DYNAMIC RESOURCE MANAGEMENT FOR 5G NETWORK SLICING USING O-RAN NEAR-RT RIC

Abstract

Network slicing in 5G New Radio (NR) requires the simultaneous satisfaction of heterogeneous Service Level Agreements (SLAs) for ultra-reliable low-latency communications (uRLLC), enhanced mobile broadband (eMBB), and massive machine-type communications (mMTC). The objective of this study is to design and evaluate a dynamic, three-layer radio-resource-management framework, built on the O-RAN Near-Real-Time Radio Intelligent Controller (Near-RT RIC) that meets these three conflicting SLAs concurrently on a shared gNB. At the data plane, three algorithms are proposed: a Weighted Proportional-Fair (WPF) scheduler that maps RIC-issued weights to proportional-fair time windows; a slice-aware pre-processor that estimates per-slice Physical Resource Block (PRB) demand; and a starvation-aware Bandwidth Part (BWP) multiplexer. At the RIC, a heuristic BWP Manager observes per-slice key performance indicators every 100 ms and updates the slice weights through an exponential-moving-average-smoothed proportional update law. The framework is implemented in ns-3 v3.40 with the 5G-LENA NR module, co-simulated with a Python RIC through ns3-gym, and compared against a static-weight TDMA-PF baseline. Preliminary results from a 10-second co-simulation, in which a single gNB serves 5 uRLLC, 10 eMBB and 30 mMTC user equipments across three independent BWPs, show that the dynamic framework attains 100% uRLLC deadline compliance (mean delay 0.49 ms), 100% eMBB throughput success (34.9 Mbps per UE) and zero mMTC packet loss, whereas the static baseline fails all three SLAs. The idealized modelling assumptions and their implications for real deployments are discussed as limitations.