REPUBLIC OF SERBIA

MINISTRY OF DEFENCE

www.mod.gov.rs

MINISTRY OF DEFENCE

Material Resources Sector

Defensive Technologies Department

Military Technical Institute

www.vti.mod.gov.rs

Porting library for wideband direction finding:
lessons learned

Marko Mišić

University of Belgrade, School of Electrical Engineering, Belgrade, marko.misic@etf.bg.ac.rs

ANĐELKO CAJKOVIĆ

UNO-LUX NS d.o.o, Belgrade, andjelko@unoluxns.com   

nadica Kozić

Department for Electronic Systems, Military Technical Institute, Belgrade, nadica.kozic@gmail.com

ivan Pokrajac

Department for Electronic Systems, Military Technical Institute, Belgrade, ivan.pokrajac@vs.rs

Abstract: Direction finders are designed for interception of radio signals and direction of arrival estimation in wider instantaneous bandwidth with several simultaneously active signals. Our direction finder is implemented as a combination of field-programmable gate arrays (FPGAs) and graphics processing units (GPUs), since they offer vast computing power and increased bandwidth over implementations that use central processing units (CPUs) only.  Dedicated FPGAs were used for signal acquisition, while direction of arrival estimation and preprocessing for signal pre-classification were implemented on the GPU. Software support was almost exclusively developed in NI LabVIEW software package, as it was suitable for virtual instrumentation development and rapid prototyping. However, for scalability, portability, and performance reasons, vendor independent solution was needed. In this paper we present our experiences with implementation of software library for wideband direction finding in C/C++ programming language based on the existing LabVIEW prototype. We discuss the architecture of the software system, problems encountered, and evaluate the implemented solution. The main benefits of the proposed solution are better scalability, performance gains due to elimination of LabVIEW code and related overheads, and vendor independent solution written in pure C/C++ programming language.

Keywords: CUDA, direction finding, FPGA, GPU, parallel computing.