RF-Based Direction Finding of UAVs Using DNN

TL;DR

This paper introduces a low-complexity, practical RF-based direction finding method for UAVs using a deep neural network that does not require phase synchronization or antenna calibration, validated through experiments.

Contribution

It proposes a novel SDAE-based DNN approach for UAV direction finding that simplifies implementation by removing the need for phase synchronization and calibration.

Findings

Effective UAV direction finding demonstrated experimentally.

Outperforms classical algorithms in practical scenarios.

Operates with a single-channel RF receiver.

Abstract

This paper presents a sparse denoising autoencoder (SDAE)-based deep neural network (DNN) for the direction finding (DF) of small unmanned aerial vehicles (UAVs). It is motivated by the practical challenges associated with classical DF algorithms such as MUSIC and ESPRIT. The proposed DF scheme is practical and low-complex in the sense that a phase synchronization mechanism, an antenna calibration mechanism, and the analytical model of the antenna radiation pattern are not essential. Also, the proposed DF method can be implemented using a single-channel RF receiver. The paper validates the proposed method experimentally as well.