The design and performance of the real time software architecture for the ITER Radial Neutron Camera

TL;DR

This paper presents the design and testing of a high-performance real-time software architecture for the ITER Radial Neutron Camera, capable of processing large data rates and providing real-time neutron emissivity profiles during plasma experiments.

Contribution

It introduces a distributed multi-core CPU based architecture and algorithms tailored for real-time neutron spectra processing and emissivity calculation in ITER diagnostics.

Findings

Achieved real-time processing at 2 MEvents/s per LOS.

Demonstrated effective neutron/gamma discrimination and pile-up correction.

Preliminary tests show the system meets ITER's performance requirements.

Abstract

The neutron detection system for characterization of emissivity in ITER Tokamak during DD and DT experiments poses serious challenges to the performance of the diagnostic control and data acquisition system (CDAcq). The ongoing design of the ITER Radial Neutron Camera (RNC) diagnostic is composed by 26 lines of sight (LOS) for complete plasma inspection. The CDAcq system aims at meeting the ITER requirements of delivering the measurement of the real-time neutron emissivity profile with time resolution and control cycle time of 10 ms at peak event rate of 2 MEvents/s per LOS. This measurement demands the generation of the neutron spectra for each LOS with neutron/gamma discrimination and pile up rejection. The neutron spectra can be totally processed in the host CPU or it can use the processed data coming from the system FPGA. The number of neutron counts extracted from the spectra is then used to calculate the neutron emissivity profile using an inversion algorithm. Moreover, it is required that the event based raw data acquired is made available to the ITER data network without local data storage for post processing. The data production for the 2 MEvents/s rate can go up to a maximum data throughput of 0.5 GB/s per channel. The evaluation of the use of real-time data compression techniques in RNC is also depicted in another contribution. To meet the demands of the project a CDAcq prototype has been used to design and test a high-performance distributed software architecture taking advantage of multi-core CPU technology capable of coping with the requirements. This submission depicts the design of the real-time architecture, the spectra algorithms (pulse height analysis, neutron/gamma discrimination and pile-up correction) and the inversion algorithm to calculate the emissivity profile. Preliminary tests to evaluate the system performance with synthetic data are presented.