Adaptive frequency allocation in radar imaging: Towards cognitive SAR

The cognitive radar (CR) vision has been recently applied to various radar applications. The cognitive property assumes the radar to be able to dynamically adapt to environment changes. In this work, we propose to exploit SAR sub-Nyquist sampling methods, that have been originally proposed to reduce the sampling rate at the receiver, in order to dynamically adapt the transmitted waveform energy to vacant spectral bands. We extend the original sub-Nyquist approach by only transmitting the spectral bands that are to be sampled and processed on the receiver side, paving the way to cognitive SAR. In addition, we investigate several subsampling schemes of a chirp signal with a linear frequency modulation (LFM), which enables the best selection of the spectral bands within the vacant holes for SAR and show that randomness, which does not limit dynamic changes, is best fit to SAR. Both software and hardware simulations demonstrate dynamic transmitted signal support, while still allowing for SAR image recovery.