Noncoherent Multisite Radar Systems for Multitarget Tracking and Identification Using Micro-Doppler Signatures
Abstract
MmWave radar is a promising sensing modality for noncontact, privacy-preserving human tracking and identification. However, single-radar systems remain limited by spatial-resolution constraints, which often lead to target merging, trajectory fragmentation, and identity discontinuity under close-proximity interactions. To address these limitations, this article proposes a noncoherent dual-radar tracking framework for multitarget tracking and identification. The framework uses two orthogonally deployed noncoherent multiple-input-multiple-output (MIMO) radars to provide complementary range observations for trilateration-based localization. Its main contribution is an event-anchored cross-view association strategy that exploits temporally stable micro-Doppler kinematic-transition events as temporal anchors to establish cross-radar correspondence before trilateration. By linking cross-view micro-Doppler consistency with dual-radar geometric constraints, the proposed method mitigates ghost-target ambiguities in multitarget trilateration. To improve robustness, sequence-level consistency is further enforced for cross-radar event verification and identity-preserving trajectory association. A multitarget human kinematic simulation model is developed for controlled evaluation, and real-world indoor experiments are conducted for validation. Results show that the proposed approach achieves a target identity consistency accuracy (ICA) of 94.0% and a multiobject tracking accuracy (MOTA) of 94.35%. These results demonstrate effective preservation of spatial and temporal continuity in challenging indoor environments.