AI-RAN backhaul saturation scenario

Wireless KPI observability pack comparing a baseline cell with a pre/post scenario event on CELL_001.
Baseline RMSE
4.39
prb_dl_util
Scenario RMSE
4.70
prb_dl_util
Risk tier
Congested
project-defined PRB threshold
Forecast RMSE ratio
1.07x
higher than baseline indicates stress
Network health summary
What this means: AI-RAN backhaul saturation scenario changes prb_dl_util from 36.81 pre-event to 35.34 during the scenario window. Operationally, the useful signal is not autonomous control; it is an early warning that lets a Non-RT workflow compare forecast error, capacity pressure, and the next monitoring action: review traffic-steering candidate timing.
RunCellTargetPre-shock MeanShock MeanPeakMAE
BaselineCELL_001prb_dl_util45.0032.8475.243.62
CongestionCELL_001prb_dl_util36.8135.3475.243.91
Throughput before/after: 127.69 to 107.15 Mbps. Latency before/after: 17.89 to 25.97 ms.
Risk tiers are project-defined over PRB DL utilization: Stable <60, Elevated 60-74.99, Congested 75-84.99, Critical >=85. They are decision-support thresholds, not operator SLA values.
Baseline forecast
AI-RAN KPI forecast: prb_dl_util actual, hold-out prediction, and forward forecast actual hold-out prediction forecast 2024-01-06 16:00:00+00:00 → 2024-01-08 23:00:00+00:00
Congestion forecast
AI-RAN KPI forecast: prb_dl_util actual, hold-out prediction, and forward forecast actual hold-out prediction forecast 2024-01-06 16:00:00+00:00 → 2024-01-08 23:00:00+00:00
Baseline impact
Before / After KPI impact historical telemetry versus forecast horizon Before: actual history After: forecast horizon Before average 51.63 After average 64.33 Impact +12.70 (+24.6%) peak before 74.17 | peak after 65.15
Congestion impact
Before / After KPI impact historical telemetry versus forecast horizon Before: actual history After: forecast horizon Before average 51.63 After average 62.58 Impact +10.96 (+21.2%) peak before 74.17 | peak after 63.87