A novel decentralized protocol is introduced to achieve heartbeat synchronization in large multi-agent systems, where agents produce brief, periodic flashes that align nearly simultaneously. The protocol operates entirely through one-way messages without requiring agents to share a common frequency or phase. These messages do not require responses and never block agents, eliminating the risk of deadlocks. The protocol proceeds as agents repeatedly contact randomly chosen peers, with a random delay before each transmission. Starting from random initial states, agents update their states upon receiving messages and emit light while a prescribed condition is satisfied. The reported study establishes a closed-form expression for the resulting flash period, showing dependence only on a protocol parameter and on the average number of messages sent per unit time per agent. Experiments conducted using a custom-built simulator confirm that the derived expression predicts the observed flash period accurately and precisely. These experiments further show that the flashes remain sufficiently brief and stable to support effective heartbeat synchronization.
Heartbeat Synchronization in Large Multi-Agent Systems Using One-Way Communication / Bergenti, F., Dallospedale, S., Monica, S.. - (2026), pp. 211-219. (25th International Conference on Autonomous Agents and Multiagent Systems, AAMAS 2026 2026) [10.65109/VFUP8012].
Heartbeat Synchronization in Large Multi-Agent Systems Using One-Way Communication
Bergenti F.;Dallospedale S.;
2026-01-01
Abstract
A novel decentralized protocol is introduced to achieve heartbeat synchronization in large multi-agent systems, where agents produce brief, periodic flashes that align nearly simultaneously. The protocol operates entirely through one-way messages without requiring agents to share a common frequency or phase. These messages do not require responses and never block agents, eliminating the risk of deadlocks. The protocol proceeds as agents repeatedly contact randomly chosen peers, with a random delay before each transmission. Starting from random initial states, agents update their states upon receiving messages and emit light while a prescribed condition is satisfied. The reported study establishes a closed-form expression for the resulting flash period, showing dependence only on a protocol parameter and on the average number of messages sent per unit time per agent. Experiments conducted using a custom-built simulator confirm that the derived expression predicts the observed flash period accurately and precisely. These experiments further show that the flashes remain sufficiently brief and stable to support effective heartbeat synchronization.I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.


