Towards compressed sensing for ground-to-air monostatic radar

• Main Authors: Dauwels, Justin, Kannan, Srinivasan
• Other Authors: School of Electrical and Electronic Engineering
• Format: Conference or Workshop Item
• Language: English
• Published: 2016
• Subjects: Remote sensing
• Online Access: https://hdl.handle.net/10356/81224; http://hdl.handle.net/10220/40678
• Institution: Nanyang Technological University

Recently, it is shown that the fundamental problem of rangeDoppler estimation can be solved efficiently by compressed sensing (CS) from single-pulse radar return. The performance of CS radar particularly degrade significantly with noise and hence the primary concern is to determine the regime where the operation of CS radar is satisfactory. Most of the studies on CS radar are often conducted under unrealistic conditions where the SNR is much higher than in typical radar applications (e.g., SNR > 5dB). In this paper, we investigate how to improve the CS reconstruction by using coherent integration over N pulses. We consider two scenarios: i) coherent integration is performed before CS reconstruction; ii) coherent integration is performed after CS reconstruction. We provide numerical results for both scenarios, and demonstrate that a proportional reduction in reconstruction error is obtained if coherent integration is carried out before CS reconstruction, corresponding to an effective gain of SNRg = 10 log10 N. On the other hand, when coherent integration is performed after applying CS reconstruction to single pulse radar returns, there is negligible gain. Both observations can be explained by the fact that CS reconstruction exhibits a threshold phenomenon with regard to SNR. By boosting the effective SNR through coherent integration, one can obtain more reliable CS reconstruction.