A Low-Cost Monopulse Receiver with Enhanced Estimation Accuracy Via Deep Neural Network

TL;DR

This paper presents a low-cost monopulse receiver with a novel comparator network design and enhances direction of arrival estimation accuracy using a deep neural network, validated through simulation and measurement at 2 GHz.

Contribution

It introduces a cost-effective monopulse comparator network with novel port-transformation rat-race couplers and applies deep learning to improve DoA estimation accuracy.

Findings

Prototype operating at 2 GHz successfully designed, simulated, and measured.

Enhanced DoA estimation accuracy achieved with DNN mapping.

Proposed design reduces complexity and fabrication costs.

Abstract

In this paper, a low-cost monopulse receiver with an enhanced direction of arrival (DoA) estimation accuracy via deep neural network (DNN) is proposed. The entire system is composed of a 4-element patch array, a fully planar symmetrical monopulse comparator network, and a down conversion link. Unlike the conventional design topology, the proposed monopulse comparator network is configured by four novel port-transformation rat-race couplers. In specific, the proposed coupler is designed to symmetrically allocate the sum ({\Sigma}) / delta ({\Delta}) ports with input ports, where a 360{\deg} phase delay crossover is designed to transform the unsymmetrical ports in the conventional rat-race coupler. This new rat-race coupler resolves the issues in conventional monopulse receiver comparator network design using multilayer and expensive fabrication technology. To verify the design theory, a prototype of the proposed planar monopulse comparator network operating at 2 GHz is designed, simulated, and measured. In addition, the monopulse radiation patterns and direction of arrival are also decently evaluated. To further boost the accuracy of angular information, a deep neural network is introduced to map the misaligned target angular positions in the measurement to the actual physical location under detection.