BEGIN:VCALENDAR VERSION:2.0 PRODID:IEEE vTools.Events//EN CALSCALE:GREGORIAN BEGIN:VTIMEZONE TZID:America/New_York BEGIN:DAYLIGHT DTSTART:20260308T030000 TZOFFSETFROM:-0500 TZOFFSETTO:-0400 RRULE:FREQ=YEARLY;BYDAY=2SU;BYMONTH=3 TZNAME:EDT END:DAYLIGHT BEGIN:STANDARD DTSTART:20261101T010000 TZOFFSETFROM:-0400 TZOFFSETTO:-0500 RRULE:FREQ=YEARLY;BYDAY=1SU;BYMONTH=11 TZNAME:EST END:STANDARD END:VTIMEZONE BEGIN:VEVENT DTSTAMP:20260713T151837Z UID:FFBCA96C-5E5E-4DC7-B90F-984D69B6FD29 DTSTART;TZID=America/New_York:20260813T113000 DTEND;TZID=America/New_York:20260813T130000 DESCRIPTION:Modern radar systems are becoming increasingly sophisticated an d responsive to their environments. While this flexibility enhances perfor mance\, it also introduces a fundamental vulnerability: behavioral adaptat ion becomes observable. An adversary can exploit radar emissions\, respons es\, and operational patterns to infer hidden states\, tracking strategies \, and intent. This talk presents a unifying perspective on radar interact ion as an adversarial inference problem\, grounded in statistical signal p rocessing and inverse reinforcement learning. Instead of estimating a targ et\, we shift the focus to inferring an adversarial radar\, using inverse filtering and natural language processing techniques to reconstruct the ra dar's estimate of the target\, its hidden beliefs\, and its operational in tent. We then address a fundamental question: how can one detect the prese nce of a cognitive radar and identify its underlying objectives? This moti vates revealed preference theory from microeconomics and inverse reinforce ment learning as principled tools for inferring radar intent from observed behavior. Finally\, we reverse the perspective: how can a radar mask its sensing strategy from an adversary while preserving operational effectiven ess? We discuss mechanisms including utility masking and information-theor etic approaches that degrade identifiability while preserving performance. \n\nSpeaker(s): \, Vikram Krishnamurthy\n\nRoom: McFawn Room\, Mile 2 Miss ion Delivery Center\, 5000 Springfield Street\, Dayton\, Ohio\, United Sta tes\, 45431 LOCATION:Room: McFawn Room\, Mile 2 Mission Delivery Center\, 5000 Springfi eld Street\, Dayton\, Ohio\, United States\, 45431 ORGANIZER:michael.callahan.10@us.af.mil SEQUENCE:9 SUMMARY:Adversarial Radars: Inference\, Intent\, and Plan Masking URL;VALUE=URI:https://events.vtools.ieee.org/m/567761 X-ALT-DESC:Description: <br /><p>Modern radar systems are becoming increasi ngly sophisticated and responsive to their environments. While this flexib ility enhances performance\, it also introduces a fundamental vulnerabilit y: behavioral adaptation becomes observable. An adversary can exploit rada r emissions\, responses\, and operational patterns to infer hidden states\ , tracking strategies\, and intent. This talk presents a unifying perspect ive on radar interaction as an adversarial inference problem\, grounded in statistical signal processing and inverse reinforcement learning. Instead of estimating a target\, we shift the focus to inferring an adversarial r adar\, using inverse filtering and natural language processing techniques to reconstruct the radar's estimate of the target\, its hidden beliefs\, a nd its operational intent. We then address a fundamental question: how can one detect the presence of a cognitive radar and identify its underlying objectives? This motivates revealed preference theory from microeconomics and inverse reinforcement learning as principled tools for inferring radar intent from observed behavior. Finally\, we reverse the perspective: how can a radar mask its sensing strategy from an adversary while preserving o perational effectiveness? We discuss mechanisms including utility masking and information-theoretic approaches that degrade identifiability while pr eserving performance.</p> END:VEVENT END:VCALENDAR