Skip to content
Open access

A DPRAM-Based Real-Time Distributed Control Architecture With Explicit Synchronization for Multi-Node Communication

2026 · IEEE Access · Vol 14, pp. 147463-147472 · 0 citations · 24 references

Abstract

This paper proposes a real-time distributed control architecture that separates Single Board Computer (SBC)-based high-level control computation from Communication Board microcontroller-based lower-level multi-node communication. Control and feedback main data are exchanged through a Dual-Port RAM (DPRAM)-based shared-memory interface. For repetitive DPRAM exchange, a software semaphore prevents concurrent read/write access, while a flag-based handshake signals data-update completion to prevent duplicate reception. Both mechanisms use designated DPRAM memory regions and require only limited synchronization overhead, supporting low-latency data exchange without additional hardware synchronization mechanisms. The software semaphore eliminated concurrent access and data mismatch with approximately <inline-formula> <tex-math notation="LaTeX">$10~\mathrm {\mu s}$ </tex-math></inline-formula> of overhead, and the flag-based handshake reduced the duplicate reception rate from 14.80% to 0.00%. System-level experiments evaluated 10 to 24 distributed nodes over a 27 Mbps communication bus. A 1 ms control period was satisfied up to 18 nodes, with a mean end-to-end closed-loop communication time of <inline-formula> <tex-math notation="LaTeX">$921.71~\mathrm {\mu s}$ </tex-math></inline-formula> and a maximum of <inline-formula> <tex-math notation="LaTeX">$931.88~\mathrm {\mu s}$ </tex-math></inline-formula>. Tests with 20 to 24 nodes used a 1.5 ms control period and showed an approximately linear increase in communication time with node count. A 24 h continuous-operation test at 18 nodes produced a mean communication time of <inline-formula> <tex-math notation="LaTeX">$922.08~\mathrm {\mu s}$ </tex-math></inline-formula> and a mean SBC CPU utilization of 65.69%, with no recorded communication time exceeding 1 ms. These results support practical timing scalability and stable extended operation under the evaluated configuration.

Read PDF

We use cookies to run the site and, with your consent, for analytics and to show ads. See our Cookie Policy.