Модель багатокритеріальної кооперативної маршрутизації на основі ієрархії цілей для мультиагентної системи

Abstract

The paper addresses the problem of QoS criteria conflict in multi-agent cooperative routing, caused by the inconsistency of quality metrics and the autonomy of agents, which gives the problem the nature of multi-criteria compromise optimization with the formation of a set of Pareto-efficient solutions. A three-level hierarchical goal structure GGG is proposed to align agents’ local actions with the global strategy through prioritized SLA compliance, maintenance of network stability, and optimization of resource utilization efficiency. The hierarchy is formalized as a graph and integrated with hierarchical reinforcement learning via the options framework, providing a structured resolution of QoS criteria conflicts in a multi-agent environment.