Priority gossip-based synchronization method for distributed datain a group of unmanned aerial vehicles

Abstract

This paper presents the development and investigation of a priority gossip synchronization method for distributed data in a group of unmanned aerial vehicles (UAVs). The relevance of the topic is driven by the rapid expansion of UAV swarm applications in monitoring, reconnaissance, search and rescue, and logistics tasks, where effective real-time data synchronization between agents is a necessary condition for successful mission execution. Each UAV continuously generates sensor data including telemetry, coordinates, and observation results that must be distributed across the swarm for coordinated decision-making. Existing synchronization methods have significant limitations: the standard gossip protocol does not account for the current state of nodes when selecting an exchange partner, while the Raft consensus algorithm incurs substantial overhead for coordinator assignment and does not ensure fault tolerance upon coordinator failure. The problem lies in the need to develop a synchronization method that ensures fast convergence and minimal data staleness in swarm nodes without a centralized coordinator. The aim of the study is to develop a priority gossip synchronization method based on a node priority function. Each agent selects an exchange partner with probability proportional to its priority value rather than uniformly at random. For verification, computer simulation was conducted for swarms ranging from 50 to 200 UAVs, averaged over 30 independent runs. The results demonstrate a convergence improvement of 2.1–5.6% and lower average data staleness compared to the standard gossip protocol. The Raft algorithm showed the slowest convergence due to coordinator assignment overhead. The proposed method requires no centralized coordinator, is fault-tolerant upon individual agent failure, and adapts naturally to changes in onboard subsystem states.