RAN KPI Reference — 3GPP-Standard Definitions
Industry-standard KPI definitions for mobile network quality (GSM/UMTS/LTE/NR), following 3GPP performance measurement frameworks. Targets shown are typical industry values — every operator tunes their own.
Why KPIs matter for engineering
KPIs translate raw network counters into engineering decisions: accessibility tells you if subscribers can connect, retainability if they stay connected, mobility if the network moves them seamlessly. Every optimization, upgrade or swap is validated against KPI baselines — a change that degrades KPIs beyond tolerance rolls back, no matter how well it "worked".
1. Accessibility — can subscribers connect?
| KPI | Formula (3GPP) | Typical target |
|---|---|---|
| RRC Setup Success Rate (LTE/NR) | successful RRC setups / attempts | > 99.5% |
| E-RAB / QoS Flow Setup Success Rate | successful E-RAB setups / attempts | > 99.5% |
| CS Call Setup Success Rate (2G/3G) | successful call setups / attempts | > 99% |
Engineering drivers: radio coverage (RSSI/RSRP/RSRQ), congestion (PRB/license limits), signaling failures (core-side), parameter errors (RACH config, admission control).
2. Retainability — do subscribers stay connected?
| KPI | Formula (3GPP) | Typical target |
|---|---|---|
| E-RAB Drop Rate (LTE/NR) | abnormal E-RAB releases / successful setups | < 0.5% |
| Call Drop Rate (CS) | abnormal releases / successful calls | < 1% |
Engineering drivers: coverage holes, neighbor relation gaps (missing/incorrect ANR), interference, hardware faults (TRX/clock), transport instability.
3. Mobility — does the network move subscribers seamlessly?
| KPI | Formula (3GPP) | Typical target |
|---|---|---|
| Intra-frequency HO Success Rate | successful HO / attempts | > 99% |
| Inter-frequency HO Success Rate | successful HO / attempts | > 98% |
| Inter-RAT HO Success Rate (LTE↔UMTS/GSM) | successful HO / attempts | > 98% |
Engineering drivers: neighbor lists (ANR efficiency), HO thresholds (A3/A5 events and hysteresis), coverage imbalances, PCI confusion (LTE/NR).
4. Utilization & Capacity — is there headroom?
| KPI | Formula | Healthy range |
|---|---|---|
| PRB Utilization (DL/UL) | used PRBs / available PRBs | < 70% peak |
| Cell Availability | available time / total time | > 99.9% |
| Active users per cell | connected users / cell | vendor/license dependent |
Engineering drivers: traffic growth vs. capacity planning, spectrum refarming opportunities, carrier adds, load balancing parameters (mobility-based balancing).
5. The KPI workflow in practice
- Baseline: record 24h KPI before any change
- Change: upgrade, swap, reconfiguration, refarm
- Post-check: compare KPIs vs. baseline (2h immediate + 24h full)
- Tolerance: degradation beyond tolerance (e.g. > 0.2% drop-rate increase) → rollback
- Trend analysis: weekly/monthly degradation trends drive optimization priorities
How this maps to the tooling in this repository
src/config_drift.pycatches configuration drift that silently degrades KPIs (missing neighbors, altered thresholds)src/subnet_prober.pygates deployments on fleet reachability before changesexamples/mop_template.mdembeds the baseline/post-check/tolerance workflow into change management