Open RAN Commercial Traffic Twinning Dataset

Dataset for the paper L. Bonati, R. Shirkhani, C. Fiandrino, S. Maxenti, S. D'Oro, M. Polese, T. Melodia, "Twinning Commercial Network Traces on Experimental Open RAN Platforms," in Proceedings of ACM WiNTECH, Washington, DC, USA, November 2024. [pdf] [bibtex]

The dataset is available at the following link. If you use it in your work, please reference the paper above.

Experimental setup

• Number of Base Stations (BSs): 1
• Channel bandwidth: 10 MHz (50 Physical Resource Blocks (PRBs))
• Number of slices for each BS: 2
• Scheduling policies available to each slice:
  • Policy 0: Round-robin (RR)
  • Policy 2: Proportionally fair (PF)
• Number of User Equipments (UEs): 8
  • Users 1-4 belong to the enhanced Mobile Broadband (eMBB) slice
  • Users 5-8 belong to the Ultra Reliable and Low Latency Communications (URLLC) slice
• UE mobility: static

UE ID	UE IMSI	Traffic Type
1	1010123456002	eMBB
2	1010123456003	eMBB
3	1010123456004	eMBB
4	1010123456005	eMBB
5	1010123456006	URLLC
6	1010123456007	URLLC
7	1010123456008	URLLC
8	1010123456009	URLLC

• Traffic: The traffic is twinned from a public dataset of LTE network from 3 BSs located in Madrid, Spain, available at this link.
  • BS1: carrier frequency at 816 MHz
  • BS2: carrier frequency at 1835 MHz
  • BS3: carrier frequency at 2650 MHz
• Number of slices for the BS: 2 (slice_id 0 for eMBB UEs and slice_id 1 for URLLC UEs)
• Slicing configurations: There are 50 PRBs in total to allocate for the two slices.

Slice Resouce Allocation	Slice ID 0 (eMBB: UEs 1-4) PRBs	Slice ID 1 (URLLC: UEs 5-8) PRBs
Slicing 1	9	41
Slicing 2	21	29
Slicing 3	30	20
Slicing 4	39	11
Slicing 5	50	0

Dataset stucture

• There are two compressed files available for downloading:
  • open-ran-commercial-traffic-twinning-dataset-kmp, which contains the metrics and the MGEN scripts.
  • open-ran-commercial-traffic-twinning-dataset-log, which contains the log files.
• The dataset has the following structure: cluster_<#>/slicing_<#>/scheduling<#>
  • 3 cluster configurations (cluster_1, cluster_2, and cluster_3)
  • 5 slicing configurations (slicing_1, slicing_2, slicing_3, slicing_4, and slicing_5)
  • 2 scheduling policies (scheduling_0, scheduling_2)
• In total there are 30 combinations of configurations in the available experiments.
• An example of a directory tree example for RESERVATION-142634 in cluster_1, slicing_1, scheduling_0 for both dataset files.

├── cluster_1
│   ├── slicing_1
│   │   ├── scheduling_0
│   │   │   ├── RESERVATION-142634
│   │   │   │   ├── bs
│   │   │   │   │   ├── 1010123456002_metrics.csv
│   │   │   │   │   ├── 1010123456003_metrics.csv
│   │   │   │   │   ├── 1010123456004_metrics.csv
│   │   │   │   │   ├── 1010123456005_metrics.csv
│   │   │   │   │   ├── 1010123456006_metrics.csv
│   │   │   │   │   ├── 1010123456007_metrics.csv
│   │   │   │   │   ├── 1010123456008_metrics.csv
│   │   │   │   │   ├── 1010123456009_metrics.csv
│   │   │   │   │   ├── enb_metrics.csv
│   │   │   │   │   ├── mgen-script-scope.mgn
│   │   │   │   ├── ue_001010123456002
│   │   │   │   │   ├── mgen.csv
│   │   │   │   │   ├── mgen-script-scope.mgn
│   │   │   │   │   └── ue_metrics.csv
...

├── cluster_1
│   ├── slicing_1
│   │   ├── scheduling_0
│   │   │   ├── RESERVATION-142634
│   │   │   │   ├── bs
│   │   │   │   │   └── enb.log
│   │   │   │   ├── ue_001010123456002
│   │   │   │   │   └── ue.log
...

• Each experiment has a directory named RESERVATION-<experiment_id>
• Inside each experiment directory, there is a directory named bs for the base station files, as well as directories named ue_<IMSI>, one for each UE, that contains the files related to the specific UE.
• Base Station files:
  • mgen-script-scope.mgen: MGEN script that twins the real-world traffic used for generating downlink traffic for the UEs in the experiment.
  • <IMSI>_metrics.csv: PHY- and MAC-layer KPMs collected at the base station for each UE.
  • enb_metrics.csv: downlink and uplink bitrate of the network, aggregated for all the UEs served by the base station.
  • enb.log: log file of the base station protocol stack.
• UE files:
  • mgen-script-scope.mgn: MGEN script running at the UE, used to start the MGEN listening process.
  • mgen.csv: Application-layer KPMs collected from the MGEN logs and converted to a .csv file using the recv-to-csv.sh script from the mgen-tools repository.
  • ue_metrics.csv: PHY- and MAC-layer KPMs collected at the UE side.
  • ue.log: log file of the UE cellular protocol stack.