window. Importantly, these upgrades are designed to operate without changing radar hardware or altering transmitted waveforms, enabling fleet-wide software updates to be deployed quickly and cost-effectively. In the broader defense architecture, better midcourse classification helps the Ground-Based Midcourse Defense system — which pairs UEWR cues with 44 ground-based interceptors — make more informed targeting and engagement decisions. Although the GMD capability is currently tailored to defend against limited ballistic missile raids rather than large salvo attacks, enhancing radar classification can reduce false alarms and improve interceptor shot selection in contested scenarios. Still, experts caution that software upgrades alone are not a panacea: comprehensive resilience typically requires sensor diversity across boost, ascent, midcourse and terminal phases, alongside layered discrimination tools. Nonetheless, AOC 1.1 represents a meaningful step to extract greater performance from existing infrastructure through improved algorithms and edge processing.
