"Q5G-ZTN: QoS-Aware 5G OAI Network Time-Series Dataset with Controlled Congestion for ZTN"

"We generate a time-series dataset of network bandwidth and latency in a controlled 5G environment. The inputs to the system are the requested bandwidth and QoS type.Network Environment (OAI):The 5G core network (CN), gNB, UE, and application server are initialized sequentially. Dedicated UDP ports are assigned to each QoS class to emulate distinct QoS bearers. The network bottleneck is enforced at the User Plane Function (UPF) using a Token Bucket Filter (TBF) queue discipline. It is configured with a fixed capacity of 3 Mbps, a burst size of 32 kbit, and maximum buffering latency of 400 ms, thereby creating a deterministic capacity-constrained link.Network CongestionTo introduce realistic congestion, a constant 300 Kbps UDP background cross-traffic flow is generated. The foreground QoS traffic is transmitted using UDP via iperf. It follows a bounded random-walk process between 1000 and 4000 Kbps to simulate stochastic variations in the offered load.Captured KPIsPerformance metrics including achieved throughput, latency (measured via round-trip time variation), and derived latency are recorded at each time step for (a) a particular QoS class and (b) varying QoS classes. The result is a structured timeseries dataset suitable for QoS aware network modeling and prediction.System configurationThe dataset was generated on a Lenovo ThinkStation P2 Tower workstation running Ubuntu 24.04.3 LTS (64-bit). The system is equipped with an Intel Core i7-14700 processor, 32 GB RAM, Intel integrated graphics, and Linux kernel version 6.17.0-14-generic."