"AMI-5G"

"The increasing reliance of Advanced Metering Infrastructure (AMI) systems on 5G connectivity creates a strong need for realistic cybersecurity datasets that capture availability-oriented attacks under practical network conditions. This paper presents 5G_AMI, a flow-based smart-grid cybersecurity dataset generated from packet captures collected on the IEEE 5G\/6G Innovation Testbed. The dataset models MQTT-based smart-meter reporting traffic delivered through a 5G core toward fog-side brokers and includes multiple availability attacks, namely SYN flood, TCP flood, ICMP flood, MQTT CONNECT flood, MQTT SUBSCRIBE flood, and TCP hold-open. Because the collected traffic includes GTP-U encapsulation, the PCAP traces are preprocessed using Python scripts to recover inner IP\/IPv6 traffic, retain other observable IP\/IPv6 packets, and convert the result into Ethernet-formatted PCAPs compatible with CICFlowMeter. The resulting flow CSV files are then labeled according to scenario type and aggregated into the final dataset. Experimental validation includes leakage-aware evaluation, probability calibration analysis, and cross-vantage testing between UE-side and UPF\/broker-side captures, demonstrating that the dataset supports robust and deployment-relevant attack detection. The dataset is intended to support future research in 5G-enabled AMI security, attack detection, and resilient self-healing mechanisms."