Integration of DPDK for Real-Time Communication in the Elettra Synchrotron Orbit Feedback Control System: Jitter and Latency Optimization
Luca Anastasio, Alessio Bogani, Mauro Cappelli, S. Di Gennaro, G. Gaio, M. Lonza · IEEE Transactions on Industrial Informatics · 2025
The scope of this work is the evaluation of the software-driven platform for the upcoming global orbit correction feedback system for the Elettra 2.0 synchrotron to be deployed in 2026. Here, a detailed analysis including latency and jitter of a full acquisition system based on Ethernet and an Intel server running Linux and DPDK is presented for the first time in the literature of particle accelerators. This assessment has primarily focused on measuring the receiving time of hundreds of remote devices based on FPGAs and their data processing. Unlike the current orbit correction system, which is distributed across multiple CPUs and involves heterogeneous hardware, the approach presented here is based on the centralization of the data acquisition and processing within a single server. The latter is to be connected to a maximum of a hundred of sensors via Ethernet ensuring, under heavy CPU and memory load, real-time communication with these peripherals through the utilization of the DPDK framework in conjunction with core isolation capabilities of multicore CPUs. The expected performance in terms of feedback repetition rate is between 40 and 100 kHz. The results show that the global orbit feedback for the Elettra 2.0 synchrotron can be definitely realized on the proposed hardware, which has many advantages over the current global orbit feedback system. This work demonstrates the viability of a software-based solution for high-frequency feedback, matching the latency and determinism of specialized FPGA systems. It introduces a scalable model for high-speed feedback applications, such as nuclear fusion and advanced manufacturing, validating DPDK as an alternative for these fields. This foundation supports future exploration of HPC-based centralized control systems in scientific infrastructures.