Ultra-wideband MIMO cognitive sensing systems: Algorithms, data processing, and testbed.

摘要

Current radar systems evolved from the adaptive radar to the radar with waveform diversity, and to the cognitive radar. The cognitive radar features cognition, which means that the radar can actively learn about the environment. The whole radar system forms a dynamic closed-loop consisting of the transmitter, environment, and the receiver.;Cognitive radar adjusts its system parameters and configurations in realtime to match its working environment and mission requirements. To build a cognitive engine, advanced mathematical tools like convex optimization are exploited to support waveform optimization. In order to provide a design philosophy and real-time demonstration of the concept of cognitive radar, an ultrawideband multiple-input multiple-output cognitive radar testbed is proposed.;In an ultra-wideband radar system, the unprecedented radio bandwidth provides advantages such as high-precision range estimation. However, the extremely high sampling rate of the analog-to-digital converter required in the radar system becomes a major challenge. Compressive sensing gives an opportunity to overcome this challenge, allowing the acquisition of signals at a much lower data rate than the Nyquist sampling rate. An algorithm is designed to get the time-of-arrival information from the sub-sampled echoed radar waveform. The effect of narrowband interference in the surveillance area is considered as well. A hardware architecture is proposed to fit into the special structure of the compressive sensing system.;The proposed cognitive radar system can be considered as a large scale sensing system as well. Random matrix theory is applied to the statistical analysis of the radar dataset. Random matrix theory can achieve better performance than that of the traditional methods in the large dataset condition. The statistical information has been reported for independent and identically distributed Gaussian signal, hardware noise, and radar waveform.