Coded Waveforms for Collocated MIMO Radar Using Sparse Modelling

Many commercial collocated MIMO radar platforms (especially in automotive applications) rely on TDMA-FMCW, which enforces sequential transmission across antenna elements and limits the unambiguous measurement space. Alternatives like PMCW-CDMA allow simultaneous transmissions but introduce elevated sidelobes that can mask weak targets. Moreover, achieving fine angular resolution by sampling at or above the Nyquist rate requires large, densely populated arrays, imposing heavy cost, storage, and processing burdens.
This study presents a novel methodology based on Sparse Annealed Projections (SAP), a compressed-sensing-native framework that co-optimizes phase-coded CDMA waveforms and nonuniform array geometries via a temperature-guided Metropolis-Hasting's optimizer. By strategically placing and selecting sampling points to minimize mutual coherence in the sensing matrix, SAP achieves up to 13 dB of sidelobe suppression, reduces storage by 60 %, and speeds reconstruction by 40 %, all without degrading detection performance at sub-Nyquist rates. SAP thus enables compact, resource-efficient radar across automotive, aerospace, maritime, and defence domains.