Abstract
This article describes research into the design and analysis of decentralized algorithms for computing convoy patterns (a density-based movement pattern amongst moving point objects, previously formalized and computed in a centralized manner). A series of decentralized algorithms are presented and discussed (naïve, tree-based, and further devolved reactive approaches) for mobile geosensor nodes to share the necessary information to enable in-network detection and monitoring of convoys; these approaches require neither centralized control nor quantitative positioning. Empirical evaluation of algorithm characteristics and performance (via simulation experiments with synthetic and real movement data) demonstrate algorithm efficiency and effectiveness (with various strategies to structure information flow and devolve decision-making considerably improving communication efficiency).

This publication has 24 references indexed in Scilit: