DRL-Based RAN Slicing with Efficient Inter-Slice Isolation in Tactical Wireless Networks

The next generation of tactical networks (TNs) is poised to further leverage the key enablers of 5G and beyond 5G (B5G) technology, such as radio access network (RAN) slicing and the open RAN (O-RAN) paradigm, to unlock new architectural options for innovative applications. RAN slicing and the O-RAN paradigm are considered game changers in TNs, where the former enables tailoring services to users' requirements, and the latter brings openness and intelligence to RAN management. In TNs, bandwidth scarcity requires dynamic bandwidth slicing strategies. Although such strategies ensure efficient bandwidth utilization, they may compromise RAN slicing isolation in terms of quality of service (QoS). To deal with this challenge, we propose a deep reinforcement learning (DRL)-based RAN slicing mechanism, termed here as sharing and inter- and intra-slice isolation (BS-IISI), which achieves a trade-off between efficient bandwidth sharing and robust inter- and intra-slice isolation. BS-IISI performs bandwidth allocation in two stages. In the first stage, bandwidth is allocated to RAN slices, while in the second stage, each slice distributes its allocated bandwidth among associated users. In both stages, the slicing operation is constrained by QoS-related considerations that enhance inter- and intra-slice isolation. BS-IISI relies on DRL algorithms to perform bandwidth sharing at each allocation stage. We propose deploying the mechanism within an O-RAN architecture and describe the relevant O-RAN functional blocks and the main phases of the DRL model lifecycle management. Finally, we develop three implementations of the proposed mechanism, each based on a different DRL algorithm, and evaluate their performance against several baselines under diverse network configurations.