Radar Identification for Noncooperative UAVs Using Multidomain Avian Activity Cognition Models

Radar identification for birds and uncrewed aerial vehicles (UAVs) are challenging for their radar signature similarity. Activity character and environmental dependency differences between two types of targets provide additional information for target distinguishing. Avian radar datasets contain abundant information to support bird activity cognition modeling. This article introduces a noncooperative UAV identification framework by characterizing bird targets as environmental-dependent organisms. A multidomain avian activity cognition model is constructed in spatial-temporal distribution and motion character domains from avian radar datasets. UAV targets are modeled as environmental independent and duty motivated. Their characters in spatial-temporal and motion character domains are extracted and applied on the bird cognition model. A matching score is calculated to quantify the similarity with bird targets. Target identity is determined according to its matching score. Historical datasets from an avian radar system are collected for verification. Identification results prove the reasonability and feasibility of bird cognition models. Compared with conventional radar signature-based target identification methods, the involvement of bird cognition models elevates the comprehensiveness of bird signatures with higher classification accuracy. The drawback of the method is its reliance on sufficient avian radar data supports. Autoadjusting weighing factor assignment and intelligent identification method for spatial grids with limited data supports are also discussed as future optimization works.