Hybrid Quantized Signal Detection With a Bandwidth-Constrained Distributed Radar System

This article addresses the hybrid quantized signal detection problem in a distributed radar system, where widely separated antennas transmit low-bit quantized data with varying quantization levels depending on the different bandwidth constraint for each channel to a fusion center. To enable a high detection performance with such hybrid quantized data, we formulate the generalized likelihood ratio, Rao, Wald, Gradient, and Durbin tests based on the derived likelihood functions, with the batch gradient descent algorithm (BGDA) being introduced to efficiently form an estimate of the unknown parameters. Furthermore, the theoretical distributions of all designed detectors are also presented, ensuring the constant false alarm rate property. Finally, the unimodality of the derived detection probability with respect to the quantization thresholds is proved, yielding the optimal quantizer to improve the detection performance, which may be efficiently solved using the BGDA. Numerical and experimental results demonstrate the performance of the detectors using the optimal quantizers, illustrating that 2-bit quantization can achieve excellent detection performance while significantly reducing the resulting data size.