Intent-Driven RAN Slice Orchestration: A Multi-Agent Deep Reinforcement Learning Based Approach

Radio access network (RAN) slicing is a promising technology for meeting various service demands by establishing multiple logical networks on a shared physical infrastructure. However, the diverse quality of service (QoS) requirements of different services pose challenges to the efficient operation of network slices. Intent-driven network (IDN) automates and orchestrates networks, which can assist in operating network slices and ensuring the QoS requirements. Therefore, this paper introduces user intent into the RAN resource slicing. To maximize the service level agreement (SLA) satisfaction degree, we propose an intent-driven multi-agent deep Q-network (MA-DQN) based algorithm for resource allocation. Simulation results demonstrate the superiority of the proposed algorithm over baseline algorithms in terms of convergence, SLA satisfaction degree (SSD), and average data rate.