Demo Proposal For Single-Input Multi-Output Sub-Nyquist Radar With Unknown Pulse Shape

In this demo, we present a single-input multi-output (SIMO) radar system that operates at a sub-Nyquist rate without assuming any priori knowledge of the transmit pulse shape. Shannon-Nyquist sampling technique is the most widely used analog to digital conversion (ADC) method where the sampling rate is equal to twice the bandwidth of the signal to be sampled. In active radar imaging, the transmit pulse is narrow in time and hence has a wide bandwidth. This results in a high sampling rate and expensive ADCs. However, in radar imaging, the received signal has a structure that can be exploited to reduce the sampling rate. Specifically, the received signal consists of a stream of time-shifted and amplitude scaled copies of the transmit pulse. By assuming that the transmit pulse is known, the received signal has a finite rate of innovation and hence can be sampled within the Xampling framework where the ADC operate in the sub-Nyquist regime. However, in practice, the transmit pulse is distorted while propagation and hence, one cannot use an existing sub-Nyquist framework. In this demo, we present the implementation of recently proposed sub-Nyquist framework with unknown pulse shape [1]. Specifically, we demonstrate that low-rate sampling is still possible if the signal is received through multiple antennas. In this case, the problem of estimating the target can be posed as a sparse multichannel blind deconvolution problem and we show that it is possible to uniquely identify the targets from compressed measurements. Our demonstration platform consists of a Vector Signal generator with delay capabilities transmitting in two different RF channels into a four channel oscilloscope. The prototype will be presented along with a dedicated GUI depicting the computed performance measures and allowing comparison to sub-Nyquist scheme with a known pulse. References: 1. S. Mulleti, K. Lee, and Y. C. Eldar, “Identifiability Conditions for Compressive Multichannel Blind Deconvolution,” arXiv preprint arXiv:2001.00613, 2020