Creative Commons source reading · Online dispute resolution
Round-Robin Arbitration Architecture Technology of System-on-Chip
Zishen Zhao
Originally published in ITM Web of Conferences · 2026
From the article
Selected source text. JournalOS removes markup and shortens excerpts where indicated. The full article provides the complete context.
Abstract
Arbiters are fundamental units to the scheduling of the core in Network-on-Chip (NoC), multi-core processors, and high-speed crossbar switching systems. Round-robin is a fair arbitration scheme, which does not have request starvation, balanced scheduling, and excellent universality, so it is the most popular scheme in the current integrated circuit resource scheduling research and engineering application. In order to systematically describe the technological development and research results of low-latency round-robin arbitration architecture, this paper introduces four mainstream round-robin arbitration architectures: conventional linear round-robin arbitration, parallel prefix arbitration, matrix-based arbitration, and binary tree hierarchical arbitration, and expounds their working principle, performance characteristics, and application range. A complete comparison is done along various dimensions like latency, complexity, hardware cost, power consumption, port scalability, and scheduling fairness. This work presents an analysis of binary-tree hierarchical arbitration and its optimization techniques, identifies some of the current bottlenecks in low- latency round-robin arbitration, and discusses the future trends towards low power, scalability, and adaptability of high-performance on-chip arbitration circuits.
Editorial reading guide
- Whose experience is measured, and who might be missing from the evidence?
- How does the proposed design affect the ability to participate, understand a decision or seek review?
